Roman Dymock: Archaeological Investigations 1995-2002, Introduction by Toby Catchpole, Tim Copeland and Ann Maxwell 2007, Vol

From the Transactions of the Bristol and Gloucestershire Archaeological Society

Roman Dymock: Archaeological Investigations 1995-2002, Introduction by Toby Catchpole, Tim Copeland and Ann Maxwell 2007, Vol. 125, 131-136

Trans. Bristol & Gloucestershire Archaeological Society 125 (2007), 131 –245

ROMAN DYMOCK: ARCHAEOLOGICAL INVESTIGATIONS 1995–2002

This set of papers brings together the results of recent work on Roman Dymock. An introductory essay, which covers geophysical surveys of the course of a Roman road, is followed by the reports on the excavations at the Sewage Treatment Works in 1995 and on land adjacent to the Rectory in 2002. The concluding papers consider the origins and function of the Roman settlement, the regional context and future research directions.

Introduction

By TOBY CATCHPOLE, TIM COPELAND and ANN MAXWELL

The village of Dymock, centred on OS Nat. Grid SO 703312, is situated in north-west Gloucestershire, between Newent and Ledbury and on the southern edge of the floodplain of the river Leadon. Dymock parish is bounded on the west by Kempley, Gloucestershire. To the north its parish extends to the Herefordshire boundary and c.3.5 kilometres to the east the course of the Leadon was until 1931 the boundary between Gloucestershire (Pauntley) and Worcestershire (Redmarley D’Abitot). The central area of the village is situated on an outcrop of Lower Old Red Sandstone. This is overlain by scattered patches of sand and gravel deposits of the Third (Main) Severn River Terrace. Alluvium overlies local deposits in the floodplain of the Leadon and its tributaries (Worssam et al . 1989). The parish church of St Mary is situated on the highest point towards the east end of the present village. It is thought that Dymock is a Welsh name and that the first syllable is from the Welsh d n, meaning ‘fort’ (although Ekwall offered the Welsh ‘ty’, meaning house’), followed by moch, meaning ‘swine’ (Smith 1964, 168). That the village has Saxon origins is evidenced by the fabric of St Mary’s church. Although substantially rebuilt in the early Norman period, the church retains some Anglo-Saxon fabric in the lower parts of its walls. In Domesday Book the manor of Dymock is recorded as having a substantial population. Although a failed borough in the early 13th century (Finberg 1957), the village grew during the medieval period before declining from the 14th century onward, possibly as a result of economic competition from nearby Newent and Ledbury.

Iron-Age Period Dymock is likely to have Iron-Age origins (Millet 1990, 35), but clear evidence for pre-Roman activity in the area is limited. It stands within the probable territory of the late pre-Roman tribe of the Dobunni (Fulford 2003, 17–19). Their boundaries have traditionally been defined by the distribution of their coins (Allen 1961, 75–149) and are seen as including Kenchester in the west and Worcester in the north (Darvill 2003, 9), although this view is now seen as problematic (Moore 04_BGAS125_131-246 16/11/07 15:49 Page 132

132 TOBY CATCHPOLE, TIM COPELAND AND ANN MAXWELL

Lyonshall Worcester N 6c

s l 63b l 180 i R

H . Stretton S e v n 63a Grandison e r r e n Kenchester v l

R a

. Hereford M L

610 a

R . W y e 613 Dymock Tewkesbury

Weston Newent R under . M on Penyard no w 611 Gloucester 612a 61

41c 614 Monmouth Forest of Dean 612

Fig. 1. Dymock: regional setting and Roman road system

and Reece 2001). At least three silver ‘Western Regular Type’ Iron-Age coins (more generally known as Dobunnic) have been found in Dymock (McWhirr 1981, 69; R. Poole pers. comm. to Maxwell) and a Gaulish base-gold coin of the Baiocasses was found there in the mid 19th century (Gethyn-Jones 1991, 98). A gold-plated stater, inscribed EISV and probably a contemporary forgery, was found just east of the village in 2006 (D. Sherratt pers. comm. to Maxwell). A copper- alloy zoomorphic fragment of a late Iron-Age or early Roman probable bucket mount was found within the village in 2003. Children and Nash (1994, 83) suggest that Iron-Age coins found in Herefordshire often in later, Roman, contexts have been curated. Indeed, it has been suggested that no coins were used in Herefordshire before the Roman conquest (Stanford 1991, 75). There are five Iron-Age hillforts within a 10-kilometre radius north of Dymock and another five within 15 kilometres. The nearest is Haffield Camp in Donnington, Herefordshire, 3.5 04_BGAS125_131-246 16/11/07 15:49 Page 133

ROMAN DYMOCK: INTRODUCTION 133

kilometres away. A small, triple-ditched, irregular enclosure located in the adjacent parish of Bromesberrow is considered likely to have been ‘a substantial settlement of Iron Age date’ (Bowden 2005, 27). Cropmarks visible on an oblique air photograph taken in the 1950s (National Monuments Record, WAB 11624/268, SO 7031/8) show possible Iron-Age enclosures in Dymock, east of the village and the cricket field. A small amount of pre-Roman rock-tempered pottery was found at the Rectory site (see Brown and Timby, below), but the occurrence of this material alongside Severn Valley wares in a number of contexts suggests that it represents a continuation of native traditions of pottery manufacturing into the early part of the Roman period rather than indicating the presence of a late Iron-Age settlement.

Roman Period

Communications The Ravenna Cosmography indicates that the road which ran from Magnis (Kenchester) to Glevum (Gloucester) passed through two now unlocated places, Epocessa and Macatonium . Gethyn-Jones (1966, 11) identified Epocessa as Stretton Grandison, Herefordshire, and suggested that Macatonium may have been the settlement at Dymock. The name ‘Macatonium’ was suggested by Rivet and Smith (1979, 405) to be of British derivation, meaning ‘place on the noble river’. If correct, a location by the Wye may perhaps be more likely than one by the Leadon. The Ravenna Cosmography was drawn up c.700 from a variety of sources and is known to be extremely corrupted through repeated copying, so that all names uncorroborated from other sources are suspect (Jones and Mattingly 1990, 29). The lack of evidence for a 3rd- and 4th-century occupation at Dymock may well indicate that it was not a settlement of sufficient significance to be recorded in contemporary Roman itineraries, although if it had indeed been deliberately founded or had a role in the Cursus Publicus (see Catchpole, discussion of the sewage treatment works excavations, below) it may be possible that it continued to be recorded in place-name lists. The identification of Dymock as Macatonium therefore remains an unverified possibility. The Roman settlement at Dymock has been said to lie at the junction of two roads (Leech 1981), one running from Stretton Grandison to Dymock and then, it has been assumed, on to Gloucester, and the other probably originating near Tewkesbury and terminating at Dymock. Margary listed the road from Stretton Grandison (number 610 in his survey) but pointed out that its route between Dymock and Gloucester was unclear for much of the way (1973, 328–9). The modern road between Stretton Grandison and Dymock lies mainly along the line of the Roman road, and where it curves away, a couple of kilometres north of Dymock, its Roman line is fossilised in field boundaries. Thirty years later the course of the road between Dymock and Gloucester has still not been traced. Although Gethyn-Jones suggested that the junction of the two roads lay under the cricket pitch (Fig. 2, site 17) and a recent geophysical survey (Maxwell 2004) adds some credence to this suggestion, clear evidence only exists for one road, one that ran from the fort and settlement at Stretton Grandison (Buteux 1996) to Dymock and then turned east towards Tewkesbury. The route at Dymock is that suggested by Gethyn-Jones (1991) and comes to within 12.5 metres of the southern side of the excavated enclosures reported below at the sewage works (Fig. 3) and probably immediately north of the Rectory site, making the identified features parallel to the road. The route eastwards from the sewage works can be traced on modern maps and aerial photographs for c.1.4 kilometre towards Crowfield Farm and the M50 motorway. The Malvern Research Group carried out intermittent work at the cricket ground over many years, the results being summarised in annual newsletters (Waters 1960–72) and a short Glevensis article (Waters 1969). Several trenches were cut through the road. One indicated occupation under the road predating its construction. 04_BGAS125_131-246 16/11/07 15:49 Page 134

134 TOBY CATCHPOLE, TIM COPELAND AND ANN MAXWELL

Key N Excavations Roman roads (known and possible locations) Gethyn-Jones site nos. references

Rose Cottage

w o r r School e no. 6 b Sewage works e n d Rectory o l t e no. 7 S i f no. 8 Cricket Ground

no. 17

Fig. 2. Location of the sewage works and Rectory excavations and of other sites in Dymock mentioned in the text.

Another suggested five or six episodes of resurfacing of the gravel road, which varied from c.3.6 to 6.0 metres in width. Although its eventual destination is unknown, the road leaves Dymock heading in the direction of the modern Haw Bridge, across the river Severn south of Tewkesbury, after which it presumably joined the Gloucester to Worcester road (Fig. 1).

Evidence for a Military Origin Leech (1981) suggested that Roman occupation at Dymock may have originated as a fort. Webster (1993, fig. 37) also identified Dymock as one of many possible sites for a Roman fort due to its situation on a ridge above a river. Dymock is c.18 kilometres north-west of the Roman legionary fortress at Gloucester and about 14 kilometres south of the known fort and suggested vicus at Stretton Grandison (Buteux 1996, 1–4) – a day’s march from each of them. However, no firm 04_BGAS125_131-246 16/11/07 15:49 Page 135

ROMAN DYMOCK: INTRODUCTION 135

evidence for the fort at Dymock is indicated by possible interpretations of its name (see above). At the cricket field, parallel ditches below the earliest road surface, reported in 1966 as triple but in 1969 as double by the Malvern Research Group, were dated from an association with an un- inscribed silver Dobunnic coin, a Roman Republican denarius and a Claudian dupondius (Waters 1969). There is no information about the depth or shape of the ditches, the distance between them, or their position in relation to the road. Multiple phases of wooden buildings lining the road were indicated by beam slots and postholes, gravel floors and daub. There is no evidence that any of the features reported (below, civilian occupation) were of military origin and the samian report from the sewage works site (below) suggests activity from c.AD 70 onwards. Until further evidence is available that has to be seen as the origin of Roman activity at Dymock.

Civilian Occupation The evidence for a Roman period civilian settlement at Dymock has been collected on a small- scale, opportunist basis and has not always been retrieved or recorded to acceptable present-day standards. For at least a century Dymock has been recognised as the site of such a settlement (Gethyn-Jones 1966, 9; Leech 1981, 30–31; McWhirr 1981, 59, 67–71) extending over a large area from the north-western edge of the existing village eastwards into the cricket field (Fig. 2). The focus of the settlement is thought to lie on the highest ground, where the church now stands. According to Gethyn-Jones (1991, 97) the occupation covered an area of 35–40 acres (14–16 hectares) and the ‘quantity and quality of the pottery, coins, glass and metal objects of the 1st to 4th centuries confirm the belief that the present village is sited upon a small Roman town’. These finds indicate that a variety of types and status of buildings existed, the discovery of flue tiles suggesting a structure with a hypocaust system. Amateur excavations in several areas of Dymock by the Malvern Research Group between 1957 and 1972 were only reported in brief and it is difficult to pinpoint their exact locations (Waters 1960–72). Apparently, they found evidence of timber-framed buildings associated with pottery dating from the 1st to 3rd centuries AD on the south side of the road. They reported a number of occupation levels, at least two bowl furnaces, a possible smithing furnace and ‘a large quantity’ of iron slag. At present any attempt at interpretation can only be conjectural, but the archive from the 1960s Malvern Research Group excavations has been located at Gloucester Museum (accession no. 2/1993), unfortunately too late for the material to be included as part of this publication. It comprises 30 boxes of finds with only one small box of site records. These may be able to add significant detail to our knowledge of Roman Dymock. Gethyn-Jones (1991) summarised his observations made over many years whilst rector of the parish. Of interest is his site no. 6, in the northern part of the graveyard, where the recovery of a flue tile suggests the presence of a property of different character to those recorded elsewhere. Sites 7 and 8 indicate stone walls present under the modern roads. Site 17, at the cricket ground, records evidence for a road junction, presumably of the roads from near Tewkesbury and Gloucester, and for pre-Roman occupation below the road in the form of ‘Belgic’ pottery and a Dobunnic silver coin, presumably a detail of the Malvern Research Group excavations. At many of the excavated and observed sites in Dymock there has been evidence for the smelting of iron. However, unlike at Ariconium (Weston under Penyard, Herefordshire) and Worcester, no slag has been noted in the excavated road surfaces (see Jackson 2003, 160; Finch Smith 1987, 313), perhaps suggesting that the road predated significant smelting activities at Dymock. There have been more recent excavations undertaken by professional teams of archaeologists. Evaluation trenches c.75 metres south-west of the sewage works on land behind the Old Forge Garage indicated the likely presence of Romano-British buildings (Derham 2001). At the western 04_BGAS125_131-246 16/11/07 15:49 Page 136

136 TOBY CATCHPOLE, TIM COPELAND AND ANN MAXWELL

end of Dymock, on land adjacent to Rose Cottage and Winserdine, evaluation trenches revealed two stone-built structures and several ditches relating to the Romano-British period, together with quantities of iron-working slag (Parry 1993). Subsequent limited excavation there did not find features that could be related specifically to settlement or industry. The earliest material comprised sherds of a large coil-built Malvernian vessel, in association with a small amount of calcined bone. These may represent part of a late Iron-Age cremation urn (Ratkai 2001). Two inhumations were also found in this location – an adult, probably in a wooden coffin, and a crouched child or juvenile with possible grave goods. The pottery assemblage is dominated by material dating to the 1st and 2nd centuries AD, especially to the period before AD 120. The tap slag was found within the fills of the earliest features, which indicates that iron smelting was being undertaken in the early Roman period (Tavener 2001). Both excavations, particularly that at Rose Cottage and Winserdine, will be referred to in more detail below. Dymock has, unsurprisingly, received the attention of metal detectorists in recent years, and local farmers and residents are aware of many finds being taken away, including coins and brooches. Most of the finds appear to have come from the area of the agger east of the cricket field, and the date range of the coins spans the 1st to 4th centuries AD, with most from the 3rd century. The brooches are mainly 1st-century, with a few from the 2nd. Two Dobunnic coins have been found recently on the south-western edge of the present village.

The Countryside around Roman Dymock There are no field names in the centre of the parish that suggest Roman occupation. However, there are a few further out, such as ‘Blacklands’ just over a kilometre to the south-west and ‘Chester Meadow’ 2 kilometres to the north (Gwatkin 1992). A mention of ‘Blacaker’ c.1240 (Smith 1964, 171) may indicate that remains of earlier occupation were visible at that period, perhaps at the ‘Blacklands’. ‘Chester Meadow’ lies next to the river Leadon at a crossing known in the late 18th century as Chester’s bridge. The crossing, presumably that known as Chester ford in the late 14th century, is not far from Roman occupation in Donnington (Ju ica forthcoming). At Donnington, c.2.5 kilometres north of Dymock, the remains of a possible Roman circular building constructed of uncoursed masonry were discovered and destroyed by the digging of a gravel pit in 1906. This has been interpreted as a well. Amateur excavations there in the 1950s apparently revealed a wall more than 40 ft. ( c.12 m) long, with ceramic roof tile fragments, tesserae, and 2nd- to 4th-century pottery (Gethyn-Jones 1991, 91–2, 95, 98). Unfortunately the exact location of the trenches in relation to the well is unknown, but the finds imply the presence of a high-status structure. Aerial photographs of the area in 1975 and 1996 (Herefordshire SMR 3713) show cropmarks of a number of features immediately north of the possible Roman well. These could be a complex of Roman buildings overlying earlier enclosures. Even though there is no evidence of contemporary field systems, the quality of the land surrounding Dymock suggests that agriculture will have been a major occupation in the Roman period and that the settlement could have been self-sufficient in food. 04_BGAS125_131-246 16/11/07 15:49 Page 137

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 137

Excavations at the Sewage Treatment Works, Dymock, Gloucestershire, 1995

By TOBY CATCHPOLE

With contributions by Marianne Cole, Hilary Cool, David Dungworth, Rowena Gale, Peter Guest, Claire Ingrem, Julie Jones, David Mullin, Joanna Richards, Jane Timby, Felicity Wild and David Williams

INTRODUCTION

Archaeological fieldwork was carried out by Gloucestershire County Council Archaeology Service for Severn Trent Water Ltd on the site of an extension to the sewage treatment works at Dymock. The works are to the east of the modern village just above the southern edge of the floodplain of the river Leadon. They are reached from the B4216 Ledbury road, via Longbridge Farm (Fig. 3). The site is situated on Silurian siltstones and mudstones of the Raglan Mudstone formation close to alluvial deposits from the Still House stream and the Leadon. Deposits of Head and a gravel terrace are located immediately to the east of the stream (BGS 1988), perhaps explaining the variation in the natural deposits encountered. Ground levels rose from c. 30 m above OD at the north-eastern extent of the excavation to over 33 m OD above at the south-western. When the sewage treatment works were constructed Gethyn-Jones (1991, 94) made a brief record of the Romano-British material that was uncovered: In the spring of 1951 finds were reported at the site of the new filter-bed. The contractors stopped work for three days; the soil already scooped out was sifted, the face of the pit was probed, and an exploratory trench was cut on the south side with the help of many willing hands. Few objects were stratified and it was mainly a case of salvage and recording; in the trench and pit, the bulk of the specimens came from a band 18 to 48 in deep. When the contractors resumed a careful watch was maintained. Thousands of fragments of pottery, numerous oyster shells and bones, several metal objects and bone pins, and much iron-slag were found.

The current location of these finds has not been ascertained. As a response to proposals to extend the sewage works, an evaluation comprising the excavation of four trenches was carried out in January 1994 (Emmanuel 1994). Subsequent open area excavation took place in two areas in January and February 1995 (Fig. 3). The main area, located to the south of the existing sewage works and centred on OS Nat. Grid SO 70243123, measured 48 m E–W by an average of 25 m N–S. The smaller area was located to the east of the sewage works and measured 18 by 14 m, centred on SO 70303 126. The surviving upper surface of the natural deposits sloped down by c. 1.75 m from south-west to north-east in the main excavation area. The north-east area was much less steep, dropping by only 0.15 m from south-west to north-east. To the north of both areas, the ground fell away steeply 04_BGAS125_131-246 16/11/07 15:49 Page 138

138 TOBY CATCHPOLE

R ive r Le adon

30m

35m Longbridge Farm Sewage De ve are lop treatment North-eastern a me nt works excavation The area S

t o Rectory n e 40m b Soakaways e r r o d w a o P R la c y rr e St Mary's Church u Main exc b avation d area L e

6 1 2

B 4215

Key m ea Str Areas excavated se ou Known l H 35m alignments of Roman roads til © CrowPnocsospsyibrilgeht. All rights reserved. S Gloucestershire County Council 100019134 2004 © Crown copyright. All rights reserved. Gloucestershire County Council 100019134 2004 0 100m

Fig. 3. Excavated areas at the sewage works and Rectory.

to the floodplain of the river Leadon, although the topography had been much altered during construction of the sewage works. The natural deposits encountered varied across the site. They mostly comprised reddish sandy clay containing bands of sandy gravel of variable depth. The upper parts of the clay had an indistinct interface with the overlying plough soil. The gravel deposits were thicker and more extensive in the north-eastern part of the site. Truncation resulting from post-medieval ploughing was also encountered to varying extents across the site, particularly in the north-east area. Excavation was carried out in difficult, wet, winter conditions. Even when it was not raining, features were difficult to identify because of the similarity of fills and the natural subsoil. These conditions restricted the amount of excavation possible. Structures A and G and all graves were fully excavated. All linear features and other structures were sampled, with the exception of ditch 04_BGAS125_131-246 16/11/07 15:49 Page 139

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 139

[1253]. The majority of the isolated pits and posts were sampled except where they were clearly of post-medieval or modern date, leaving several pits in the south-eastern part of the site and some smaller posts and stakes associated with the gated entrance through the Phase 1 enclosure unexcavated.

THE EXCAVATIONS

Phase 1: late 1st – early 2nd century The major features recorded comprise the southern and eastern boundary ditches of a 1st-century enclosure, and the group of rectangular timber buildings which it contained. The north-eastern part of the enclosure had been destroyed during the construction of the sewage works but the remainder should lie preserved under the field to the west of the excavated area.

The Enclosure A rectilinear enclosure was indicated by three ditches, [1087], [1156] and [208/1267] (Figs. 4 and 8a–c). Its southern arm was represented by [1087], which was evident over a distance of 17 m aligned NW–SE across the south-west side of the main excavation area. It measured 2.25 m in width and 0.8 m in depth, with a steeply cut slope on its south (outer) side and a much more gradual slope on its north side. The majority of the feature was filled with a single homogenous deposit (1086) interpreted as the result of rapid backfilling. (1086) sealed a thin primary silting (1109) datable only broadly to the later 1st century. An upper fill (1088) probably filled a depression resulting from slumping of the main fill. It is assumed that [1087] continued north-eastwards as [1156], although they did not meet within the excavated area. [1156] ran SW–NE to a rounded terminal, which formed the south side of an entrance, c. 18 m from the site boundary. It measured 2.5 m in width and 1.1 m in depth. The majority of the feature contained a single homogenous fill (1131), interpreted as the result of rapid backfilling and dating to the late 1st century. The shallow upper fill (1130) is interpreted as either natural or deliberate filling of a depression resulting from slumping of (1131), and pushes the final backfilling of the feature into the mid 2nd century (Timby, below). The alignment of [1156] was continued by ditch [208/1267] to the north of the entrance. It measured 2.5 m in width and 1.7 m in depth and had a flat base 0.5 m in width. Seven individual fills were recorded, the lowest (220) being a silty clay with no finds, while the upper six fills all produced material of the late 1st century as well as fuel ash slag and a fragment of iron ore. Although the fills differed from those in the rest of the enclosure ditch it is thought that this represents a localised change in the materials used to backfill the ditch, rather than any difference in phasing or function. Both [1156] and [208/1267] were cut more steeply on their outer than inner sides although not to the marked extent visible in [1087]. There was no surviving evidence of any bank or rampart associated with the enclosure ditches.

The Gate The east-facing enclosure entrance between the terminals of ditches [208/1267] and [1156] appeared to have been closed by a timber gate, with fence lines behind the ditch to either side. The gate posts were between the ditch terminals not behind them, as would have been the case if a rampart were present. 04_BGAS125_131-246 16/11/07 15:49 Page 140

140 TOBY CATCHPOLE s e r s u n t o a i 1 h e s 4 f c u 1 r n 1 n t r

e 2 3 n r i e 4 t

7 h 1 d / 1

d c 1 o 1

n n n 1

a m e o

i r

t t s & a e

l n v g

o i a

v t c

e a x i

e u

e n d

r l i s

f e

u a a n

s v o r m o o

- i E l e e t t h c c s g c c t n n i o d e F E d P E U S ? X e

r X u 1 t 6 c 8 2 u 0 r 1 t 2

/ c S

7 8 X

6 . 5

2 g i 8

1 F 2 1

4 X 1 5 6 7 6 e 2 t 1 2 1 1 a 1 7 3 G 6 m 2 6 5 1 2 1 9 7 6 e 2 r 5 1 3 u 1 7 2 1 s 2 0 8 s o 1 2 l 8 h s 1 2 c c e t 1 n i c E d c 0 a 0 t 2 0 1 n 2 a l 1 7 7 p 2 1 2

1 X 3 n 1 0 o i 1

C b t 8 7 c h . 2 c u

t g . e 2 i r i r t ) 1 d F s u l d 6 t n a 5 c X e v o 1 u 2 e d 1 r i C 5 t a d 0 S e 5 h 1 D 8 5 s m 8 1 - ( 2 t e 1 4 2 3 r s s 1 3 1 8 o u 1 2 t 1 n X p 2 c 1 4 o 1 u i 5 r d t 0 0 X t 0 8 9 1 a 0 S . 2 3 v 1 g 1 1 i 1 3 a 1 7 F 8 1 c 5 0 5 1 x 0 5 1 0 1 1 e 2

X 1 1 1 8 2 n 9 1 8 i 1 8 X 1 e a 1 2 1 8 6 1 . 0 m 6 6 8 g 5 1 A i 2 8 8 5 F n 1 2 1 e 1 i 1 r 1 1 1 8 u s t 6 e c 2 r u 1 r u t 8 t S 0 0 6 9 a 2 1 6 4 6 3 e 5 1 1 9 0 1 2 8 f 1 9 0 2 1 1 1 0 3 7 1 1 1 1 1 1 7 1 1 1 1 3 1 e e 3 6 r s 0 7 u a 1 1 s 9 1 h 6 o l h 0 P c c 1 t n : i 7 s 8 E d 4 0 k 1 1 r 2 6 o 1 1 2 w e 1 r e u t g n c o a i u t 1 r G w t a B h 0 e u S 7 l c e a 9 S r 2 a n 8 4 7 0 1 u v . 9 e 8

t 5 2 r g 2 0 E t i c 9 1 5 7 1 2 1 F

u . 9 0

r 1 1 t X 2 4 1

S X X .

X g N i . g F i f F 8 04_BGAS125_131-246 16/11/07 15:49 Page 141

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 141

The gap between the ditch terminals measured 6 m. The features interpreted as gate post pits were [1237] on the north side and [1264] on the south side. The gate so formed would have been no more than 2.7 m in width, assuming posts 0.5 m in diameter placed at the outer extremities of the post pits. Significant spaces would have been left on either side between the gate posts and ditch terminals and it is uncertain how these gaps were closed. It is possible that barriers were attached to the outside of the gate posts and that these left no archaeological trace. The only dated feature in the gateway area was [1237], which was backfilled in the late 1st century. A further unexcavated post [1261] was located c. 3.5 m outside the centre of the gateway. A linear slot representing a fence line [1200] with associated posts [1271], [1272] and [1273],was located inside ditch [1156] on the south side of the entrance. A second shorter fence line [1266] with stake [1285] and post [1265] was located inside ditch [208/1267] on the north side of the entrance. Several other posts and stakes were located to the west of the gate. These may indicate that a gatehouse structure was present, although no clear pattern has been identified to explain what form this may have taken. If however, as proposed below (Fig. 7), several of the cut features in the area inside the gate relate to the north-east corner of Structure D, there is no need to suggest a gatehouse, merely a fence represented by a line of stakes running from the south-west side of the gateway to the east side of Structure D.

Timber Buildings A group of rectangular timber buildings stood inside the enclosure, only one of which (Structure A) was well defined. Other adjacent buildings were either partly outside the excavation or survived only in part. The evidence for the buildings comprised postholes and narrow gullies with flattened ‘U’-shaped profiles and flat bases.

Structure A Structure A was rectangular, aligned SW–NE, its sides parallel with enclosure ditches to the south and east (Figs. 4 and 8d–e). Its west side measured 22.5 m in length and it was an average of 4.5 m in width. There were internal partitions, one, [1061], 16.5 m from the south end and a second, represented by construction trench [1268] and posthole [1269], 6 m from the same end and apparently continuing the alignment of Structure B. The structure was built on a slope, the surviving levels dropping by 0.81 m from the south-west to north-east corner (Fig. 5). The majority of the structure was represented by a single construction trench cut by later features. The elements of the trench, given different context numbers ([1069], [1146], [1148], [1061] and [1155]), were ‘U’-shaped in section and 0.4–0.5 m in width and they survived to depths of 0.2–0.4 m. Irregularly spaced posts and stakes were excavated in the base of the trench but these only extended as far north as the northern side of adjacent Structure B. The east side of construction trench [1148], at the north-east of the building, did not align well with the rest of the building to its south. Construction trench [1155], which represented the majority of the east side of the structure, turned westwards as [1061] rather than continuing north towards [1148]. An ephemeral drip gully, [1150] and [1188], was identified to the east of the structure. The fact that it appeared to end at pit [1054] may be of relevance to the interpretation of the relationship between Structures A and C. An indistinct rectangular pit [1100], containing no finds, was cut through the north side of trenches [1146] and [1148], removing the relationship between these two contexts. Consideration was given to the possibility that it represented a northern entrance but as the truncated terminals 04_BGAS125_131-246 16/11/07 15:49 Page 142

142 TOBY CATCHPOLE

of the construction trenches were only 0.40 m apart it is more likely to have been a shallow pit of unknown function truncating the earlier continuous construction trench. A single phase of backfilling was identified in most of the construction trenches. The fill was a single continuous deposit, apart from in the south-east corner, and contained large quantities of burnt daub, especially over its southern half. The homogenous nature of the fill suggests that timber was systematically removed and the site of the building cleared. A better preserved sequence was encountered on the east side of the building, to the south of post-medieval ditch [1031]. The earliest deposit in trench [1069] was a silty fill (1114) , which was cut by two postholes, [1120] and [1129], and therefore may have been contemporary with construction; unfortunately it was not closely datable and a sample taken from it provided little information. When the building went out of use a deposit of burnt daub (1083) collapsed around a line of stakes ([1105], [1106], [1107] and [1108]), preserving their outlines and sealing (1114) and postholes [1120] and [1129]. The collapsed daub (1083) was sealed by the general burnt backfill found elsewhere.

Structure B Structure B was a poorly preserved rectilinear timber structure, the outline of which was represented by construction trenches [1207/1209], [1214] and [1216] and post [1033]. The features were insubstantial and ill defined, only surviving to depths of 0.1–0.2 m. Despite the poor level of preservation at least four spaces (B1–B4) are defined and these are indicated on Fig. 6. B1 was enclosed by [1216] to the west and [1214] to the south and east, forming a roughly square space measuring 2.65 m across. The full northern extent of [1216] had not survived but the northern arm of [1214] measured 2.9 m in length. An entrance into the south-west side of B1 was suggested by the shape in plan of the construction trenches . B2 measured 2.95 × 2.9 m and was defined by [1214] to the south and west and post [1033] to the north. The presence of post [1033] and the lack of a northern construction trench to B1 and B2 suggests a roof above an open fronted structure. The eastern extent of B2 was formed by Structure A construction trench [1146] between posts [1176] and [1179]. The gap at the east end of [1214] may represent a further access to B3 to the south. B3 measured c. 2 × 7 m and was defined by [1207] to the south, [1209] to the west, [1216] and [1214] to the north and construction trench [1146] of Structure A to the east. The eastern part of [1207] had been lost and it is uncertain whether it continued as far east as Structure A. It was physically continuous with [1209]. B4 was of indeterminate extent as construction trenches [1209] and [1216] were truncated to the north and west and only [1207], which formed its south side, continued to the western edge of the excavation. It likely that [1216] turned west, continuing the line of the division between B1 and B3. A single backfill was present in [1207/1209] and also in [1214]. The fills were similar to those of the Structure A construction trenches. The fill (1215) of trench [1216] was clearly later than (1208), which filled [1209]. The deposit was stony and compacted and it was indicated by the excavator that it constituted ‘packing’, suggesting a foundation for a later structure, for which there was no further evidence.

Relationship between Structures A and B Structure B appears to have formed a block at right angles to Structure A. No physical relationship had survived between Structures A and B but the following evidence suggests that they were continuous. Structure A posts [1173], [1176] and [1179] were on the alignment of the sub-divisions of Structure B. Structure A construction trench [1268] and posthole [1269] continued the line of 04_BGAS125_131-246 16/11/07 15:49 Page 143

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 143

Fig. 5. Sewage works: Structure A during excavation looking north-east.

1146

1033 B4 B1 1179

1216 B2

9 1214 12 0 07 2 1178 1 1177 B3 Structure A

1176 1 1269 207 1268 Structure B 1174

1173

0 2.5m

Fig. 6. Sewage works: detail of Structure B. 04_BGAS125_131-246 16/11/07 15:49 Page 144

144 TOBY CATCHPOLE

[1216] and [1214] across Structure A. [1214] clearly ended c. 1.2 m to the west of Structure A, but the continuation of the above ground structure may be indicated by post [1176]. The construction trenches of Structure A were fully excavated and the posts and stakes shown on Fig. 4 indicate the full surviving extent of those features within the building. No posts or stakes were present to the north of Structure B. Therefore, it seems likely that Structures A and B were built at the same time as part of the same building complex. The function of Structure B remains unclear as no evidence was uncovered that pointed to a specific purpose. The structure resembles stabling or animal pens, B1 and B2 being only slightly larger than the 2.74 × 1.82 m required for a Roman cavalry horse (Frere and St Joseph 1974, 24), in which case B3 represents an access corridor. Less likely, given the absence of evidence for hearths or furnaces, the presence of open-fronted structures might be explained in relation to the evidence for metal working from the site as the burning of significant quantities of charcoal required well ventilated structures due to the amounts of carbon monoxide produced (Hammer 2003, 21).

Structure C This structure was represented by a ‘T’-shaped construction trench [1052]/[1113], the eastern arm of which had not fully survived. Posts ([1289] and [1290]) were located at either end of the E–W member. The layout of the feature is best ascertained from its plan (Fig. 4). It was cut by a shallow rectangular pit [1057] which produced no finds. Structure C is difficult to explain as a free-standing structure. As the ground level drops by 0.8 m from south-west to north-east across Structure A, the construction of a level boarded floor would require it to be raised at the north-east side of the building and Structure C may therefore represent the foundations of a wooden external stair giving access to Structure A. This interpretation of Structure C is complicated by the presence of a large shallow pit [1054], which had been cut into the backfilled Structure A construction trench [1148] and was itself cut by the construction trenches of Structure C. A case can be made for Structures A and C being elements of the same building if the pit is interpreted as the result of water from a roof collecting in the area. The ground is unusually level at this point, only dropping by 2 cm from the south to the north side of the pit. As drip gully [1150] runs along the east side of Structure A only as far north as pit [1054] it is possible that the pit resulted from the pooling of water, which created an area of disturbance around the base of the building and the stair. If the area had dried out before demolition of the building the construction trench might look to be later than the disturbed ground that had in fact surrounded the in-situ timbers. Pottery evidence indicates a late 1st-century date for the backfilling of the Structures A and C and the pit that separates them.

Structure D This poorly preserved structure was represented by a single ‘L’-shaped construction trench [1183] measuring c. 19 m in length, with a short return to the east at its southern end, truncated after 2 m by undated pit [1205]. It was also cut at its south-west corner by pit [1126]. Structure D was oriented SW–NE, roughly parallel to both Structure A and enclosure ditch [1156]. A linear feature, [1181]/[1286], identified to the west of the structure could have been an associated eaves drip gully. The single homogenous backfill (1182) was superficially similar to those in Structures A and B in that it contained charcoal and burnt daub. It was not closely dated as it only produced undiagnostic Severn Valley ware but there is no reason why it could not have been contemporary with the other Phase 1 features (Timby, below) and its location and alignment strongly suggest that it was part of the same group of structures. 04_BGAS125_131-246 16/11/07 15:49 Page 145

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 145

1284

1202 1283 1281 1282

1227 Structure D Structure A 1181 1183

1225

1286 1183

1223 Surviving element of Str D 1205 11 26 Possible outline 1182 0 5m

Fig. 7. Sewage works: possible layout of Structure D based on Structure A.

The only other features which could have been part of Structure D were a series of post- and stake holes to the east of [1183]. These features ([1223], [1225], [1227], [1281]–[1284] and [1202]: Fig. 7) do not fall into an immediately recognisable pattern either in being regularly spaced along the length of [1183] or in providing the width of the structure that can be plotted. It is possible, however, that Structure D was similar to Structure A, although there was not sufficient space for an equivalent of Structure C. A postulated outline for the building can be seen on Fig. 7. Structure A was far from perfectly rectangular and there is no reason to assume that Structure D was any different. Other posts could equally be suggested as part of the structure, particularly to the north- east. The suggested layout implies that post [1202] may have been part of a strip building rather than part of a gatehouse structure. 04_BGAS125_131-246 16/11/07 15:49 Page 146

146 TOBY CATCHPOLE

Structure G A large flat bottomed negative feature [1195] (Figs. 4 and 8f), only partly within the excavated area, was interpreted as a possible sunken floored building (SFB). The feature measured 2.6 m N–S and at least 1.8 m E–W, continuing beyond the west edge of excavation, and was c. 0.4 m in depth. The eastern corners were rounded, suggesting a sub-rectangular shape. The south and north sides were almost vertical, the east less so. It was cut by small pit [1270], which was thought to be later and unrelated. The base was flat and contained no post- or stake holes. Stake holes [1294] and [1295] were located by the east and north sides of the feature respectively but may not have been associated with it. The lower fill of the feature (1198) was burnt and possibly represented the remains of a timber superstructure; it was recorded as comprising mostly ash, especially at its base, and contained daub. This burnt material appeared to have been deliberately sealed by a layer of gravel (1197) to allow continuing use of the feature. The upper fill (1196) was a homogenous silty clay, typical of Phase 1 backfills. Pottery evidence suggests that the backfill of the feature is of Flavian or Trajanic date (Timby, below), and its abandonment may therefore have been slightly later than the majority of features in this phase.

Other 1st-century Features within the Enclosure A number of pits, postholes and gullies were present within the enclosure. Not all were excavated and several produced no finds. Only one was datable to the 1st century. [1185], a circular pit within the area of Structure D, was c. 1.0 m in diameter and 1.1 m in depth with vertical sides. It produced Flavian samian and an earlier stamped sherd dated to AD 50–65 (Wild, below). The feature’s lower two fills, (1186) and (1191), contained much ash and charcoal and produced one of the better assemblages of archaeobotanical remains from the site (Jones, below) and quantities of burnt brushwood (Rowe, below), suggesting its use for disposal of domestic waste.

1st-century Features outside the Enclosure A number of features outside of the enclosure, comprising posts, pits and a probable well, can be allocated to Phase 1. The evidence is poorly defined but clearly indicates activity to the east of the main enclosure. No such features were encountered north of the gateway or in the north-east excavation area, perhaps suggesting that the activity was focused on the road located as close as 12.5 m to the south of the enclosure ditch (discussion, below). Pit [1143], c. 1.5 m in diameter, may have been a well. It was excavated to the modern water table, encountered at a depth of 1.25 m. Only a single backfill (1144), dated to the 1st century (Timby, below), was encountered. [1117], a small circular pit or posthole, dated broadly to the 1st century. It contained burnt material together with rodent and frog bone (Ingrem, below) and small amounts of tap and fuel ash slag. An adjacent small shallow pit, [1141] containing only Severn Valley ware, similarly had a fill comprising mostly ash and charcoal. [1171], a small oval pit or post, was clearly dated to the late 1st century. It had been backfilled with refuse which included sherds of Flavian samian and tap slag. The north and west sides of a possible structure (F) can be formed of three posts and a small linear feature but no feature indicating a south-east corner was noted. Only the linear feature [1243] produced any finds, dating it only broadly to the late 1st–early 2nd century (Timby, below). It is worth noting that small post-built structures were typically intended to keep rain from smelting furnaces (Hoyle et al . 2004, 103), although there was no further evidence to support such an interpretation in the case of Structure F. 04_BGAS125_131-246 16/11/07 15:49 Page 147

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 147

S N NW SE 32.08 (1130) 33.43 (1203) (1012) (1019) (1088) (1131) (1084) [1020] (1086) [1156]

(1109) [1087]

8a. E facing trench section across enclosure ditch [1087] 8b. SW facing across enclosure ditch [1156]

W E

(201) (top soil)

[205] (202) (plough soil) tree 32.24 (202) root? (209) modern drain pipe (203) (natural) (214) (204) (natural) (215/6) (218) (221) [208] (219)

(217)

(220)

8c. S facing section across enclosure ditch [208]

NE SW E W 32.24 32.36 (1147) (1154) [1148] [1155]

8d. NE facing across structure A construction trench [1148] 8e. N facing across structure A construction trench [1155]

SN 33.11 (1196) (1197) (1198) [1195]

8f. E facing section across structure G [1195]

0 1m

Fig. 8. Sewage works: Phase 1 sections.

Phase 2: mid 2nd century Phase 2 comprises features post-dating the backfilling of the Phase 1 enclosure (Figs. 9 and 10). Three features, pits [1038], [1043] and [1007], are thought to have been used for the disposal of Phase 1 material but are themselves later in date, possibly indicating a short hiatus followed by 04_BGAS125_131-246 16/11/07 15:49 Page 148

148 TOBY CATCHPOLE

l u

t i o

t r

1026

Fig. 12a post-medieval X X ditches Robber Pit 1043 pit1036 405

Pit 413 1055 1007 Rob ber tren ch 1 073 Robber trench 1010 Structure H

0 5m

Fig. 9. Sewage works: all features in north-eastern excavation area.

clearance of the site in the mid 2nd century. Ill-defined ditches [1020] and [103]/[1124] may also have served the same purpose when backfilled.

Pit containing Bronze Casting Debris

In the north-east excavation area a large amorphous pit [1043] measuring 4.9 × 2.0 m in plan and 0.65 m in depth included debris from the production of bronze brooches and other tools (Figs. 9 and 12a). The pit is interpreted as having been originally dug to provide gravel but finds from its backfill included three brooch moulds and a mould of less certain type, furnace lining, tap slag, and iron ore (Dungworth, below), and a terminus post quem ( t.p.q. ) for the feature was provided by Central Gaulish samian form 33, thought to be of Antonine date (from AD 138) (Wild, below). There was strong evidence that one of the moulds was used to produce an unstratified Chester type trumpet brooch, of late 1st–early 2nd-century date, from the excavation (Cool, below: catalogue no. 7: Dungworth, below: Rf. 23). The two other brooch moulds from the feature were probably used to produce Colchester derivative brooches of late 1st-century date (Dungworth, below). This suggests that datable bronze working activity is contemporary with the structures excavated in the main excavation area, rather than later, as implied by the t.p.q. for the pottery from the feature in which the moulds were found. 04_BGAS125_131-246 16/11/07 15:49 Page 149

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 149 t x m e t 5 y e r h u t 0 t 1 0 n n 1 i 5 e 1 c o 1 t d D d n A e 2 r r - r e d e t i f 4 f e 0 0 m a r 1 , , 1 s e s s 9 v e e e e r v 5 a r r 9 r a 2 u 8 u u r t t t G 1 1 a G 1 a a e e e f f f d 2 3 e t e e 1 a s s 2 d a a y 2 n h h e

P U P K 3 E

5 e e

2 v 2 r

4 a 1 r u

2 t

G 1

t S 5 4 2 1 8 6 1 1 e 5 v 5 a r 2 1 G 2 2 . 1 1 a e e r v 4 a 9 a r 1 1 G n o i t a v a c x e n . i b g i 2 a F 1 8 5 m 1 e 1 v X e 3 a 5 h r 1 t 1 G 1 n 6 i X 5 1 0 1 s 2 e 1 r u t 6 a 2 e 1 f 1 r 3 e 2 t 0 a 1 l 8 e 3 v d 0 a r n 1 a G 2 e s e a 1 v 8 a h 4 r 0 3 P G 1 0 : 1 4 s 2 k 1 r 1 o w 3 e 0 g 1 a 9 2 w 2 8 4 e 0 n 1 0 1 S o 2 1 i t 1 1 a 0 . h 2 u l c 0 0 a n 1 N 1 v e r . E t g i F 04_BGAS125_131-246 16/11/07 15:49 Page 150

150 TOBY CATCHPOLE

The lowest fill of the feature (1066) comprised much charcoal, including pieces of small roundwood thought to have been used in the production of bronze items (Gale, below), as well as a mould for a lunate fitting (Dungworth, below). The deposit is therefore interpreted as waste from metal working. The upper layers within the pit contained far less charcoal but produced the three brooch moulds already mentioned, daub and tap and undiagnostic iron slag. Pit [1043] was cut by a shallow and truncated pear-shaped pit, [1007]. The fill (1005) produced pottery dated to the early 2nd century but stratigraphically it was later than [1043] and therefore must have been mid 2nd-century or later in date. Careful consideration was given to the feature during excavation since its shape suggested that it was a possible hearth or furnace, the narrow north-west extension of the feature representing a flue or tapping pit. No evidence was found to support such an interpretation; there was no indication of in-situ burning and it was one of the few features that produced no slag. The pit was thought more likely to have resulted from slumping of the fills of pit [1043] below.

Inhumations Five adult and three infant inhumations were excavated (Figs. 10 and 11). Where datable they were of mid 2nd-century or later date (Timby, below). Infant graves [1023] and [1081] were cut through the backfill of enclosure ditch [1087]. [1023] also cut the spread upper fill of Phase 2 pit [1038]. The adult inhumations were all outside the Phase 1 enclosure, suggesting that the enclosure was still considered an area inappropriate for adult burial. All graves were shallow. No grave goods were retrieved except coffin fittings and hobnails.

Adult Inhumations Skeleton (1241) in rectangular grave [1242] was aligned N–S. It was supine, with its head to the north, its arms were folded, the left leg was straight with the right leg bent across it and the chin rested on the left shoulder. The fill around the skeleton (1232=1240) produced a small fragment of iron ore and coffin nails (Cool, below: catalogue no. 58). The grave cut ditch [1104]. Skeleton (1258) in grave [1259] was aligned NW–SE. It lay on its left side, with its head to the north-west, the right arm flexed, the left arm bent underneath the ribs, and the legs straight with the right foot on the left. The skull was damaged during machining. No evidence for a coffin was present in fill (1257). (1258) may have been paired with (1190), an adjacent skeleton on the same alignment in grave [1189]. (1190), with the head to the north-west, was in poor condition, lying crouched face down with its left arm around the neck, the right arm bent and pointing upwards and the knees tightly bent with the right leg over the left. Fill (1170) produced 36 registered finds, suggesting a slightly more ornate coffin (Cool, below: catalogue no. 56) as well as tap and undiagnostic iron slag. The relationship of the grave with ditch [1104] could not be determined. Skeleton (1193) in grave cut [1194] and skeleton (1167) in grave [1168] lay adjacent and on the same N–S alignment. (1193) was supine with the head to the north, the arms were folded across the waist and the legs were straight. The skeleton was in good condition except for the spine and ribs. The fill (1192) produced coffin nails and a number of hob nails (Cool, below: catalogue nos. 13 and 57) as well as several pieces of iron slag. Skeleton (1167) was supine, the upper body at the north had been ploughed away. The legs were bent over to the left and the left leg was also damaged. The fill (1166) produced coffin nails and a number of hob nails (Cool, below: catalogue nos. 14 and 55). 04_BGAS125_131-246 16/11/07 15:49 Page 151

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 151

(1241) (1258) (1190)

(1193) Infant (1080)

Fig. 11. Sewage works: inhumations.

Infant Inhumations [1081], a small sub rectangular grave, contained infant (1080) and cut backfilled boundary ditch [1087]. The skeleton was in poor condition, aligned E–W with the head to east and lying on its left side in a crouched foetal position. Circular grave cut [1023] contained infant (1021). It cut the backfills of pit [1038], which was cut through backfilled boundary ditch [1087]. (1021) was in poor condition, aligned E–W with the head to east and lying on its left side in a crouched foetal position. Infant (1152) was in circular grave cut [1153]. It was in a very fragile state, aligned SW–NE with the head to the south-west and lying on its left side in a crouched foetal position. 04_BGAS125_131-246 16/11/07 15:49 Page 152

152 TOBY CATCHPOLE

Other Features In the main excavation area (Fig. 10), feature [1038] was either an amorphous pit cut into the backfill of the Phase 1 southern enclosure ditch [1087] or a concentration of material filling a hollow caused by slumping of the ditch fill. As excavated it was only 0.20 m in depth by c. 1.00 m in diameter, but it produced a very rich assemblage apparently of material cleared from the burnt Phase 1 structures. The lowest fill (1059) appeared to be a layer of decayed timber, possibly the remnant of a wood lining or discarded timber from the adjacent structures. The upper two fills were remarkably rich. (1045) produced 70 pottery sherds dating it to the mid 2nd century (Timby, below), the only stratified coin from the excavation (a sestertius of Domitian: Guest, below), the only well-preserved assemblage of cereal grains from the site (Jones, below), charcoal interpreted as fuel debris from iron working (Rowe, below) and much burnt clay. (1037), the top fill, spread over the sides of the feature and produced 38 sherds of pottery (see Fig. 14, nos. 20 and 21), a mould for the production of a conical headed pin (Dungworth, below: Rf. 17), a spindle whorl (Cool, below: catalogue no. 16) and tap slag. Sub-rectangular pit [1122], may also be of this date. It was cut into the backfills of the eastern enclosure ditch [1156] and contained a single backfill, comprising mostly charcoal and burnt clay/daub. Ditch [1020] was recorded in the south-west part of the main excavation area, but was not visible to the east of evaluation trench 1. Its eastward continuation may have been truncated or it may be represented by gully [103]/[1124], which is broadly contemporary. [1020] was cut into the backfill of boundary ditch [1087] (Fig. 8a). The lowest fill was silty with few finds but the upper backfills, (1084), (1019) and (1203), produced large amounts of pottery, including samian of AD 120–130. Small finds from the feature included a penannular brooch and hairpin (Cool, below: catalogue nos. 8 and 12), of earlier date than the pottery and thus a further instance of mid 2nd-century features containing residual 1st-century finds. The upper fills contained a high proportion of ash, daub and charcoal. Gully [103]/[1124] also cut the backfilled enclosure ditch [1087]. It produced a significant pottery assemblage of mid 2nd-century date, together with a large amount of charcoal and burnt daub. The gully apparently merged with [1286], a series of drip gullies to the west of Structure D, although it is unlikely that they were contemporary. Their relationship was not explored. On the site of the Phase 1 Structure D was [1161] (Figs. 10 and 12b), an irregular shallow-sided and clay-lined feature interpreted as a hearth or oven. The charcoal rich upper fill (1159) produced an iron goad, a couple of fragments of ceramic building material and two sherds of Severn Valley ware. It therefore could not be closely dated but it was cut by [1158], a very similar oval pit without a clay lining but containing a burnt fill and pottery of mid 2nd-century date. These features were similar in form to casting hearths (Hammer 2003, 16–20) but, in the absence of conclusive evidence, their purpose is unknown. The stratigraphic relationship with undated pit [1205] was unclear. [1034] was a shallow gully located in the north-western part of the main excavation area. No more than 50 mm in depth, on site it was interpreted by its alignment as a medieval or later plough furrow but it produced 74 sherds of Roman pottery, dating the feature to the early 2nd century, as well as 1077 g of tap slag and a small amount of other slags. [1029] was an irregular pit, not closely dated, adjacent to the west end of [1034]. The fill (1028) produced a fragment of iron ore and some daub. A small adjacent posthole [1042] produced no finds. Despite the finds retrieved from them the forms of this group do not resemble metal-working features recorded elsewhere. [1034] was compared with known roasting hearths but was too shallow, displayed no sign of burning and contained tap slag and not crushed ore. A few other excavated features were firmly dated to this phase but were isolated and produced no clear evidence of function. 04_BGAS125_131-246 16/11/07 15:49 Page 153

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 153

E W 30.05

+ + + (1044) + + + + + + + + + (1064) +

+ + (1065) + + + + + + + + + + (1066) + + [1043]

12a. N facing section across pit [1043]

SE NW

32.60 + + (1159) (1256) + + (1157) + + + + + + clay lining (1160) [1158] [1161]

12b. NE facing section across pits [1158] and [1161]

0 0.5m + = Charcoal concentration

Fig. 12. Sewage works: Phase 2 sections.

Possible Secondary Enclosure A group of three ditches, [1253], [1245] and [1104], may represent an enclosure of this phase. Ditch [1253] ran NNE–SSW immediately to the east of Phase 1 enclosure ditch [1156]/[208] and across its entrance. The northern extent of [1253] is unclear. It appeared to cut Phase 1 ditch [208] although it was indistinct throughout and was not noted at all during the evaluation when the southern terminal of [208] was excavated. It is possible that feature [221], identified as animal burrow disturbance on the east side of [208], represented the ephemeral remains of this ditch. It was truncated to the south by post-medieval ditch [1031], but its continuation was probably represented by ditch fragment [1245]. This was truncated to the south so that only 1.2 m of its length was apparent. It produced significant amounts of 2nd-century pottery. Ditch [1104] ran NW–SE in the south-east part of the main excavation area. Truncated to the north-west by post-medieval ditch [1031], it was at right angles to, and may have terminated at, ditch [1253]. Two excavated segments of the feature produced no datable finds but its backfill was cut by a mid 2nd-century grave. The enclosure formed by [1104], [1245] and [1253] was on almost the same alignment as the Phase 1 enclosure but both would have been roughly parallel to the road to the south and the earlier enclosure may have not determined the layout of the later.

Phase 3: Roman activity after AD 150 Evidence for activity at this date is restricted to a single pit and an undated structure robbed out in the later Roman period. 04_BGAS125_131-246 16/11/07 15:49 Page 154

154 TOBY CATCHPOLE

Structure H Ephemeral rectilinear structure H (Fig. 9) was located in the north-east excavation area. On the north side an unstructured sandstone gravel and rubble deposit (413), 0.52 m in width, was recorded during the evaluation. It was interpreted as a sub-base for a foundation in cut [412]. No other in- situ remains of the structure survived below robber trenches. The majority of the structure was indicated by those trenches: [1010], [405]=[1073] and possible quoin or post robbing pit [1036]. More substantial sandstone foundations were suggested by the rubble in the backfills of the robber trenches and in small spreads within the post-medieval ploughsoil to the north-west of the feature. As no finds were retrieved from (413) the date of the original structure is unclear. Its alignment was similar to Roman features in the main excavation area rather than to those of post-medieval features in both trenches. Only the west and north sides of the structure were within the excavation trench, the area of the structure within the excavation measuring 10 m ESE–WNW by 5 m NNE–SSW. The western half of the structure’s north wall could not be located. Either there was a gap in the building wall at this point or, more likely, the robber trenches had been entirely truncated. Where excavated the robber trench on the north side of the structure was 0.15–0.20 m in depth and that on the west side 70 mm–0.14 m in depth. The nature of the structure is hard to clarify. It had been almost entirely robbed out and the robber trenches had then been heavily truncated. On the west side of the structure it was uncertain if the robbing had been made through the post-medieval ploughsoil or predated it. With the exception of a single sherd of (presumably intrusive) modern china the latest material from the robber trenches was Roman, of 3rd-century or later date (Timby, below). The other finds from the robber trenches included 1st–2nd-century pot, tap slag, iron ore, daub and a single sherd of Central Gaulish stamped samian dating to the Hadrianic or early Antonine period, all of which have to be regarded as residual as do similar finds from the grid of post-medieval ditches recorded in the trench. The Phase 2 pits [1007] and [1043] shared the alignment of, and had an uncertain relationship with, Structure H. It seems probable that they were built over by it, but this could not be demonstrated stratigraphically. In the main excavation area, a circular pit, [1126], was cut through the south-west corner of the backfilled Structure D construction trench. The fill, (1125), produced 3 fragments of tap slag but they may well have been residual as the fill also contained samian of both pre-Flavian and late 2nd- century date. The feature produced no evidence of function beyond its final use for rubbish disposal.

Undated and post-Roman Activity

Structure E (Fig. 10) was very small (1.3 × 0.8 m) and formed by four posts. All were excavated but no datable finds were retrieved. The fill of one post produced a small amount of tap slag. A small oval pit, [1255], was in close proximity but also produced no indication of date and function. A fragment of a pit, [1218] located to the south of Structure B, was truncated to the east and south. The fill (1217) produced no finds. It appeared to have been burnt in situ but contained no burnt grain (Jones, below). A scatter of isolated post- and stake holes was also present. They were either not excavated or produced no finds.

Medieval and Later Features The only recorded medieval feature was a gully producing 14th–16th-century pottery. It was identified in evaluation trench 3 outside the excavation areas. 04_BGAS125_131-246 16/11/07 15:49 Page 155

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 155

Post-medieval ditches were mostly on N–S or E–W alignments (Figs. 4 and 9). They included E–W ditches along the northern edge, and [1031] across the width, of the main excavation area. A grid of closely spaced post-medieval ditches recorded in the north-east excavation area yielded much residual Roman material and slag. It included ditch [1055] which produced several of the illustrated Roman sherds. Several shallow E–W aligned features were interpreted as remnants of post-medieval plough furrows. A modern fence line and square pit were recorded in the north-western part of the main excavation area near Structure A. A large modern pit was also cut into enclosure ditch [1156].

THE FINDS

Pottery by Jane Timby with notes on the samian by Felicity Wild and on an amphora sherd by David Williams

An assemblage of some 2,865 sherds of Roman pottery weighing c 67.65 kg was recovered. In addition a small group of 95 sherds dating to the medieval and post-medieval periods was also found, largely from topsoil/ploughsoil or cleaning contexts. The part of the assemblage from the latter contexts was significant, accounting for 34 per cent of the total assemblage by count and 41.65 per cent by weight. A further 8 per cent of the Roman sherds came from post-Roman features. Overall the pottery shows a date range spanning the pre-Flavian period through to the 3rd century AD. The sherds were well preserved with a good average sherd weight of 23.6 g. There are a number of joining sherds and some reconstructable profiles. The pottery was sorted into fabric types on the basis of the type, size and frequency of the inclusions. The fabrics were coded according to either the Gloucester City type fabric series (Ireland 1983) or the National Roman reference collection (NRFC) (Tomber and Dore 1998). Fabrics not familiar to either series and specific to this assemblage are described separately.

Description of Fabrics and Associated Forms Pottery fabrics are summarised on Table 1, which includes a list of the abbreviations used in the text.

Continental imports: finewares SAMIAN by Felicity Wild The site produced 62 sherds from about 52 vessels, ranging in date from the pre-Flavian period to the late 2nd century AD (see Table 2). Forms were as follows:

South Gaulish : 29 (5), 37 (2), 27 (2), Ritt. 8? (1), 18 (3), 18–18/31 (2), 35/36 (1), Ritt. 12 or Curle 11 (1), uncertain cup (1), enclosed vessels (3), dish (1). Central Gaulish (Les Martres-de-Veyre ): 37 (2), 24/25 (1), 46 (1), dish (cf. Lud. Tb) (1) Central Gaulish (probably all Lezoux ): 37 (3), 33 (4), 18/31 (4), 18/31–31 (1), 31 (1), 31R (1), 35 (1), 42 (1), 79 (2), bowl (38, 44 etc.) (1), scraps (6).

Although the assemblage is too small for statistics to be reliable, 22 vessels, 42 per cent of the total material, were South Gaulish and of 1st- to early 2nd-century date. The earliest material is likely to have been made in the pre-Flavian period: the base of a small cup from Structure A construction 04_BGAS125_131-246 16/11/07 15:49 Page 156

156 TOBY CATCHPOLE

Table 1. Sewage works: quantification of pottery by sherd count, weight (in grams) and estimated vessel equivalents. Fabric NRFC Description No. % Wt. % EVE % Imports 8A Central Gaulish samian 38 1.3 466 0.7 88 2.2 8B South Gaulish samian 24 0.8 213 0.3 52 1.3 12Q KOL CC Cologne colour-coat 4 0.1 17 0.0 0 0.0 12S ARG CC Argonne colour-coat 1 0.0 2 0.0 0 0.0 10 Dressel 14 1 0.0 55 0.1 0 0.0 10A BAT AM Dressel 20 amphora 15 0.5 1,807 2.7 0 0.0 9C NOG WH North Gaulish mortaria 1 0.0 96 0.1 8 0.2 29 ?imported white ware 1 0.0 17 0.0 0 0.0 Native ware 2 hm grog-tempered 92 3.2 1,016 1.5 24 0.6 18 MAL RE Malvernian rock tempered 76 2.7 995 1.5 122 3.0 30 sandstone-tempered grey ware 11 0.4 151 0.2 27 0.7 31 shale and limestone/calcite 6 0.2 23 0.0 0 0.0 33/216 Malvernian limestone-tempered 27 0.9 304 0.4 55 1.4 L00 Jurassic limestone-tempered 1 0.0 4 0.0 0 0.0 Regional 3 ?Wiltshire mica-slipped 11 0.4 96 0.1 50 1.2 4 DOR BB1 Dorset black burnished ware 132 4.6 1,702 2.5 299 7.4 6 SAV GT Savernake ware 10 0.3 237 0.4 41 1.0 9A OXF WH Oxon whiteware mortaria 1 0.0 24 0.0 0 0.0 9F VER WH Verulamium whiteware mortaria 1 0.0 122 0.2 9 0.2 12A OXF RS Oxon colour-coated ware 6 0.2 30 0.0 0 0.0 13 OXF WH Oxon white ware 6 0.2 65 0.1 0 0.0 15 SOW OX SW oxidised ware 4 0.1 44 0.1 8 0.2 19 Malvernian ware 19 0.7 273 0.4 16 0.4 201 Wiltshire black burnished ware 30 1.0 357 0.5 25 0.6 Severn Valley 11B SVW OX Severn Valley ware 1,220 42.6 20,114 29.7 1,729 42.6 11BG SVW black/grey variant 58 2.0 484 0.7 73 1.8 11CC SVW colour-coated variant 3 0.1 18 0.0 0 0.0 11D early SVW variant 16 0.6 249 0.4 66 1.6 17 SVW charcoal variant 53 1.8 1,359 2.0 23 0.6 23 SVW OX SVW hm storage jar 583 20.3 32,502 48.0 679 16.7 Local/ 9 unknown mortaria 1 0.0 56 0.1 3 0.1 Un- known 12 misc. colour-coated ware 5 0.2 26 0.0 0 0.0 21 oolitic limestone-tempered 2 0.1 3 0.0 0 0.0 24? Kingsholm-type flagon ware 2 0.1 11 0.0 0 0.0 GREY miscellaneous grey wares 64 2.2 538 0.8 41 1.0 GREY1 grey, micaceous variant 261 9.1 3,240 4.8 501 12.3 GREY2 grey sandy wares 37 1.3 587 0.9 55 1.4 BW black ware 15 0.5 109 0.2 0 0.0 MICBW micaceous blackware 1 0.0 4 0.0 0 0.0 MICGYF micaceous fine greyware 12 0.4 109 0.2 22 0.5 OXIDF fine oxidised sandy 4 0.1 56 0.1 41 1.0 MICOX micaceous oxidised ware 3 0.1 13 0.0 2 0.0 WSOXID white-slipped oxidised ware 3 0.1 29 0.0 0 0.0 WW miscellaneous whiteware 4 0.1 24 0.0 0 0.0 Total 2,865 100.0 67,647 100.0 4,059 100.0 NRFC = National Roman Fabric Reference Collection 04_BGAS125_131-246 16/11/07 15:49 Page 157

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 157

Table 2. Sewage works: samian by context Context Origin Form Date Other 104 CG 18/31 Hadrianic or early gully [103] Antonine 104 CG 33 Antonine gully [103] 104 CG 42 Hadrianic gully [103] 104 CG bowl or dish Hadrianic (or Antonine) gully [103] 215 SG Cup Flavian? or earlier enclosure ditch [208] 406 CG uncertain Hadrianic or Antonine wall robbing [405] 1003 CG 79 2nd half 2nd century cleaning 1003 CG 33 Antonine cleaning 1003 CG (MdV) 37 (decorated) 100–125 cleaning 1003 CG (MdV) 37 (decorated) 100–125 cleaning 1003 CG 37 (decorated) 120–130 cleaning 1003 SG 37 (decorated) 67–75 cleaning 1003 SG 37 Flavian or Trajanic cleaning 1003 SG 27 Flavian? cleaning 1003 CG uncertain Hadrianic or Antonine cleaning 1004 SG Inkwell? Flavian? cleaning 1005 CG 18/31 Hadrianic or early pit in NE area [1007] Antonine 1019 SG 35/36 Flavian or Trajanic ph 2 ditch [1020] 1024 CG 18/31 130–160 stone spread in [1036] (stamp Attius) 1039 SG 29 70–85 at latest cleaning 1046 CG bowl Antonine post-med ditch [1055] 1046 CG (MdV) dish – uncertain Trajanic post-med ditch [1055] form 1046 CG 18/31 or 31 Hadrianic or Antonine post-med ditch [1055] 1046 CG uncertain Hadrianic or Antonine post-med ditch [1055] 1047 CG 31R 2nd half 2nd century robber trench [1073] 1053 SG 29 Neronian or early Flavian pit/pooling [1054] 1060 SG prob Ritt 8 Claudian construction trench [1061] 1065 CG (MdV) uncertain – 46? Trajanic – early Antonine pit [1043] 1067 CG 33 prob Antonine pit [1043] 1067 SG dish footstand? Flavian or Trajanic pit [1043] 1071 CG (MdV?) 37 Trajanic-Hadrianic post-med ditch [1072] 1084 CG 37 (decorated) 120–130 ph 2 ditch [1020] 1086 SG 27g pre-Flavian or Flavian ditch [1020] 1086 SG bowl (Ritt 12 pre-Flavian or Flavian ditch [1020] or Curle 11) 1088 SG 29 (decorated) 50–65 ditch [1020] 1125 SG 29 prob pre-Flavian pit [1126] 1125 SG beaker? prob Flavian or earlier pit [1126] 1125 CG 79 2nd half 2nd century pit [1126] 1130 CG 18/31 Hadrianic or early SE enclosure ditch [1156] Antonine 1169 SG 18 prob Flavian small pit [1171] 04_BGAS125_131-246 16/11/07 15:49 Page 158

158 TOBY CATCHPOLE

1169 SG 18 prob Flavian small pit [1171] 1184 SG 18 (stamp C) prob Flavian pit [1185] 1186 SG 18 prob. Flavian pit [1185] 1191 SG 29 (decorated – stamp Murranus) 50–65 pit [1185] 1192 CG? 35 Hadrianic or Antonine grave [1194] 1196 SG 18 or 18/31 Flavian – Trajanic Structure G [1195] 1198 SG 18 or 18/31 Flavian – Trajanic Structure G [1195] 1206 CG (MdV) 24/25 Flavian – Trajanic? but Structure B [1207] after 100 1210 prob CG Scrap 2nd century furrow? [1211] 1230 CG 37 Hadrianic or early ditch [1245] Antonine 1239 SG 67? Flavian? cleaning 1257 CG 31 Antonine grave [1259]

trench [1061], probably Ritt. 8, in a pale fabric suggesting Claudian, or at latest Claudio-Neronian, manufacture and two decorated fragments of form 29 (nos. 1 and 2 below), one stamped by Murranus, c. AD 50–65. The other fragments, particularly of plain forms, are less closely datable, but, of the decorated forms, form 29 outnumbers form 37, which rapidly superseded it as the commonest decorated form after c. AD 70. Of particular interest in this connection is no. 3 below, which is likely to be among the earliest examples of form 37. The presence of so high a proportion of South Gaulish pieces seems to suggest activity on the site certainly from the early Flavian period. The relatively high number of decorated vessels (five examples of form 29, including base fragments, two of form 37) may suggest contact with the military. The three scraps of enclosed vessels pose more of a problem. None shows decoration. The commonest beaker form of the Flavian period is form 67, though the mouth of the beaker is wide enough to allow the slip to coat the interior of the vessel without forming an airlock. On two of the scraps the interior is not completely coated. It is possible that one at least may come from a form such as an inkwell, where the interior is often not fully slipped. If this is so, and it is by no means certain, its presence may also suggest the military. The other 30 vessels are Central Gaulish, though the majority are likely to date from the first half of the 2nd century. Five vessels in the fabric of Les Martres-de-Veyre include an example of form 24/25. The form, of South Gaulish origin, is normally pre-Flavian when found in Britain, though some examples continued to be made in southern Gaul during the Flavian period and the form was occasionally copied in Central Gaul. Export from Les Martres-de-Veyre to Britain is not thought to have started until c. AD 100. Only five of the Central Gaulish vessels are from decorated forms, all form 37, and all but one (from Phase 2 ditch [1245], fill (1230)) are likely to be Hadrianic at latest. Two, possibly three, are in the fabric of Les Martres-de-Veyre. All three sherds showing decoration (nos. 4–6 below) are in a style associated with Rogers’s (1974) potter X.13. Only about five vessels (forms 31, 31R, 79 (2) and the bowl fragment (38, 44 or similar)), all plain, are likely to date to the second half of the 2nd century AD. The Roman material previously found at Dymock has been usefully collected and published by Gethyn-Jones (1991). Although the largest collection of samian ware to be reported upon, by Brian Hartley from work in 1951 at the sewage treatment works, contained no 1st-century material (ibid. 94, cat. no. 9), Gethyn-Jones describes it as ‘a representative collection’ of the items found and 04_BGAS125_131-246 16/11/07 15:49 Page 159

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 159

draws a contrast with the coarse pottery, which clearly did contain 1st-century material. However, it is clear from the other finds that he lists that 1st-century, and indeed pre-Flavian, material has already been recorded from Dymock. The excavation of Grave 97 in the churchyard (ibid. cat. no. 6) produced samian forms 18 and 27, both 1st-century forms, the 27 with a stamp identified by Dr Grace Simpson as of Volus of La Graufesenque, c. AD 45–55. Also mentioned is the finding of various pre-Flavian coins, a denarius of Tiberius in ‘almost mint condition’ (ibid. cat. no. 15), two coins, one a dupondius , of Claudius I (ibid. cat. nos. 17, 21), and an as of Nero (ibid. cat. no. 4). Although coins remain long in use and it is possible that they only reached the site well after their date of minting, this seems less likely for the samian cup. That there should be occupation at Dymock by the Flavian period at latest should not therefore cause undue surprise. The Roman army was present in the Gloucester area from c. AD 50, first at Kingsholm and from the late AD 60s at Gloucester itself. Contact with the military need not necessarily imply a fort on the spot. The absence of decorated ware of Antonine date from the present collection may simply be fortuitous, as Antonine decorated sherds have been recorded by Gethyn-Jones: an early Antonine form 37 in the group studied by Brian Hartley (ibid. cat. no. 9) and ‘two Antonine decorated pieces, form 37’ from the school garden (ibid. cat. no. 21). It may well have been that less of this prestige ware was reaching Dymock from the mid 2nd century than hitherto, though the quantities involved, in all cases, are too small and, in the case of the older finds, possibly selective for an accurate assessment to be made. The presence of rivet holes in two of the three 2nd-century decorated bowls (nos. 4 and 6 below) suggests that such bowls were carefully preserved and not easily replaceable.

Catalogue of illustrated decorated samian ware (Fig. 13) Figure types are quoted from Oswald 1936–7 (O.), Central Gaulish decorative motifs from Rogers 1974 (Rogers).

1. Form 29, South Gaulish. Fragment of lower zone, showing straight, plain, gadroons, with a stamp of Murranus in the mould, parallel with the gadroons. c. AD 50–65. (1191) [1185]. 2. Form 29, South Gaulish. Fragment of upper zone showing a scroll with tendrils ending in rosettes, of a type common in the Neronian-early Flavian period. The small dots were also used by a variety of pot - ters, e.g. Albus (Knorr 1952, Taf. 2) and Calvus (Knorr 1919, Taf. 17, 23), and on a bowl from Kingsholm (Wild 1985, D25). There is little here to indicate a particular potter, but the date is likely to be c. AD 50–70. (1088) [1087]. 3. Form 37, South Gaulish. Rim fragment from a small bowl. It is likely that this is a very early example of the form, as the rim is very short, slightly incurved and more similar in form to the rim on form 29, though it lacks the rouletting. Finds at La Graufesenque have revealed cases where the bowl finisher, clearly unfamiliar with the new form, has given the bowl the rouletted rim, base and footstand of form 29. This piece is likely to be in the same general tradition. The ovolo, with trident tongue, is a com - mon one in the early Flavian period. It is generally considered that form 37 was first made at La Graufesenque c. AD 65, though it is unlikely to have reached Britain before c. AD 70, certainly not in any quantity. A single example of an early form 37, though with normal rim, occurs at Kingsholm (Wild 1985, D18), which, from its context, may or may not have reached the site before the evacuation of the pre-Flavian fortress. The date of the present piece is likely to be c. AD 65–75, though more probably after AD 70. (1003) cleaning. 4. Form 37, Central Gaulish. Two fragments, including base containing a lead rivet, probably from the same bowl in the fabric of Les Martres-de-Veyre. The ovolo (Rogers B14) was used by his potter X.13, as was the Triton (O.18) (Stanfield and Simpson 1958, pl. 45, 525). The motif at the left-hand side of the sherd may be the same leaf tuft (Rogers L19) as appears with the Triton on the bowl published in 1958. c. AD 100–120. (1003) cleaning. 04_BGAS125_131-246 16/11/07 15:49 Page 160

160 TOBY CATCHPOLE

4 6

Fig. 13. Sewage works: decorated samian. 04_BGAS125_131-246 16/11/07 15:49 Page 161

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 161

5. Form 37, Central Gaulish, in similar style to no. 4, showing neat, beaded panels with the rosette (Rogers C229) masking the junctions, the beaded ring (Rogers C293) and a saltire with the acanthus (Rogers K11). All the motifs occur on pieces in the style of X.13. The fabric is heavily over-fired, so that it is not possible to tell whether the piece is likely to have come from Les Martres-de-Veyre or Lezoux, where work in this style was being produced c. AD 120–130. It is not likely to have been from the same bowl as no. 4. c. AD 100–125. (1003) cleaning 6. Form 37, Central Gaulish, with a neat ovolo, probably Rogers B14, as on no. 4 above, but with a bead row above it as well as below. Although clearly a different bowl, the parallels are again with X.13, who sometimes uses a bead row above the ovolo (Stanfield and Simpson 1958, pl. 48, 567 etc.). There is a rivet hole through the rim, just above the ovolo. The fabric here seems more likely to be that of Lezoux. c. AD 120–130. (1084) [1020].

Other finewares Cologne colour-coated ware (KOL CC) (Tomber and Dore 1998, 57). Four sherds are present, all from beakers. Three came from gully [103], two from an indented beaker and one from a beaker with roughcast decoration (see Fig. 14, no. 17). The fourth sherd, also with roughcast decoration came from ditch [1020]. The indented beakers are more likely to be mid–later 2nd century. The roughcast beakers could date from the Flavian period into the 2nd century. Argonne colour-coated ware (ARG CC) (Tomber and Dore 1998, 46). A single small sherd of roughcast beaker came from gully [1124]. Probably early 2nd century. DRESSEL 14 AMPHORA by David Williams A plain amphora bodysherd, possibly from the lower section of the vessel, in a coarse sandy fabric with some rock inclusions, light buff outer surfaces (7.5 YR 8/6) and light red inner surface and core (2.5 YR 6/8), was recovered from fill (1086) of enclosure ditch [1087]. A small sample from the sherd was made into a thin-section and then studied under the petrological microscope. This showed ill-sorted grains of quartz, a number of them large and polycrystalline in texture, flecks of mica, sparse discrete grains of plagioclase feldspar, and with some fragments of chert, metamorphic rock and limestone. Given the small size and featureless nature of this sherd, it is difficult to be certain of the particular form involved here. However, consideration of the fabric, both in the hand specimen and in thin-section, suggests that it may belong to the Dressel 14 type of amphora (Peacock and Williams 1986, Class 20). This form seems to have carried various types of fish sauce such as garum, muria, liquamen and hallex and was current during the first three centuries AD. Kilns producing Dressel 14 are known at Calahonda in southern Spain (Beltrán Lloris 1970) and along the banks of the river Sado in Portugal (Mayet et al . 1996). The petrology of the Dymock sherd would appear to favour a Portuguese origin, although as yet no fabric details are available for the Spanish vessels.

Continental imports: other amphorae Dressel 20 amphora (BAT AM) (Tomber and Dore 1998, 84–5). A total of 15 sherds (1,807 g) of this Southern Spanish olive-oil amphora are present. This is the commonest amphora type to be found on Romano-British sites with an exporting period spanning the 1st–3rd centuries. A handle recovered from cleaning has a battered illegible stamp (Fig.15, no. 47).

Continental imports: mortaria North Gaulish mortaria (NOG WH4) (Tomber and Dore 1998, 75). A single rim was recovered from post-medieval ditch [1055]. 04_BGAS125_131-246 16/11/07 15:49 Page 162

162 TOBY CATCHPOLE

Coarsewares: native wares Native wares, that is to say hand-made wares with pre-Roman origins that continue to feature up until the 2nd century AD, are moderately well represented in the assemblage accounting for just under 8 per cent by sherd count. Grog-tempered ware (Glos TF 2). Hand-made jars, mainly small vessels with everted rims (see Fig. 14, no. 18) but with at least two very large hammer-rim bowl fragments, both from enclosure ditch [1087]. The fabric first appears in Gloucestershire in the early years of the 1st century AD and continues to occur into the Roman period. Malvernian rock-tempered ware (MAL RE A) (Tomber and Dore 1998, 147). Represented in similar quantities to the grog-tempered ware and broadly contemporary. Forms include squat jars (Fig.15, no. 42), everted rim jars, plain-rimmed dishes and lids. A perforated lid knob came from cleaning (1003). A sherd from gully [1034] has a burnished line lattice decoration. Palaeozoic limestone-tempered ware (Glos TF 33/216) (Peacock 1968, Malvernian B ware). Represented by 27 sherds in forms similar to the grog-tempered wares with at least one large hammer-rim bowl from pit [1038]. Sandstone-tempered ware (Glos TF 30). Eleven sherds from hand-made jars. 1st century. Shale and limestone-tempered ware (Glos TF 31). Six hand-made sherds, probably from the Malvern area. Jurassic limestone-tempered ware . A single sherd of hand-made ware came from cleaning.

Coarsewares: regional Regional imports account for a similar proportion of the assemblage as the native wares, 7.5 per cent by sherd count. The commonest by far is Dorset black burnished ware which accounts for 4.7 per cent of the total assemblage, 63 per cent of the regional imports. Smaller quantities of material came from the Oxfordshire industries, Savernake kilns, the Verulamium region, the Malverns and possibly Wiltshire. Savernake ware (SAV GT) (Tomber and Dore 1998, 191). Represented by just 10 sherds, mainly storage jar, probably dating from the 1st–early 2nd century. South-west oxidised ware (SOW OX) probably dating to the later 2nd–3rd centuries is represented by just four sherds. A single mortarium from the Verulamium industries (Fig. 15, no. 30) came from cleaning.

Dorset black burnished ware (DOR BB1) (Tomber and Dore 1998, 127). A range of forms is present spanning the 2nd to later 3rd century, for example jars (Fig. 15, no. 37), flat rim bowls/dishes (Figs. 14, no. 16; 15, no. 36), plain-rimmed dishes and grooved-rim dishes (Fig. 15, no. 27). No featured sherds from flanged-rim conical bowls which become common from the later 3rd–4th century are present. Part of a miniature jar was recovered from ditch [1020]. Over 40 per cent of the BB1 was recovered from cleaning layers. Oxfordshire wares (OXF RS; OXF WH) (Tomber and Dore 1998, 175–6). Represented by just 15 sherds in total. A single whiteware mortarium sherd came from grave [1189]. There were seven unfeatured whiteware sherds, probably from flasks or similar, one each from graves [1242] and [1259] and pit [1185], three from pit [1007] and one from cleaning. Six colour- coated sherds came from the cleaning horizons, one from a beaker with painted and barbotine decoration. Black sandy ware (Glos ?TF 201). Thirty sherds of a black sandy ware similar to a 1st- century–early 2nd-century burnished wheel-made ware thought to come from Wiltshire were present (cf. Rigby 1982, fabric 5). The fabric appears a little harder than the usual TF 201 and 04_BGAS125_131-246 16/11/07 15:49 Page 163

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 163

could come from a slightly different source. Forms include bowls, some with burnished lattice decoration (Fig. 14, no. 15), lids (Fig. 15, no. 45) and closed forms. Also possibly a Wiltshire product is a small oxidised sandy beaker with traces of a mica slip (Fig. 14, no. 13) from ditch [1020]. A few sherds (19) came from vessels belonging to the later Malvernian industry (Glos TF 19), which includes both wheel-made and hand-made forms.

Coarsewares: Severn Valley wares The largest single component of the Dymock assemblage comprises wares belonging to the Severn Valley industry within which there are a number of variants. As a group SVW accounts for 69 per cent by count, 82 per cent by weight. Severn Valley ware (SVW OX). (Tomber and Dore 1998, 148–50). The standard oxidised wheel- made SVW is numerically the largest category with some 1,226 sherds. A diverse range of forms is present with particularly high numbers of tankards/or carinated bowls/cups (Figs. 14, nos. 5, 11 and 24; 15, nos. 28, 34–5 and 43), necked jars (Fig. 15, no. 31), carinated bowls, handled jugs (Fig. 15, no. 29), dishes/platters (Figs. 14, no. 12; 13, no. 38), handled bowls (Fig. 15, no. 32), other small bowls/dishes (Figs. 14, nos. 10 and 12; 15, nos. 33, 38 and 41), colander and beakers (Figs. 14, no. 4; 15, no. 46). The emphasis is very much on products typical of the earlier Roman industry. Also present is a smaller amount of the same fabric in a reduced version occurring in the same form range. Early SVW (Glos TF 11D ). Earlier SVW is distinguished by a much less consistent fabric with clay pellets and organic matter in oxidised and, less commonly, black or brown fired fabrics. It generally dates from around the middle of the 1st century AD, possibly earlier (cf. Timby 1990). Most of the featured sherds are from carinated bowls or everted rim jars. Hand-made SVW (Glos TF 23). A hand-made variant generally reserved for large storage vessels (Figs. 14, no. 3; 15, no. 30). These are exceptionally well represented in this assemblage with several slight seconds, which begs the question as to whether they might be locally made. One vessel had VI incised into the rim (Fig. 14, no. 25) Charcoal-tempered variant (Glos TF 17). This variant, also generally a feature of the earlier industry, often occurs in a grey fabric and can be hand- or wheel-made. Mainly found in jar forms. Colour-coated variant . Three sherds of what appears to be a colour-coated SVW variant were noted, all from pit [1038]. This rare facet of the SVW industry has recently been recognised in Warwickshire in a distinct SVW form (P. Booth pers. comm.).

Coarsewares: unknown/local Several sherds encountered were not easily identifiable to the Gloucester fabric series or were perhaps subsumed into some of the less well-defined or less common categories. Most of these have been given broad generic terms here reflecting the broad category. Two fabrics have been identified. Three very small sherds were of Glos TF 21, an oolitic limestone-tempered ware, and two sherds were of a fabric greatly resembling Glos TF 24, the Kingsholm military fabric, both from ditch [1087]. Unknown mortaria : An oxidised mortarium was recovered from ditch [1156]. Moderately soft fabric with sparse dark orange-red sub-angular to rounded ?iron up to 2 mm, fine quartz and occasional fine white inclusions. No trituration grits visible. Grey micaceous ware (GREY1). A moderately soft greyware, finely micaceous with few other visible inclusions. The range of associated forms suggests a date for this ware of Flavian to early 04_BGAS125_131-246 16/11/07 15:49 Page 164

164 TOBY CATCHPOLE

2nd century. Vessels include carinated and other bowls (Fig. 14, nos. 2 and 19), rusticated jars (Fig. 14, no. 22), tankards in the SVW style, flanged handled bowls, curved wall and other dishes (Fig. 14, no. 21), dishes imitating moulded forms (Fig. 14, no. 9), beakers (Fig. 14, no. 20) and necked jars and a possible carinated jar (Fig. 14, no. 23). Decoration includes impressed comb, combed wavy line, rustication, irregular ridges (Fig. 14, no. 7) and burnished line. Fine micaceous greyware (MICGYF). A small group, including a necked jar and a platter, all from ditch [1020]. Micaceous black ware (MICBW). Moderately fine, highly micaceous black ware. No featured sherds. Black sandy ware (BW). A fine–medium black sandy ware. No featured sherds. Fine–medium grey sandy ware (GREY2). A moderately hard pale grey sandy ware with fine, but macroscopically visible quartz sand. Forms include a flanged bowl from cleaning, everted rim jars and a sherd decorated with a barbotine line. Other miscellaneous greywares (GREY). Featured sherds include a carinated bowl (Fig. 14, no. 1) and a hand-made dish decorated with radiating burnished lines (Fig. 14, no. 14). Fine oxidised ware (OXIDF). Just four sherds but including a ring-necked flagon and a beaker from ditch [1087], a curved-wall dish (Fig. 15, no. 44) and a segmental bowl from ditch [1020] with white painted decoration (Fig. 14, no. 8). The last is possibly an Oxfordshire or North Wiltshire product (cf. Young 2000, fig. 72, type 39). Micaceous oxidised ware (MICOX). Three sherds, one from a curved wall dish, possibly originally mica-slipped, from ditch [1020]. White-slipped oxidised ware (WSOXID). Three sherds, probably flagon. Miscellaneous white ware (WW). Four sherds from a fine white, globular-bodied, beaker decorated with white barbotine circles and raised dots (Fig. 14, no. 6) from ditch [1087]. Possibly from the North Wiltshire/Oxfordshire region.

Forms Table 3 presents a breakdown of the main forms present based on rim estimated vessel equivalents (EVEs). Overall, jars account for 55 per cent of the assemblage followed by beakers at 22 per cent and bowls/dishes at 18.3 per cent. Other forms, platters, cups, flagon, jugs, mortaria and lids are only present in minor amounts. Within the jar category storage jars constitute just over 30 per cent followed by necked everted rim jars at 26 per cent. High proportions of storage jar tend to be a feature of rural as opposed to urban sites perhaps reflecting the activities carried out at such sites. Normally assemblages are dominated by jars followed by bowls/dishes but a high proportion of drinking vessels is a phenomenon of the Severn Valley ware region (Evans 2001, 30) and is thus not that unexpected at Dymock.

Discussion The pottery recovered from the fills of the construction trenches associated with Structure A suggests a 1st-century date for the use of the building. No pottery was recovered from the posts and stake holes associated with the construction trenches. Pottery was recovered from the backfilled construction trench [1061[/[1068[/[1069], a total of 46 sherds (866 g). Most of the group, some 34 sherds are unfeatured Severn Valley wares. Associated with these is a single piece of South Gaulish samian of probable Claudian date, three sherds of Dressel 20 amphora and sherds of GREY1, 201 variant, MAL RT and two pieces of DOR BB1. Pit [1054] cutting the backfilled construction trench produced a sherd of Neronian or early Flavian South Gaulish decorated samian (Drag. 29), SVW OX and a carinated bowl (Fig. 14, no. 1). 04_BGAS125_131-246 16/11/07 15:49 Page 165

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 165

Table 3. Sewage works: pottery forms.

Form Sub-type EVE % % total JAR 202 8.9 necked, everted rim 596 26.2 beaded rim 134 5.9 neckless 113 5.0 storage jar 694 30.5 expanded rim, necked 77 3.4 bifid rim 15 0.7 hooked rim 13 0.6 cavetto rim 263 11.6 native internally expanded 165 7.3 Sub-total 2,272 100.0 55 BEAKER general 217 23.8 poppyhead beaker 29 3.2 tankard 666 73.0 Sub-total 912 100.0 22 BOWL general 167 21.9 fineware bowl 6 0.8 beaded rim 3 0.4 bifid rim 21 2.8 flat rim 81 10.6 carinated 48 6.3 flanged 38 5.0 hammer-rim 24 3.2 bowl/tankard 33 4.3 DISH general 16 2.1 fineware dishes 86 11.3 straight-sided plain rim 85 11.2 curved wall 103 13.5 grooved rim 34 4.5 flanged 8 1.1 bifid rim 8 1.1 Sub-total 761 100.0 18.3 PLATTER 40 0.96 CUP coarseware 2 fineware 10 0.3 FLAGON ring-necked 45 1 JUG 74 1.8 MORTARIA 27 0.65 LID 11 0.26 Total EVE 4,154 100.0 04_BGAS125_131-246 16/11/07 15:49 Page 166

166 TOBY CATCHPOLE

The only pottery associated with Structure B came from [1214], [1209] and [1207], a total of 35 sherds. Most are SVW OX, the exception being a Central Gaulish cup (Drag 24/5) of presumed Flavian/Trajanic date and two native ware sherds (TF 30). The group would appear to date well within the second half of the 1st century. Probable later plough furrow [1211] yielded a single small sherd of probable Central Gaulish samian broadly of 2nd-century date. The small group of 15 pottery sherds recovered from construction trench [1052], Structure C, appears also to date to the 1st century. Stratigraphically it is suggested that C postdates Structure A. SVW OX dominates, as with Structure A, but at least four sherds are of the earlier charcoal variant (TF 17). These are associated with a sherd of Dressel 20 amphora, a GREY 1 jar and a grog-tempered (TF 2) sherd. If fabric GREY1 dates to the Flavian period as is likely then this suggests a Flavian or later date for the backfilling of the structure. Structure D produced just 12 sherds, all SVW OX and including a curved wall dish. All again potentially could potentially date to the 1st century but equally be early 2nd century. The rectilinear enclosure defined by ditches [1087], [1156] and [208]/[1267] produced a good assemblage of pottery, particularly [1087] and [208]. Ditch [208]/[1267] yielded 121 fairly well- preserved sherds (2,483 g) of which 108 are SVW OX accompanied by native wares (TF 2, 33), Dressel 20, one sherd of fabric GREY 1 and two small sherds of fine greyware barbotine dot- decorated beaker. All these wares suggest a late 1st-century date. Ditch [1087] produced 123 sherds (5,354 g), a very high average sherd weight again being partly the result of a high proportion of SVW OX (TF 23) storage jar. This ditch also produced the Dressel 14 amphora sherd, decorated white ware beaker (Fig. 14, no. 6), a ring-necked flagon, possible Kingsholm flagon and three sherds of South Gaulish samian of pre-Flavian or Flavian date. The group might suggest a Flavian or slightly later date of deposition. The 32 (1,248 g) sherds from ditch [1156] mainly comprise SVW OX, particularly storage jar, accompanied by a flat rim DOR BB1 bowl, a mortarium sherd and a Central Gaulish samian dish (Drag. 31). The samian and DOR BB1 push the date into the first half of the 2nd century, conflicting slightly with the evidence from ditches [1087] and [208] suggesting an unrecognised later cut, disturbance or ongoing accumulation as those sherds derived from the upper fill (1130). Gatepost [1237] contained 15 sherds, mostly TF 23 and nothing that need date later than 1st century. Of the other structures no pottery was associated with Structure E and just seven sherds came from gully [1243], possibly linked to Structure F. All seven are of broadly later 1st- or early 2nd- century currency. Pit [1195] (Structure G) produced 94 sherds (2,651 g). The high average sherd weight is again the result of several substantial sherds of TF 23 storage jar, including that with the number incised into the rim (Fig. 15, no. 47). In contrast to many of the above-mentioned groups the assemblage, although again dominated by SVW OX, contains a significant amount, 21 sherds, of the grey micaceous fabric (GREY1) including rusticated ware. Two South Gaulish samian dishes (Drag. 18/31) suggest a date in the Flavian–Trajanic period. Two early DOR BB1 sherds are also present. The robber trenches associated with Structure H, [405]/[1010]/[1073], collectively produced 52 sherds (722 g) in more fragmented condition. Apart from a piece of modern china the latest material is DOR BB1 jar with oblique lattice, which must be 3rd century or later. Pit [1036] produced a single sherd of Central Gaulish stamped samian dating to the Hadrianic or early Antonine period. Of the eight inhumation burials, five contained pottery in the backfills of graves [1168], [1189], [1194], [1242] and [1259]. That pottery, 105 sherds in total, suggests that all the burials belong to the 2nd century. Graves [1242] and [1259] with mainly SVW OX and native ware may be of slightly earlier date (2nd century); [1259] contained a sherd of Central Gaulish Antonine samian and both graves had very small sherds of whiteware, possibly from the Oxfordshire kilns. Grave 04_BGAS125_131-246 16/11/07 15:49 Page 167

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 167

[1168] included three sherds of DOR BB1 whilst grave [1189] contained DOR BB1 and a sherd of Oxfordshire mortarium Among other assemblages of note, 135 sherds recovered from gully [103] (104) in evaluation trench 1 included three sherds of Cologne beaker and seven sherds of samian indicating a date from around the middle of the 2nd century. Ditch [1020] cut into the backfill of [1087] produced 163 sherds with a significant number of vessels in GREY1, the mica-slipped beaker (Fig. 14, no. 13), a miniature DOR BB1 jar, Cologne ware and a sherd of decorated Central Gaulish samian (Drag. 37) dated AD 120–130, broadly contemporary with gully [103]. Gully [1124] also produced a significant quantity of pottery, 94 sherds, with examples of rusticated greyware, North Gaul roughcast ware suggesting it is contemporary with ditches [103]/[1020]. Of the pits on the site [1171], with two sherds of South Gaulish samian, and [1185], with five such sherds including a stamped bowl (Fig. 13, no. 1) and a sherd of probable imported whiteware flagon, are probably the earlier, dating to the Flavian period. Pits [1007] and [1038] would seem to be early 2nd century in date. The gravel extraction pit, [1043], contained several sherds of 2nd- century DOR BB1 including a lid fragment alongside SVW OX and a sherd of mica-slipped oxidised ware. Pit [1126] cutting Structure D seems to contain one of the later stratified assemblages on the site with a grooved rim DOR BB1 bowl and a samian dish (Drag. 79) indicating a 3rd-century date. Possible well [1143] produced just 54 sherds, a mixture of SVW OX and native ware suggesting it is quite early in the sequence and was functional in the 1st century. The burnt contents of pit [1117] produced 20 sherds, including five SVW OX pieces alongside native wares indicating a likely 1st-century date.

Conclusions The assemblage recovered from the sewage works is slightly enigmatic in its composition and character. The presence of several sherds of South Gaulish samian, along with a smattering of other curiosities such as the Dressel 14 amphora, stamped Dressel 20 amphora, North Gaulish mortaria and possible imported whiteware, would suggest that this is not a typical rural site. On the other hand the dominance of Severn Valley wares, particularly the storage jars and the native wares, is completely in keeping with a 1st–2nd-century rural site. The incidence of samian at 2.2 per cent by sherd count is also perhaps a level to be expected from a rural site in Gloucestershire but the pre-Flavian element is odd and suggests something official. The overall composition of the assemblage and the forms do not argue for a military group although reference has been made in the past for the siting of a military fort at Dymock. If this were the case one might expect to see a much greater proportion of specialist wares and imports and a higher incidence of flagons as at sites like Kingsholm, Gloucester and Usk. There is also nothing present to point necessarily to a pre-Conquest origin for the occupation as the native wares always occur alongside SVW proper or other wares. Also of interest is the presence of a greyware industry dating to the later 1st–early 2nd century. Appearing to be local, it produced a completely different repertoire of forms compared to the indigenous SVW industry. It is an introduced Roman pottery production not dissimilar to some of the greywares produced at the Gloucester kilns although not as well fired. Until recently there has been very little work done at Dymock to understand the chronology, development and nature of the settlement. Most published knowledge derives from a gazetteer of information assembled and published by Gethyn-Jones (1991). The pottery data provided there would suggest a 1st-century Roman origin. Other recent excavations on land adjacent to the Rectory (for report, see Simmonds, below) and adjacent to Rose Cottage (Tavener 2001) have also yielded small assemblages of pottery. The pottery from the former only comprised some 339 sherds and appears to date to the later 1st–2nd century through to the 3rd century. The assemblage from 04_BGAS125_131-246 16/11/07 15:49 Page 168

168 TOBY CATCHPOLE

Rose Cottage is of comparable size to this group and again suggests a focus of activity in the 1st and 2nd centuries with limited evidence of 3rd-century occupation. Neither assemblage contained early samian or other unusual imports. Aside from the pre-Flavian component the sewage works groups broadly fit in with the slightly later chronology of these other assemblages, suggesting a focus of activity in the later 1st–2nd centuries, limited 3rd century activity and a complete decline by the 4th century.

Catalogue of illustrated sherds (Figs. 14 and 15) 1. Carinated bowl. Fine, dark grey fabric with a gritty texture: GREY. [1054] (1053). 2. Carinated bowl with combed wavy-line decoration. GREY1. [1061] (1060). 3. Hand-made storage jar. Glos TF 23. [1087] (1086). 4. Globular jar/beaker with a vertical rim. SVW OX. [1087] (1086). 5. Carinated cup. SVW OX. [1087] (1086). 6. Fine white ware beaker bodysherd decorated with barbotine circles and raised dots in white. WW. [1087] (1086). 7. Bodysherd from a jar or beaker decorated with irregular raised ridges. GREY1. [208] (215). 8. Segmental bowl in a fine oxidised ware with white painted decoration on the flange. OXIDF. [1020] (1084). 9. Platter imitating a moulded form. GREY1. [1020] (1084). 10 . Small bowl. SVW OX. [1020] (1084). 11 . Small carinated cup. SVW OX. [1020] (1084). 12 . Small flanged dish. SVW OX. [1020] (1084). 13 . Small mica-slipped beaker, slightly distorted. Fine oxidised sandy ware: Glos TF 3. [1020] (1019). 14 . Hand-made dish imitating a moulded form. Decorated with radiating burnished lines on the interior. Burnished exterior. GREY. [1020] (1019). Similar vessels are documented from Wanborough, Wiltshire (Seager-Smith 2000, figs. 85 and 91). 15 . Short flanged wheel-made bowl with burnished lattice decoration. Glos 201 variant. [1020] (1019). 16 . Flat rim bowl. DOR BB1. [1020] (1019). 17 . Cornice rim beaker with a roughcast finish. KOL CC. [103] (104). 18 . Hand-made grog-tempered jar. Glos TF 2A . [1020] (1019). 19 . Flanged, ‘S’-profile bowl. Fine, grey, slightly micaceous fabric: GREY1. (1037). 20 . Beaker or small jar with everted rim. GREY1. (1037). 21 . Dish with faint traces of rouletted decoration. GREY1. [1121] (1122). 22 . Globular jar with rusticated decoration. Fine greyware: GREY1. 23 . Fine grey sandy ware jar, probably carinated. GREY1. [1195] (1998). 24 . Large tankard. SVW OX. [1195] (1198). 25 . Hand-made storage jar. Glos TF 23. Incised VI on rim made after firing. [1195] (1198). 26 . Jar. Hard, fine light greyware with sparse grog: ?SAV GT. [1195] (1198). 27 . Grooved rim dish. DOR BB1. [1055] (1046). 28 . Tankard. SVW OX. [1055] (1046). 29 . Handled jug. SVW OX. [1055] (1046). 30 . Hand-made storage jar. Glos TF 23. [1055] (1046). 31 . Everted rim jar. SVW OX. [1055] (1046). 32 . Handled bowl. SVW OX. [1055] (1046). 33 . Grooved rim, hemispherical bowl. SVW OX. Cleaning (1003). 34 . Carinated bowl. SVW OX. Cleaning (1039). 35 . Small carinated cup. SVW OX. Cleaning (1039). 36 . Flat rim bowl with burnished line lattice decoration. DOR BB1. Cleaning (1039). 37 . Hand-made beaded rim jar. DOR BB1. Cleaning (1039). 38 . Dish. SVW OX. Cleaning (1039). 04_BGAS125_131-246 16/11/07 15:49 Page 169

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 169

Fig. 14. Sewage works: pottery. 04_BGAS125_131-246 16/11/07 15:49 Page 170

170 TOBY CATCHPOLE

Fig. 15. Sewage works: pottery. 04_BGAS125_131-246 16/11/07 15:49 Page 171

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 171

39 . Everted rim wide-mouthed jar. Glos TF 19. Cleaning (1039). 40 . Whiteware mortaria, probably VER WH. Flint trituration grits with some iron. Cleaning (1039). 41 . Small necked bowl. SVW OX. Cleaning (1039). 42 . Hand-made Malvernian jar. MAL RT, Glos TF 18. Cleaning (1039). 43 . Carinated cordoned bowl. SVW OX. Cleaning (1039). 44 . Curved wall shallow dish. Fine oxidised sandy ware: OXIDF. Cleaning (1239). 45 . Black sandy lid with a sooted interior. Glos 201 variant. Cleaning (1239). 46 . Butt beaker. SVW OX. Cleaning (1239). 47 . Dressel 20 handle with damaged illegible stamp. BAT AM. Cleaning (1003).

Metal and Glass by Hilary Cool The excavations at Dymock produced a small but interesting assemblage of finds of 1st and early 2nd-century date (Figs. 17, 18 and 19). Those from the non-funerary contexts are summarised in Table 4 according to whether they were found stratified or not and to function. In addition, four graves produced iron coffin fittings (nos. 55–8) and two had groups of hobnails (nos. 13–4). In the following report the material will first be discussed by functional category with a brief note on definite post-Roman items placed at the end. Following this the assemblage will be considered as a whole.

Table 4. Sewage works: summary of the non-funerary metal and glass finds.

Category Stratified Unstratified Total Personal Ornaments 39 12 Toilet equipment 1– 1 Textile equipment 1– 1 Household items 31 4 Structural items 81 9 Fasteners –6 6 Agricultural items –2 2 Miscellaneous 14 3 17 Post-Roman items 12 3 Total 31 24 55

At the outset it is appropriate to signal that the group includes finds of national importance as for the first time a site has produced both a mould and the brooch (no. 7) that may have been made in it (Fig. 16). The implications of this are considered below.

Personal Ornaments and Clothing The items in this category consist of ten brooches, two hairpins and, in two of the graves, groups of hobnails from shoes. Most were found unstratified and thus cannot be associated with particular buildings or areas of activity, but they do cast interesting light on the general occupation of the site. The number of brooches in the assemblage is not surprising as the West Country is an area of very high brooch use in the 1st and 2nd centuries. The types recovered strongly support the pottery evidence with regard to the date of the occupation, in that this is a mid 1st- to 2nd-century assemblage with the emphasis on occupation in the 1st century. One brooch, the rosette (no. 1), 04_BGAS125_131-246 16/11/07 15:49 Page 172

172 TOBY CATCHPOLE

16a. Moulds, clockwise from top left rf. nos. 23, 22, 21, 17 and 176

16b. Brooch cat no. 7 in mould rf. 23

Fig. 16. Sewage works: moulds. 04_BGAS125_131-246 16/11/07 15:49 Page 173

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 173

may be dated to the mid 1st century and thus precede the main date range suggested by the pottery. It is of the type with an arched bow and lozenge-shaped plate (Hattatt 1987, 48 class B; Feugère 1985, 288 type 19d). On the whole the rosette family appears to be a pre-Conquest type going out of use during the mid 1st century. There is evidence, though, that the form with the lozenge- shaped plate was one of the later variants in the sequence. Feugère (1985, 292) notes that in Gaul rosette brooches are most commonly recovered from contexts belonging to the second and third quarters of the 1st century. At the King Harry Lane cemetery at Verulamium most of those with lozenge plates come from phase 3 contexts and the majority of those with circular plates from the earlier phase 2 contexts (Stead and Rigby 1989, 93–4). No. 1, therefore, cannot be taken as definite evidence of pre-Conquest occupation, but certainly it would not have been in use after the third quarter of the 1st century. Nauheim Derivative brooches, represented here by no. 2, were another pre-Conquest type which became increasingly common after the Conquest (Hull types 10–11: see Bayley and Butcher 2004, 147). No. 2 has a rod bow with an unusual moulded decoration on the upper part. Rod bow Nauheim Derivatives continued in use into the later 1st century and were out of general use by the early 2nd. Three brooches belong to post-Conquest types in use during the second half of the 1st century. One (no. 5) is an Aesica brooch (Hull type 37: see Mackreth 1982; Hattatt 1987, 54). The other two (nos. 3 and 4) are Colchester derivatives. Generally these may be dated from the mid 1st century into the 2nd century; those with perforated catch-plates tend to be early examples and may be dated to the 1st century. The form found at Dymock is the Polden Hill (Bayley and Butcher 2004, 159) where the spring is held by a perforated lug and a bar in the perforated ends of the spring cover. Polden Hill brooches were the preferred form of Colchester derivative in the West Country. Both Dymock examples belong to a variant that appears to be particularly common in the Gloucestershire region (Cool 2007 ; see also Mackreth 1998, 118, nos. 17–20), scant dating evidence suggesting that it was in use during the second half of the 1st century. One (no. 4) was among the very small number of brooches found stratified. Coming from the backfill of the enclosure ditch together with late 1st-century pottery, it provides useful additional confirmation of this dating. No. 3 was in poor condition and was subjected to investigative conservation. As far as can be seen neither the head loop nor the ends of the spring covers have been perforated to take the chord of the spring and the ends of the spring bar respectively. The return plate also appears more of a solid ridge and would have provided very little seating for the pin. This, therefore, appears to be brooch from a mould that needed to be finished. The ridge down the catch-plate, for example, would have been hammered out, the head loop and spring covers perforated, the spur on the catch- plate filed away, and the central rib on the bow probably transversely grooved. Though none of the mould fragments need have come from a brooch of this type, it seems very likely that the workshop also produced this form of Polden Hill brooch Two trumpet brooches (nos. 6 and 7) have a slightly later floruit as they belong to a variant in use during the later 1st and into the early 2nd century. This variant (Hull type 154: see Hattatt 1985, 109) is commonly called the Chester trumpet brooch, though there is increasing evidence that its main concentration was in the Gloucestershire region. The concentration in the Cirencester area strongly suggested that there was a workshop in the vicinity (Mackreth 1998, 134; Cool 2007); and the mould evidence from this site (see Dungworth, below) has confirmed this, Rf. 23 being part of a mould for a Chester brooch. In as far as is possible to tell given that it retains its spring, no. 7 fits very neatly into the mould no. Rf. 23, and it seems reasonable to conclude that it was made in that mould or in another created by the same craftsman. To the author’s knowledge, the possibility of directly linking a finished Romano-British brooch and the mould in which it could have been made has never been possible (for a survey of brooch 04_BGAS125_131-246 16/11/07 15:49 Page 174

174 TOBY CATCHPOLE

moulds see Bayley and Butcher 2004, 35–6). These discoveries, therefore, provide a unique opportunity for tracing the distribution area of a particular workshop. The finished trumpet brooches have two very distinctive features. Both have a band of rocker arm ornament on the catch- plate either side of the junction with the bow. No. 6 was definitely coated with white metal and a sheen on the surface of no. 7 hints that it might have been. The use of a white metal coating and the rocker arm ornament do not appear to be normal on brooches of this type. It was certainly not present on the seven published examples from the Kingscote area (Mackreth 1998, 134 nos. 102–7, 326 no. 19) and on eleven from the Cotswold Waterpark sites (Cool 2007 ). These features might well be diagnostic for the Dymock workshop, as might the highly arched bow seen on no. 6, and future research should make it possible to plot the distribution of the products. Unfortunately the other brooch mould fragments from the site are not sufficiently diagnostic to link them to particular types of brooch. The other two brooches (nos. 8 and 9) are both penannular brooches where the notched terminals are bent back onto the hoop (Fowler 1960, type D5). No. 8 came from a ditch infill together with 2nd-century pottery and a hairpin (no. 12 below). Penannular brooch types tend not to be closely datable but it has been noted the Fowler D5 form has quite often been found in mid to late 1st-century contexts (Manning et al. 1995, 94 no. 76; Mackreth 2000, 157). These two examples would fit happily into this assemblage predominantly of the second half of the 1st century. This is a small brooch assemblage so conclusions about the occupation on the site drawn from it have to be cautious. The 1st-century date though is very clear as is the paucity or absence of common local types that might be expected in the 2nd century such as the lower Severn ‘T’-shape brooches (Hattatt 1987, 120; Bayley and Butcher 2004, 167). Also noticeable is the absence of Hod Hill brooches (Hattatt 1985, 56; Bayley and Butcher 2004, 152). This may be a function of chronology as they would not have been in use in the final quarter of the 1st century, but as already noted the assemblage does contain some brooches which would have been contemporary with them. Their absence, however, does fit a pattern observed in studying the Cotswold Water Park sites (Cool 2007 ) where they did not seem to be a type particularly popular amongst the rural population in the area. As well as the brooches, there are two copper-alloy hairpins. No. 11 is a 2nd-century form, common across the province (Cool 1991, 154 Group 3A). No. 12 is a slightly unusual variant. The cross-hatched unit recalls a common West Country hairpin type (ibid. 170 Group 23) though differing in precise details. The nearest parallels fall within my Group 5c where the head is only slightly larger than the shank (ibid. 157). Examples with a variety of terminal types have been recovered from Shakenoak (Brodribb et al. 1973, fig. 55 no. 203). Alchester (Iliffe 1932, pl. XVIII nos. 2 and 3) and Ditchley (Cool 1991, fig. 4 no. 7). The Shakenoak pin came, as did one of the Alchester examples, from a context with a t.p.q. of the early 2nd century, the context predating a house dated to the Domitianic/Hadrianic period. No. 12 adds to this dating evidence as it came from a 2nd-century context. At present a late 1st- and 2nd-century date may be suggested for the type, which is clearly a regional one. Very possibly it was made by the workshops making the grooved collar form of nail cleaner (Crummy and Eckardt 2004, 54), which was in production by the mid 2nd century. The recovery of two 2nd-century hairpins at Dymock fits a regional pattern in which the Romanised hair-styles requiring them were adopted only in the 2nd century (Cool 2007). Finally, two of the graves contained small numbers of hobnails (nos. 13 and 14). In neither case have enough nails survived to guarantee that shoes were deposited with the body. Even a lightly nailed shoe can have 25 nails, and numbers of c. 100 are not unusual (Cool and Booth 2007). Although the site was plough damaged, the lower parts of the bodies in the two graves were well preserved and it might have been expected that more hobnails would have been recovered. Second- 04_BGAS125_131-246 16/11/07 15:49 Page 175

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 175

century pottery recovered from the fills of both the graves suggests a 2nd-century or later date for the burials. Philpott (1991, 171) has noted that nailed shoes started to be placed in burials from the 2nd century onwards.

Catalogue of personal ornaments and clothing (Figs. 17 and 18) 1. Rosette brooch. Copper alloy. Remnants of cylindrical spring cover with groove along upper edge retaining fragments of spring; curved reeded bow; diamond-shaped plate with grooved semi-circle, punched oval motifs with transverse ribs; upper part of catch-plate on back of plate. Present length 28 mm. (1000): Rf. 34. 2. Nauheim Derivative brooch. Copper alloy. Spring of four turns with chord passing below bow, pin missing; ‘D’-sectioned tapering rod bow angled at head; broken at upper part of catch-plate. Transverse mouldings on upper part of bow below angled head. Present length 55 mm. (1000): Rf. 33. 3. Polden Hill brooch. Copper alloy. Short semi-cylindrical spring covers with closed ends; cast loop on head extending as raised rib on upper half of bow; ‘D’-sectioned bow tapering to broken foot; triangu - lar catch-plate with triangular perforation; groove down return plate which appears closed with spur at top. Much corroded surface and details obscured. Wing width 27 mm, original length c. 59 mm. (1003): Rf. 155. 4. Polden Hill brooch. Copper alloy. Short semi-cylindrical spring covers with perforated ends and verti - cal groove at either end; one end broken; cast loop on head extending as raised finely ridged rib on upper half of bow; ‘D’-sectioned bow tapering to foot; triangular catch-plate with triangular perfora - tion. Much corroded and bent out of shape. Wing width 19 mm, original length c. 55 mm. (1086): Rf. 46. 5. Aesica brooch. Copper alloy. Open-backed semi-cylindrical spring case with closed ends with notch at margins in which rest bar runs through centre of spring of three turns on either side; chord held by stump of forward facing hook; curved upper bow expanding to small knobs at corners with central lug for junction to (missing) lower bow. Length 24 mm, maximum width 29 mm. (1000): Rf. 35. 6. Trumpet brooch. Copper alloy. Small trumpet head with flange around edge, perforated lug behind retaining central bar, probably lower edge of loose wire headloop, and spring of one turn on either side with broken pin. Arched and waisted upper bow with central vertical rib, moulding above and below central button, pair of lentoid mouldings and narrow rib on lower part of button, upper part obscured by corrosion; triangular-sectioned lower bow with central moulding and groove parallel to each edge; small double ribbed foot trapezoidal catch-plate with band of rocker arm ornament on each side by junction with bow. Traces of white metal coating in places including lug. Length 42 mm. (1000): Rf. 31. 7. Trumpet brooch. Copper alloy. Small trumpet head; perforated lug behind retaining central bar, prob - ably lower edge of loose wire headloop, spring of one turn on either side with pin. Waisted upper bow with central vertical ridge, moulding above and below central button with narrow rib on top and bot - tom of rib; triangular-sectioned lower bow; small double ribbed foot trapezoidal catch-plate with band of rocker arm ornament on each side by junction with bow. Possible traces of white metal coating. Length 42 mm. (1004): Rf. 5. 8. Penannular brooch. Copper alloy. Circular-sectioned hoop with terminals bent back on top of hoop, one terminal has transverse depression; circular-sectioned pin tapering to point at one end and flattening at other where it wraps around hoop. Diameter 26 mm, hoop section 2 mm. (1019): Rf. 15. 9. Penannular brooch. Copper alloy. Circular-sectioned hoop with flattened terminals bent back over hoop; terminals have three transverse grooves. Circular-sectioned pin flattening and wrapping around hoop. Diameter 37 mm, hoop section 3 mm. (1004): Rf. 18 . 10. Brooch (not illustrated: stolen during an open day). (1003): Rf. 10. 11. Hairpin. Copper alloy. Circular-sectioned tapering shank with decoration for head cut into top of shank; pointed knob terminal with two ribs below; biconical unit; two grooves below forming two ribs. Bent out of shape. Original length c. 105 mm, head section 3 mm. (1003): Rf. 14. 04_BGAS125_131-246 16/11/07 15:50 Page 176

176 TOBY CATCHPOLE

Fig. 17. Sewage works: copper-alloy objects. 04_BGAS125_131-246 16/11/07 15:50 Page 177

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 177

12. Hairpin. Copper alloy. Cylindrical head with conical terminal, sides decorated with very fine diagonal cross-hatched grooves with horizontal groove above and below cross-hatched zone; circular-sectioned shank tapering to point. Length 89 mm, head diameter 4 mm, shank section 2 mm. (1019): Rf. 16. 13. Hobnails (not illustrated). Iron, in poor condition. At least 10 head fragments, also shank and uniden - tifiable pieces. Head diameters c. 10 mm. (1192): Rf. 120. 14. Hobnails (not illustrated). Iron. Eight retaining conical heads, also shank fragments. Head diameters 9 mm. (1166): Rf. 14.

Toilet Equipment The only item of toilet equipment is a nail cleaner with a bone head (no. 15). This is a form that was in use during the later 1st and 2nd century and has a regional distribution centred on the Gloucestershire/Oxfordshire area (Crummy and Eckardt 2004, 53, illustration 5). The find, given the associated pottery, appears to be from a 1st-century ditch fill. The green staining to the bone head is almost certainly deliberate rather than the result of corrosion; the habit of staining bone appears to be an early Roman fashion (Crummy 1983, 20).

Catalogue of toilet equipment (Fig. 18) 15. Nail cleaner. Circular-sectioned shank flattening to notched end; upper end has cross- hatched groove pattern; ovoid bone knob head fitted over end, head stained green. Length 46 mm, knob head section 6 mm. (1086): Rf. 45.

Textile Equipment Of the three whorls (nos. 16, 41 and 42) recovered, only no. 16, which is symmetrical and carefully made, could have functioned effectively as a spindle whorl. The other two would not have provided an even twist. The perforation diameter is large for a Roman spindle which normally requires perforations of c. 5–7 mm (Walton Rogers 1997, 1735). However, given the rarity of medieval material on the site, and the fact that it was found stratified in a Roman context, its Roman date is undoubted.

Catalogue of textile equipment (Fig. 18) 16. Spindle whorl. Oxidised coarse pottery (SVW OX) sherd cut into disc with smoothly ground sides and cen - tral cylindrical perforation. Diameter 40 mm, thickness 8 mm, perforation diameter 8 mm. (1037): Rf. 51.

Household Equipment The assemblage of household items hints at an establishment of some pretensions. No. 17 is the neck of a globular or conical jug of Isings’s (1957) form 52 or 55 (Price and Cottam 1998, 150–7). In general this was in use in the second half of the 1st century with some variants continuing into the mid 2nd century. Despite being found in a 2nd-century context, the very dark yellow/brown glass of this example indicates it is a 1st-century vessel as such strong colours disappear by the end of the century. No. 18 may have been from the same jug but it appears to be of a less intense shade and so may be from another vessel. The fact that a glass jug is the only identifiable vessel from the site is interesting as it is far more common for rural sites in the 1st century to have utilitarian glass bottles or large bowls, if they have glass vessels at all (Cool and Baxter 1999, 84). Jugs are more typical of urban sites. 04_BGAS125_131-246 16/11/07 15:50 Page 178

178 TOBY CATCHPOLE

No. 19 is a metal shank of an item such as a hairpin, a toilet implement or a spoon. The white metal coating strongly suggests a spoon, the only one in the range of possibilities that is regularly treated in this way. If it is from a spoon this again argues for the pretensions of the settlement, as during the 1st to 2nd centuries spoons are much more associated with urban than rural sites (Cool 2004, 29–30) and this was found in a late 1st-century context. The final item (no. 20) that may be assigned to this category is a fragment of lead alloy that might come from a dish or bowl. It is too small and distorted to be identified with certainty.

Catalogue of household equipment (Fig. 18) 17. Jug neck fragment. Dark yellow/brown glass. Cylindrical neck expanding out slightly to constricted tooled junction with missing body. Very faint traces of diagonal ribs on lower part of neck. Present length 88 mm, neck thickness 3 mm. (1123): Rf. 128. 18. Body fragments (2: not illustrated). Dark yellow/brown glass. Convex-curved, broken at carination; parts of two vertical optic blown ribs. Dimensions (larger) 32 × 24 mm. (1039): Rf. 44 19. Spoon handle (not illustrated). Copper alloy coated with white metal. Circular-sectioned tapering to point at one end and broken at other. Present length 58 mm, section 2.5 mm. (1088): Rf. 50. 20. Bowl or dish rim fragment (not illustrated)? Lead alloy. Rounded rim bent out slightly; the whole frag - ment bent out of shape. Dimensions 30 × 11 mm, wall thickness 1 mm. (1004): Rf. 2.

Structural Equipment Iron nails came from a variety of pits, gullies and other features. Where they are complete, or near complete, most (nos. 22–3, 27, 29) are of lengths typically used for timber cladding and similar purposes (Manning 1985a, 291 group E) and tend to be very common in Roman assemblages. A much longer nail (no. 28) would have been used for a more substantial piece of joinery. If this is indeed a later 1st- to mid 2nd-century site as the pottery and other finds discussed so far suggest, the number of nails is of some interest. There is some evidence that iron did not become plentiful on some rural sites in the area until well into the Roman period (Cool 2007). Here the inhabitants were clearly using Roman building methods earlier than some of their neighbours.

Catalogue of structural equipment 21. Nail. Iron. Two head and shank fragments also one shank fragment. (1066): Rf. 166. 22. Nail. Iron. Tip missing. Length 55 mm, head diameter 11 mm. (1065): Rf. 165. 23. Nail. Iron. Complete. Length 50 mm, head diameter 16 mm. (1157): Rf. 170. 24. Nail. Iron. Shank, two fragments. (1045): Rf. 164. 25. Nail. Iron. Shank fragment. (1035): Rf. 163. 26. Nail. Iron. Shank fragment. (1005): Rf. 162. 27. Nail. Iron. One complete and one shank fragment. Complete – length 50 mm, head diameter 11 mm. (104): Rf. 160. 28. Nail. Iron. Complete. Length 125 mm, head width 20 mm. (1019): Rf. 156. 29. Nail, Iron. Complete, bent. Length 68 mm. (1000): Rf. 44.

Fasteners and Fittings The underlying theme of modest affluence continues with the fasteners. The presence of a lift key (no. 30; see Manning 1985b, 90) indicates at least one door with a simple tumbler lock. The studs (nos. 31 and 32) are likely to have come from a box or chest. No. 31 is not intrinsically datable, but no. 32 is from a stud or mount comprising a copper-alloy sheet over a lead-alloy core into 04_BGAS125_131-246 16/11/07 15:50 Page 179

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 179

Fig. 18. Sewage works: copper-alloy, glass and other objects. 04_BGAS125_131-246 16/11/07 15:50 Page 180

180 TOBY CATCHPOLE

which an iron shank would have been inserted. This construction technique was used on box fittings throughout the Roman period as can be seen on a casket used in a mid to late 2nd-century burial at Skeleton Green, Hertfordshire (Borrill 1981, 305), and on another from a 4th-century burial at Colchester (Crummy 1983, 85, nos. 2179–82). It is possible that no. 33 came from a similar stud or mount as clearly the interior was not designed to be seen. Alternately, given the stump of a projection on one side, it might have been half of a pendant or bulla . The repoussé palmette design on no. 34 strongly suggests it is of Roman date, though it is difficult to see whether it would have been mounted flat or would have acted as a terminal like a strap-end. No. 35 does appear to have acted as a strap-end but it is not intrinsically datable and may not be of Roman date.

Catalogue of fasteners and fittings (Figs. 18 and 19) 30. ‘T’-shaped lift key. Iron. Handle with loop end tapering evenly to stem with single tooth on each side. Loop at 90 degrees to teeth. Length 178 mm, width teeth 50 mm, diameter of loop 16 mm. (1003): Rf. 8. 31. Stud. Copper alloy. Hollow hemispherical head; square-sectioned tapering shank. Length 32 mm, head diameter 30 mm. (1000): Rf. 37. 32. Stud head or mount (not illustrated). Copper-alloy shallow domed circular sheet head, approximately two thirds extant; fragments of lead alloy infill. Diameter 31 mm. (1003): Rf. 11. 33. Mount? Copper alloy. Domed hollow oval with flat flange and broken stump at one side; interior less well-finished. Length 29 mm, width 22 mm. (1003): Rf. 12. 34. Mount. Copper alloy. Rectangular sheet strip with one corner missing. Two thirds decorated with repoussé palmette design, plain end has small central perforation. Fragment now bent in two across pal - mette design but crease at end of the design suggests the piece may originally have been bent at that point. Total length 37 mm, length of repoussé panel 22 mm, width 11 mm, thickness 0.5 mm. (1000): Rf. 41 . 35. Strap-end (not illustrated). Copper alloy. Rectangular plate with small perforation in each corner and lower end tapering to point. Bent in half. Original length c. 60 mm, width 14 mm. (1000): Rf. 39.

Agricultural Equipment Agricultural activities on the site are represented by a goad (no. 36) and a reaping hook (no. 37). The goad is clearly of Roman date given its context. The hook, an unstratified find, is more open to question. It is in relatively good condition, still retaining the nail that held the handle in place. Roman reaping hooks are more often tanged than socketed as here (Manning 1985b, 53). Whilst many undoubted Roman items have come from the unstratified finds, reaping hooks had a long life, and the possibility that this find was associated with post-Roman agriculture must be borne in mind.

Catalogue of agricultural equipment 36. Goad. Iron. Single ring of spiral with pointed tang extant. Length 25 mm, diameter of spiral 16 mm. (1159): Rf. 171. 37. Reaping hook. Iron. Curved blade tapering to point; open socketed handle with perforation for fixing nail retaining nail shank. Length 97 mm, handle width 19 mm. (1000): Rf. 32.

Catalogue of miscellaneous items The following cannot be dated other than by their contexts. 38. Chape? Copper alloy. Rectangular sheet bent over at one long edge, broken on lower edge and on other long side; upper edge also broken but retains two small perforations at side and centrally (correspond - 04_BGAS125_131-246 16/11/07 15:50 Page 181

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 181

Fig. 19. Sewage works: iron latch lifter. 04_BGAS125_131-246 16/11/07 15:50 Page 182

182 TOBY CATCHPOLE

ing area on other side missing); possibly broken across lower edge of much larger central perforation above. Bent and distorted. Present length 40 mm, width 26 mm. (1000): Rf. 40. 39. Chain. Iron. Four inter-locked figure of eight links. Dimensions of links 38 × 17 mm, 40 × 38 mm, 40 × 18 mm, 42 × 21 mm. (1084): Rf. 167. 40. Ring. Copper alloy. Diamond-sectioned. Diameter 27 mm, section 5 mm. (1005): Rf. 27. 41. Whorl. Lead alloy. Irregular plano-convex with off-centre square perforation. Weight 19g. Diameter 22 × 20 mm, thickness 8 mm, perforation diameter 4 mm. (1000): Rf. 38. 42. Whorl fragment. Tile. Retaining one original face, other side mainly spalled and missing; edges bro - ken; irregular perforation with conical cross section. Dimensions 41 × 40 mm, thickness 17 mm, max - imum perforation diameter 11 × 8 mm. (1046): Rf. 24. 43. Sheet. Copper alloy. Rectangular with one end bent over. Original length c. 52 mm, width 20 mm, thickness 0.5 mm. (1043): Rf. 25. 44. Strip. Copper alloy. Narrow rectangular strip rolled into tapering point over two thirds of length. Length 43 mm, maximum width 3 mm. (1039): Rf. 49. 45. Strip. Copper alloy. Oval-sectioned and curved; in two pieces all ends broken. Present length c. 55 mm, section 3.5 × 3.0 mm. (1060): Rf. 20. 46. Rod. Iron. ‘L’-shaped. Length 94 mm. (1086): Rf. 168. 47. Rod. Iron. Tapering and bent. Length 265 mm, width 6 mm. (1003): Rf. 9. 48. Bar. Iron. Length 72 mm, width 15 mm, thickness 12 mm. (1226): Rf. 174. 49. Bar. Iron. Present length 75 mm. (1198): Rf. 172. 50. Bar or nail shank. Iron. Curved. Present length 30 mm. (216): Rf. 161. 51. Plate. Iron. Folded fragment. Dimensions 30 × 27 mm. (1045): Rf. 175. 52. Plate. Iron. Fragment. Dimensions 60 × 34 mm. (1125): Rf. 169. 53. Fragment. Iron. (1125): Rf. 157. 54. Off-cut. Lead alloy. Tapering rectangular with bevelled long edges. Weight 3 g. Length 43 mm, width 7 mm, thickness 2 mm. (1039): Rf. 48.

Funerary Finds Iron fittings for coffins were found in four graves. In three cases (nos. 55, 57 and 58) only nails were found but in the fourth (no. 56) there were additional fittings such as a loop-headed spike (56a), two possible angle brackets (56b and c), two unusual nails (56d and e) which may have been decorative rather than strictly functional and two possible decorative plates (56f and g). This coffin also stands apart in the length of the nails used in its construction. Table 5 summarises the minimum number of nails in each coffin judged by the extant heads, the length of complete and near complete nails, and the average diameter of the head. The coffin represented by no. 56 (1189), as well as having possible decorative elements, was more substantial than the other three judged by the lengths of its nails and the sizes of their heads. Interestingly it was associated with the tallest of the females. She was 9 cm shorter than the only male and his coffin (grave 1242) in comparison was of light construction. As stature does not appear to have been the reason for her having a more substantial coffin, the elaboration may have been because of her position in the community.

Table 5. Sewage works: comparison of the sizes of nails from the coffins.

Grave No. of nails Length: more Length: less Average head than 80 mm than 80 mm width in mm 1168 17 2718 1189 25 9 11 21 1194 15 3316 1242 20 – 20 15 04_BGAS125_131-246 16/11/07 15:50 Page 183

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 183

Summary catalogue of funerary finds A full catalogue of coffin nails is to be included in the site archive. 55. Iron fittings from Grave 1168 a–t Twenty-one coffin nails. (1166) 56 Iron fittings from Grave 1189 a Loop-headed spike. Rectangular-sectioned bar bent into almost closed loop with tapering spike. Length 50 mm, loop diameter 28 mm, loop width 16 mm. (1170): Rf. 97. b Angle bracket? ‘L’-shaped with shorter arm wider than longer. Length 40 mm, maximum width 22 mm. (1170): Rf. 129. c Angle bracket? Maximum length 37 mm. (1170): Rf. 130. d–e Two coffin nails. (1170). f Coffin fitting. Bar with ends bent up and in. Length 63 mm, depth 25 mm. (1170): Rf. 121. g Coffin fitting? Bar tapering to one end. Length 50 mm, maximum width 12 mm, thickness 10 mm. (1170): Rf. 132. h–aj Twenty-nine coffin nails. (1170).

57 Iron Fittings from Grave 1194 a–v Twenty-two coffin nails. (1192). 58 Iron fittings from grave 1242 a–x Twenty-four coffin nails. (1240).

Post-Roman Finds Compared with the number of undoubted Roman finds, post-Roman material was rare. The frame of a post-medieval jew’s harp (no. 59; see Lawson 1990) was unstratified and a button of relatively modern date (no. 60; see Biddle and Cook 1990, 573 type F) came from the fill of a post-medieval ditch. The binding (no. 61) is also probably of modern date as it appears to be machine made.

Catalogue of post-Roman finds 59. Jew’s harp. Copper alloy?. Diamond-sectioned frame with slightly asymmetrical legs; filed rebate on loop for seating of tongue. Length 52 mm. (1000): Rf. 36. 60. Button. White metal. Flat head with loop behind and traces of maker’s name. Diameter 21 mm. (1071): Rf. 26 61. Binding. Copper alloy. Straight length of semi-circular binding; small strip wrapped over at one end retaining rivet. Length 27 mm, width 5 mm. (1004): Rf. 4.

Conclusions These finds support the evidence of the pottery (Timby, above) that occupation started in this part of Dymock in the later 1st century and that there may have been an establishment of some status in the area. Household items such as the glass jug and the spoon certainly indicate occupation with pretensions above a basic rural existence as may the lift key, the mounts and studs from furniture, and possibly the early adoption of Roman joinery techniques indicated by the nails. In general the finds are typical of this region in the 1st to 2nd century and Provide useful confirmation of the patterns that have been observed further east in the county as part of the Cotswold Water Park project. Of great importance is the discovery of not only moulds for making copper-alloy articles but also at least one finished brooch which would have come from them and an unfinished brooch of a different form. Future research to track the distribution of products from this workshop might 04_BGAS125_131-246 16/11/07 15:50 Page 184

184 TOBY CATCHPOLE

prove informative, but to date extensive enquiries amongst fellow specialists have not yet produced other brooches with the distinctive features of nos. 6 and 7.

Slag and Moulds by David Dungworth

Iron-Working Slags In all, 192 pieces of slag and other debris weighing in total 11,485 g were recovered (see Table 6). A large portion of the slag is tap slag that would have been produced while smelting iron from ore. Given the lack of any distinctive smithing slags, the large quantity of undiagnostic slags probably relate to smelting. The small size of this assemblage (in particular the small quantity of furnace lining) suggests that the smelting took place not within the area excavated but near by. It is possible that the slags from post-Roman contexts are Roman in origin and residual in the later contexts. The large quantity of undiagnostic slag in post-Roman contexts, however, may relate to other iron working, e.g. iron smithing, but the lack of diagnostic iron-smithing slags (in particular smithing hearth bases) makes this no more than a possibility.

Table 6. Sewage works: weight of slag (in grams).

Coal Ore Daub Furnace Fuel Tap Undia- Total Lining Ash Slag Slag gnostic Roman 175 34 23 61 1,979 507 2,779 Post-Roman 1– 86 ––1,279 3,791 5,157 Unphased 2 406 – 130 – 1,665 1,346 3,549 Total 3 581 120 153 61 4,923 5,644 11,485

Explanation of Terms Used The fragments of dense, fayalitic (iron silicate) tap slag show a characteristic ‘ropy’, flowed morphology on their upper surface and low vesicularity at their fracture surfaces. They are diagnostic of iron smelting (i.e. primary extraction from the ore) and are typical waste products of the bloomery furnace, in use during the Roman period, from which the molten slag was tapped and run out rather than collected within its interior. Fifteen iron-rich stones were examined and classed as ores. These were either reddish-orange and moderately friable or grey-black and very hard. Analysis of two fragments, one of each type, by X-ray diffraction indicated that the reddish-orange ore is hematite and the grey-black ore goethite. Both hematite and goethite can be found in the Forest of Dean (Fulford and Allen 1992, 188). Vitrified furnace lining is produced by a high temperature reaction between the clay lining of a hearth or furnace and the alkali fuel ashes or fayalitic slag. It can be formed by iron smelting, iron smithing, non-ferrous metal working and other pyrotechnical processes. Its composition is usually graded, from un-modified clay on one side to a glazed surface or irregular cindery material on the other. Fuel ash slag is the very light, in weight and colour (grey-brown), and highly porous material that results from the reaction between alkaline fuel ash and silicates from soil, sand or clay at elevated temperatures. The reaction is shared by many pyrotechnological processes and the slag is not diagnostic. Energy Dispersive X-ray Fluorescence (EDXRF) analysis shows the presence of silicon and alkalis such as calcium, potassium and sodium with little or no iron. 04_BGAS125_131-246 16/11/07 15:50 Page 185

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 185

Most assemblages of slag include undiagnostic iron-working slag, which is also of fayalitic composition and can be formed during iron smelting or iron smithing. However, in the absence of any clear evidence for the latter, it is probable that the undiagnostic slag also derives from smelting. It is possible that while some slag was removed in the molten state from the furnace, some may have remained inside the furnace.

Moulds Four mould fragments were submitted for examination and a fifth was discovered during the assessment of the iron-working slag (Fig. 16). The first four are all in the same condition and were all made from a grey-green clay tempered with mica, limestone, sandstone and an unidentified black mineral. The mica and the black temper are generally fine (up to 0.5 mm across) while the limestone and sandstone temper is coarser (up to 3 mm across). The softness of the clay fabric of the moulds suggests that they have not been fired to a high temperature. The fifth fragment is superficially similar to the others; the outer surface is a grey-green colour but the inner surface is black. The range of tempers used is similar but the particle size is smaller. The first four moulds would have been used in pairs (front and back or left and right) as two-piece moulds; not enough survives of the fifth mould to be certain whether it was a piece mould or a lost wax mould. The shapes of the individual moulds (and of the artefacts which would have been cast from them) are discussed below. The EDXRF spectra collected from each mould showed the presence of zinc, lead and copper at levels far above what might be expected naturally in the clay or temper. While tin was not detected, previous work has established that tin rarely penetrates mould fabrics to the same extent as lead and zinc. The differences in vapour pressure and other factors make the reconstruction of the alloy used (bronze, brass, gunmetal, etc.) impossible. Nevertheless the moulds clearly have been used, probably to produce copper-alloy objects.

Rf. 17 (context 1037) Dimensions 41 mm by 25 mm by 13 mm. The impression is of a conical-headed pin. The head is 10 mm in diameter at its base and 13 mm high. The shaft is incomplete but c.3 mm in diameter and at least 23 mm in length.

Rf. 21 (context 1066) Dimensions 35 mm by 20 mm by 6 mm. The impression is of a fairly thin curvilinear artefact (surviving dimensions are roughly 25 mm by 20 mm by 2 mm). The depth of the impression and the flatness of the mould suggest that this portion of the mould was the back into which the pattern was first impressed. The surviving portion can be reconstructed as a lunate fitting such as those used to decorate horse harness (cf. Allason-Jones 1991, fig 112, no. 57) or as a brooch (cf. Hawkes and Hull 1947, plate XCVIII, 1703.)

Rf. 22 (context 1067 ) Dimensions 22 mm by 20 mm by 10 mm (approximate). The impression is of a rib or fin emerging from a larger curved object. One possible artefact which could have been produced from such a mould is a Colchester derivative brooch, the fin being the end of the rib running along the bow and terminating at the spring end of the brooch. This mould appears to have been subjected to some localised heating, which has turned the clay orangey-brown. This may be associated with the use of the mould or may have occurred accidentally after the mould was discarded. The shape of the mould, in particular its curvature, suggests that it was the front valve of the two mould halves. 04_BGAS125_131-246 16/11/07 15:50 Page 186

186 TOBY CATCHPOLE

Rf. 23 (context 1065) Dimensions 28 mm by 21 mm by 11 mm. This mould is one side of a left–right mould of a Trumpet brooch (see Hattat 1982, 106) and is similar to one found at Prestatyn (Bayley 1988, fig. 7). The mould is incomplete but shows the brooch from the head to approximately half way down the catch-plate. The discernible features are described in turn, starting at the head. The flat lug protruding from the back of the head has a single opening for an axial pin. The head shows the usual trumpet expansion and the tight angle to the axis of the bow. The central moulding is fairly plain and conforms to Collingwood’s R(i). Ceramic brooch moulds have also been found at Compton Dando, Somerset (Bayley 1985), and Prestatyn, Clwyd (Bayley 1988). See also Cool (above: catalogue no. 7) for evidence of a brooch made in this mould.

Rf. 176 (context 1067) Dimensions 30 mm by 16 mm by 8 mm. This fragment appears to have been fired. The interior surface is reduced-fired black although the outer surface is still a grey-green colour. The fragmentary nature of the mould makes it difficult to be sure that this was a piece mould rather than a lost wax mould. Given the nature of the other moulds from this and other Roman sites in Britain a piece mould is the more likely. The impression is of a curving bar with a single rib. One artefact type that could possibly have been made using this mould is a Colchester derivative brooch but not enough survives of the mould for this to be a certain interpretation.

Conclusions The iron-working slags from Dymock that can be clearly related to a particular process are of types indicative of iron smelting, a finding supported by the recovered ore fragments. The iron was smelted from hematite and goethite using the bloomery process. The small amount of furnace fragments and the small size of the slag assemblage as a whole suggest that smelting did not take place within the area excavated. As it is unlikely that such material would be transported over large distances, the smelting would have taken place somewhere close to the area excavated. The five moulds recovered indicate that copper-alloy casting took place near by. They do not include joining fragments and may represent a small fraction of the total number of moulds originally produced. Full quantification of slags and other metallurgical debris was included in the assessment report (Catchpole 2000) and will be contained in the archive. The 12 kg of slag collected represent a small fraction of that encountered on site.

Coins by Peter Guest Twelve Roman bronze coins and a single 19th-century farthing were recovered during the excavations (see Table 7). The Roman coins include four 1st-century issues, three coins struck during the later 3rd century and four coins of the period 330–348. Although it is statistically unwise to attempt to extract too much information from such a small group, the four coins struck during the 1st century are particularly noteworthy. They include a copy of a Claudian as , dupondii of Nero and Vespasian, and a sestertius of Domitian. All would have been in circulation during the 1st and early 2nd centuries and are most commonly found in south-west Britain on military and urban sites occupied during the early years of the Roman period. Excavations at the legionary fortresses at Usk and Kingsholm, the Sacred Spring at Bath and the fort/town at Cirencester all produced relatively large quantities of the high-value bronze denominations issued by the Roman mints during the 1st century (Boon 1982; Reece 1985; Walker 1988; and Reece 1998). It would appear, therefore, that in terms of coin loss the settlement at Dymock is similar to settlements that were 04_BGAS125_131-246 16/11/07 15:50 Page 187

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 187 t ; s m u a b s i k d g r n a m i m c g m a m f 0 e 6 . . – – – – – – – – R – – l – 1 8 0 8 e 0 – c 1 2 5 0 7 n : 4 7 0 3 4 e 3 4 6 C 1 8 r / : : : I : : e 2 f R C C C K K 2 e I I I s 5 – – – – – R R R a R H – – H g m u u L n / u e e r r t d e e n m m g i i i r r o o u – – – – – R R L M T – T – ] . . P k R … ] r [ T a / P / / / R o ] M h . h . T c t [ r n / n - … i i / a [ r h / – – – – – – – – – M M c / b I ) . C C C s R d S S t S S S n s O E i – – – N V V A A 1 o T I ( a I I c N T T v G C s I I T R R n n n A A a r I : i i i i e Q I I u S V e s C C r ) a a a t s O O V i t t t ) s R R n k A r G V o x R R i r r r c I l T T r d d e a r e e e O O t t G G E E C X v o V e s s C C P M c c c G D L L S I e I X x X V V A n n n s s s w O 1 2 – u D A E ( E E ( u a u – I P A V a R V G a G e g S a U w S s N e N e e e f u A N s N A S I I n n n o A n u i i i I i A I i f f . t t t R R n n r e e S i i d o o s T T n n n 7 o s a a O O I A u t a a a u r S r t t e e O t t t a c e a r r P o s s e T T l s s s i l M s N R e e v S u u n n n e b C C V C c c H E b o o O o o o O a I I E a n n s s s V u u V a D V N a O H C C H a C C C T y r / u t d r n 8 0 8 5 6 0 6 3 e 4 4 4 3 9 7 9 4 e c 2 – – – – – – – 7 6 e t t 8 7 7 0 0 0 8 8 – – h a a 8 t 4 3 3 3 7 6 6 7 1 6 3 1 4 L 3 3 3 3 2 2 2 8 7 6 4 D n o i e e t t t a a a s s i i n y s u u i d d i i p u g a a i d d y o m r r e t n 4 c n n t p i r o . . / a o o o e h n b b i 3 3 3 4 3 t t c p p r r e d s r E E E E E a a u u s a e a f A A A A A b b r s d d a D n o t i x t a e t u l 1 1 4 4 4 0 3 5 0 0 n a 0 0 0 0 0 0 0 4 0 0 s s o 2 v / / 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 u 1 1 u 1 1 C E T d e r e t s i d g 8 9 n e i 5 3 2 3 6 7 0 5 3 9 1 8 9 1 4 4 0 0 0 3 1 1 1 0 2 2 F R 04_BGAS125_131-246 16/11/07 15:50 Page 188

188 TOBY CATCHPOLE

supplied with and used coins from the beginning of the Roman period. This suggests either that Dymock itself was a relatively Romanised settlement (i.e. urbanised or militarised) as early as the later 1st century (Claudian copies do not seem to have circulated long into the 2nd century) or that these coins arrived at Dymock as a result of the settlement’s position on the junction of two Roman roads connecting other urban and military sites.

Flint by David Mullin A total of 28 pieces of struck flint was recovered from the excavation at Dymock. Many (12) were from unstratified context (1003). The assemblage is dominated by waste flakes and narrow blades of Late Mesolithic/Early Neolithic date. Of particular note is an ?Early Neolithic side scraper from context (1114) and a small, Early Neolithic, borer from context (215). A single burnt fragment of flint was recovered from context (219).

Other Finds A very small assemblage of ceramic building material was retrieved, mostly from post-medieval contexts. The Phase 1 enclosure ditch backfills produced 6 fragments with a combined weight of 127 g. Four Phase 2 contexts produced 9 fragments with a combined weight of 185 g. In contrast a total of 10.6 kg of daub was recovered during the excavation and from slag samples. Due to the high quantities present not all daub was collected during the excavation and a sample amount was collected from each context found to be rich in the material. Of the daub collected 74 per cent came from Phase 1 features, 7 per cent from dated later Roman features and the remainder from undated, post-Roman and unstratified contexts. It seems likely that all the daub on the site originated from the Phase 1 structures. Given the amount of iron smelting being carried out in the vicinity there is a possibility that a proportion of this material may have originated from the structure of shaft furnaces but in the absence of any identifiable furnace lining material this is unproven. Querns were notably absent from the finds assemblage.

ENVIRONMENTAL EVIDENCE

Human Skeletal Remains by Marianne Cole Eight articulated skeletons comprising five adults and three sets of infant remains were recovered. The adult skeletons were found in discrete grave cuts all concentrated in the eastern area of the excavation, outside the boundary ditch of the rectilinear enclosure. There was evidence that four of the inhumations were coffin burials. The infant remains were also found in discrete cuts, two of which were circular in form. Two of the burials were aligned east–west. Some of the grave cuts had been truncated by later activity which had an effect on the condition of some of the remains. All eight sets of remains were presented for examination to establish the sex of each individual, the age at death and the living stature. Note was made of the presence of any non-metric traits and also of any changes to the bones indicating trauma or disease. Summaries of all data are shown on Tables 8 and 9. 04_BGAS125_131-246 16/11/07 15:50 Page 189

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 189 ) h l t d a n l w r i e e o t r m M a h g a l t . i r s e w o B e t f ( o y d r l e h o a g l t p a i N e i o t b s e s r y l m – – – t e r s o h o r - o r o p a u e e c o t r m d a n n l e a p h i t o o s e u c s – – M – N N F O S S o a i b i d e T s s U & u . r s r r s e n u u a e i o m n i m m b u t i o e e a H – F – F – T B m u h 6 n i 2 6 . 7 e . 3 h 3 – t / – n f / + o w + e 8 o r y 8 6 r n . . u a t k 6 8 3 8 a n 5 6 5 5 m t . . . . s s s 1 1 1 1 U S m h h h t t t u s n n n n ) ) ) s s s : o o o o o o s h h h w r r r t t t k m m m o e e e r n n n 5 0 5 0 t t t n 2 2 2 o o o o 3 3 3 3 u u u k e – – – – w – – – n m m m g n n n 5 5 5 5 / / / i i i e 2 2 2 2 U A + ( 8 + ( 8 + ( 8 g a w e n n n S w w w . e e e e o o o 8 l l l l a a a a n n n e e l k k k l x m m m m a n n n b e e e e e a U F U M U F F F S T t e n n e o s b % % % % % % % % e r 0 0 0 8 0 5 5 0 5 9 4 9 8 7 9 5 % p d n o o i o t i G d d d d d – r o o o o n r r r o i i i o o o o o o a a a G F F G G F G P C 1 0 2 2 8 5 n 0 0 1 1 1 1 o t t t t e n n n l 8 1 3 0 7 a a a e 5 4 9 9 6 f f f k 2 2 1 1 1 n n n I 1 I 1 I 1 1 1 S 04_BGAS125_131-246 16/11/07 15:50 Page 190

190 TOBY CATCHPOLE

Table 9. Sewage works: human adult burial dental health and non-metric traits.

Skeleton Ante- Caries Calculus Cranial non- Post-crania mortem Loss metrics l non-metrics 1167 No data ––Traits absent – 1190 06Medium Traits absent – 1193 20Medium Traits absent – 1241 03Slight Lambda wormians (3) Parietal Foramen (2) – 1258 15Medium Supra-orbital Os acromiale foramen (1) (Bilateral)

Condition Apart from (1167), the overall condition of the assemblage was good. The skeletons were from 40 to 98 per cent complete. Evidence of post-mortem breakage, in some cases extreme, was possibly due to post-deposition ploughing that had truncated some of the graves.

Sex and Age The completeness of the assemblage enabled sex and age to be determined with a degree of accuracy. The features used to establish the sex of the skeletons included pelvic morphology, notably the sub-pubic angle (Bass 1995), the sciatic notch (Brothwell 1981) and the presence of a pre-auricular sulcus. Cranial morphology was also used by looking at such features as the prominence of the brow ridges, mastoid process and occipital protuberance (Bass 1995). In addition, measurements were taken of the femoral and humeral heads to confirm the sex using standards set out by Buikstra and Ubelaker (1994). It is widely believed that using the pelvis and crania together can determine the sex with 90–-95 per cent accuracy. Each of the adult skeletons possessed at least one bone usable for determining sex. No sex estimation was attempted for the infant remains. Determining sex in sub-adult remains is highly unreliable because the skeletal traits that are used do not usually manifest themselves until the onset of puberty. Age determination for the adult skeletons was performed using the teeth and the pubic symphysis. For the teeth the amount of wear present on the occlusal surface can be attributed to age (Brothwell 1981). The morphology of the pubic symphysis, the surface of which changes with age (Webb and Suchey 1985), can be used for both males and females but the degree to which it is affected by pregnancy and childbirth should be considered in studying females. Only skeleton (1167), a very fragmentary set of remains lacking the necessary criteria, could not be aged Infant ages were established using a chart devised by Powers (unpublished). This chart was constructed using modern infant measurements and this may affect its applicability to this group. The measurement of surviving long bones is a standard technique for estimating sub-adult age, although at present no technique for infant ageing is considered as reliable as those used on adult skeletons. All three infant skeletons possessed at least one bone that could be measured. All three were of approximately the same age, 8 months in utero , suggesting that all died at or around birth. It is likely that Romano-British infant birth lengths were shorter than those of modern infants, which would explain the slight underestimate of infant age. One infant skeleton contained a jaw fragment and the developing crowns were used to confirm the age estimate (Ubelaker 1978). 04_BGAS125_131-246 16/11/07 15:50 Page 191

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 191

Stature Stature could be calculated for four of the five adult skeletons. Again skeleton (1167), due to its fragmentary nature, was the exception. The stature for the other four was calculated using long bone measurements. Where possible the long bones with the least margin of error were used. The measurements were taken and then compared to an equation set out by Trotter and Gleser (1952). Given the data from other contemporary sites the skeletons from Dymock are within the normal range of expected stature.

Dental Health All of the adult skeletons except 1167 had dentitions available for examination. The four adults possessed both mandibles and maxillae in varying states of completeness. The number of teeth present out of a total of 128 (4 × 32) was 103. Some 2.9 per cent of the teeth (n=3) were lost ante mortem , 4.8 per cent (n=5), wisdom teeth, had not erupted, either bilaterally or on one side only, and 19.4 per cent (n=20) were lost post mortem and were unaccounted for. There was no evidence for periodontal disease, enamel hypoplasias or abscesses. Three of the skeletons showed evidence of dental caries. Skeleton 1190 displayed decay on 25 per cent (n=6) of the teeth present. Calculus deposits were recorded on all of the dentitions (Brothwell 1981). The quantity varied from slight to moderate in most cases. Calculus is a deposit of mineralised plaque that tends to be very brittle when found on buried remains. It often flakes off during excavation or examination and the level of calculus found in death is not a reliable indicator of the amount in life. The build up of calculus is thought to reflect poor dental hygiene. Infant 1152 possessed a mandible in which four developing crowns were found. The other two infants had loose crowns but no mandibles or maxillae associated with them.

Non-Metric Traits Non-metric variation ‘is generally taken to encompass any minor anomaly of skeletal anatomy not normally recorded by measurements’ (Mays 1998). There is a large quantity of features that classify as non-metric traits, for example parietal foramen, lumbar sacralisation and the retention of the metopic suture. For the purposes of this study 21, made up of 11 cranial and 10 post-cranial, traits were noted only for the adult skeletons. Skeletal non-metric traits would have had no effect on the living individual. Some research suggests that they may be genetically predetermined and could thus be of use in establishing familial relationships. This is by no means conclusive and it is currently accepted that non-metric traits could simply be a manifestation of natural variation within a population.

Evidence of Trauma or Disease There was no evidence of trauma on any of the remains. There was limited evidence of spinal change in some of the adult skeletons. Skeleton (1241) displayed Schmorls Nodes on the lumbar spine. They represent the herniation of an intervertebral disc into the body of the vertebrae and the proposed causes of this phenomenon include extended activity, trauma, degenerative disease and possibly metabolic disorders. Several of the adult skeletons also showed the beginning of osteophyte growth around some of the vertebrae. This is usually found in greater profusion as the individual ages. The small amount found on this set of remains is consistent with the proposed ages of the skeletons. There was no further evidence of disease. 04_BGAS125_131-246 16/11/07 15:50 Page 192

192 TOBY CATCHPOLE

Conclusions This set of human remains represents five adults, one male and four female, and three infants. All the adults were aged between 25 and 35 and none of their remains showed any evidence as to mode of death. There were no indicators of disease or poor health and, consistent with their age, there was little or no degenerative change to the skeletons. The dental health of the adult assemblage remains within the range of variation displayed at sites of this period. The Dymock skeletons had generally good dental health with no abscesses, and the distribution of caries was again within the normal range. The stature of the individuals from Dymock falls within the standard range and the non-metric traits found on the assemblage is again within the expected range. The presence of os acromiale on skeleton 1258 is a reasonably rare finding but would have had no effect on the individual during life. The group also included three infant skeletons all approximately 8 months ( in utero ) in age. This finding is based on a chart devised using modern data and longer birth lengths in modern infants will account for the discrepancy in age. It is likely that these infants died at birth. Mays in his 1995 paper argues strongly in favour of infanticide as a major cause of infant death in this period. He suggests that infanticide ‘until recently was a widely practised and tolerated means of population control’. The age at which such deaths commonly occurred corresponds to the ages at death of the infants at Dymock. As no actual manner of death is agreed upon for cases of infanticide it is difficult to prove infanticide conclusively. There was no cause of death visible on these remains and it is impossible to tell whether these infants were victims of infanticide, were stillborn or died naturally.

Animal Bone by Claire Ingrem Animal bone was recovered from numerous features, including ditches, pits, other linear features and a possible well, and was analysed with the aim of investigating animal husbandry and general economic practices. Details of a small quantity of animal bone recovered from medieval and post- medieval deposits can be found in the earlier assessment report (Baxter 1999).

Methodology The animal bones were identified and recorded at the Centre for Applied Archaeological Analyses (CAAA), University of Southampton. All anatomical elements were identified to species where possible with the exception of ribs and vertebrae, which were assigned to size categories. Mandibles and limb bones were recorded using the zonal method developed by Serjeantson (1996) to calculate the minimum number of individuals (MNI); this is based on the most numerous zone of a single element taking into account side. In addition, all bone fragments over 10 mm in the hand-recorded material were recorded to species or size category to produce a basic fragment count of the Number of Identified Specimens (NISP). Fragments categorised as large mammal are likely to belong to horse or cattle and those in the medium mammal category to sheep/goat or pig. For the purposes of this report these are included in the count of identifiable fragments. The presence of gnawing and butchery together with the agent responsible was recorded as was evidence for burning. Measurements were taken according to the conventions of von den Driesch (1976). The wear stages of the lower cheek teeth of cattle, sheep and pig were recorded using the method proposed by Grant (1982) and age was attributed according to the method devised by Payne (1973) and Legge (1992). The fusion stage of post-cranial bones was recorded and age ranges estimated according to Getty (1975). Measurements of the crown height of horse teeth were recorded and age was estimated according to the method of Levine (1982). 04_BGAS125_131-246 16/11/07 15:50 Page 193

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 193

A selected suite of elements was used during recording to differentiate between sheep and goat according to the methods of Boessneck (1969) and Payne (1985). These were the distal humerus, proximal radius, distal tibia, metapodials, astragalus, calcaneus and deciduous fourth premolar. No elements belonging to goat were positively identified. Birds identified as galliforms can include domestic chicken, pheasant, and guinea fowl, but as there is little or no good evidence for either of the last two in Roman Britain, assemblages from this period will mainly be chicken or ‘domestic fowl’. Amphibians were differentiated according to the morphology of the illium (Holman 1985).

Data A total of 966 fragments of animal bone was recovered by hand collection of which 63 per cent is identifiable to taxon or size category. Most of the animal bone came from deposits dated to the 1st–early 2nd century AD and mid 2nd century AD with only a small quantity recovered from late 2nd-century or later deposits (Table 10a). Overall, sheep/goat are the most numerous taxa according to NISP, although a consideration of the large and medium mammal categories suggests that cattle may have been equally numerous. Sieved samples taken from selected features produced 919 fragments of animal bone of which 177 (19%) were identifiable to taxa or size category. Again, most came from deposits dated to the 1st–early 2nd century AD and mid 2nd century AD (Table 10b). The species lists for the two recovery methods are similar although sieving has allowed the recovery of a few bones belonging to small species such as woodcock ( Scolopax rusticola), rodents and amphibians which may have been missed by hand collection.

1st–early 2nd century AD Caprines dominate the remains from 1st–early 2nd-century AD deposits. They are almost twice as numerous as cattle (Table 11a) and represent 21 per cent of the identifiable assemblage. Pig is the least frequent of the major food animals (8%). Horse is represented by a few bones and, apart from eleven galliform specimens, is the only other domestic animal present. The remains of wild mammals are scarce. Roe deer ( Capreolus capreolus ) is represented by a fragment of antler, red deer ( Cervus elaphus ) by a metacarpal with an unfused distal epiphysis, and hare ( Lepus europaeus ) by a radius and ulna and a molar tooth. Several bones belonging to rodent and frog ( Rana spp.) were recovered from a pit and probably represent pit fall victims. Cattle and caprines are fairly equally represented in the boundary ditch and beam slots but cattle are almost absent from non-boundary ditches and poorly represented in the pits (Table 11a). The predominance of the most robust elements such as loose teeth, mandible and tibial fragments is an indication that the assemblage has been biased by the effects of density mediated preservation. Despite this bias, and the small sample size, cattle, caprines and pig are represented by elements from all parts of the body – both major meat bearing bones and extremities (Table 12a). Tables 11a and 12a also indicate that whilst horse is present in small numbers their bones were recovered from a variety of feature types. Most of the galliform remains came from a pit (context 1198) and probably represent a partial skeleton. Ageing data from both tooth eruption and wear (Table 13) and epiphyseal fusion are scarce. A single cattle mandible is able to provide ageing data; it belonged to an animal aged between 15 and 26 months at the time of death. Thirteen sheep/goat specimens provide dental ageing data: five derive from animals slaughtered between one and two years of age; one belonged to a caprine that died between six and twelve months; and the remainder were from individuals aged over three years at the time of death. Both of the two ageable pig mandibles belong to immature animals. Epiphyseal fusion data, although scarce, suggests a similar picture indicating that at least one of 04_BGAS125_131-246 16/11/07 15:50 Page 194

194 TOBY CATCHPOLE

Table 10. Sewage works: animal bone species representation according to phase (NISP).

a) hand 1st–early 2nd mid 2nd late 2nd Unspeci- ?RB Total collected century AD century century fied or later Roman n% Horse 8– – – – 81 Cattle 46 42 21394 16 Sheep 11 4–– – 15 2 Sheep/goat 72 51 – – 4 127 21 Pig 29 15 1 – – 45 7 Red deer 1 – – – – 1 <1 Hare 11 – – – 2 <1 Galliform 10 2 – – 2 14 2 Bird 22 – – – 41 Large mammal 72 89 9 – 10 180 30 Medium mammal 74 38 – – 1 113 19 Small mammal 1 – – – – 1 <1 Unidentifiable 237 117 125362 Total 564 361 13 3 25 966 No. identifiable 327 244 12 1 20 604 % identifiable 58 68 92 33 80 63

b) sieved samples 1st–early 2nd century mid 2nd century Total – – No. % Cattle – 332 Sheep/goat 8 17 25 14 Pig 43 7 4 Roe deer 1 – 11 Hare 1 – 11 Galliform 11 2 1 Woodcock – 221 Bird 2 10 12 7 Rodent 3811 6 Frog 2 – 21 Amphibian 54 – 54 31 Large mammal 1 11 12 7 Medium mammal 12 8 20 11 Small mammal 22 3 25 14 Unidentifiable 323 419 742 Total 434 485 919 No. identifiable 111 66 177 % identifiable 26 14 19 04_BGAS125_131-246 16/11/07 15:50 Page 195

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 195

Table 11. Sewage works: animal species representation according to feature (NISP).

a) 1st–early Beam Boundary Ditch Gully/ Pit Probable 2nd century slot ditch linear well Total AD feature n % Horse 22 3 – 1 – 8 2 Cattle 9 19 2311 2 46 11 Sheep/goat 8 15 21 3 39 5 91 21 Pig 54 9 111 3 33 8 Roe deer – – – – 1 – 1 <1 Red deer – – 1 – – – 1 <1 Hare – – – – 2 – 2 <1 Galliform 11 – – 9 – 11 3 Bird – 2 – – 2 – 4 1 Rodent – – – – 3 – 3 1 Amphibian – – – – 54 – 54 12 Frog – – – – 2 – 2 <1 Large mammal 15 29 2222 3 73 17 Medium mammal 11 12 6250 5 86 20 Small mammal – – – – 23 – 23 5 Total 51 84 44 11 230 18 438

b) mid 2nd Ditch Grave Gully/ Pit Total century AD linear feature No. %

Cattle 4332 6 45 15 Sheep/goat 9 26 21 16 72 23 Pig 14 8 518 6 Hare – – 1 – 1 <1 Galliform 1 – 2 – 31 Woodcock 2 – – – 21 Bird 81 3 – 12 4 Rodent 43 1 – 83 Sm. mammal 21 – – 31 Med. mammal 9716 14 46 15 Lg. mammal 10 6 53 31 100 32

Total 50 51 137 72 310 04_BGAS125_131-246 16/11/07 15:50 Page 196

196 TOBY CATCHPOLE

Table 12. Sewage works: anatomical representation of major species (NISP).

a: 1st–early Horse Cattle Sheep/ Pig Roe Red Galli- Hare Large Medium 2nd century Goat Deer Deer form Mammal Mammal AD – – – – – – – – – – Zygomatic – 1 – – – – – – – – Occipital condyle – 1 – – – – – – – – Frontal – 1 – – – – – – – – Premaxilla – 11– – – – – – – Maxilla – 215– – – – – – Mandible 14 17 4 – – – – – – Teeth 3 16 12 1 – – – 1 – – Hyoid – – 1 – – – – – – – Atlas – 1 – – – – – – – Scapula – 342– – – – 21 Coracoid – – – – – – 1 – – – Humerus – – 42– – 1 – – 1 Radius – 2 12 – – – 1 – – – Radius & ulna – – – – – – – 1 – – Ulna – 115– – 2 – – – Pelvis – 252– – – – 1 – Femur 2 – 1 – – – – – – – Tibia – 3 13 3 – – – – 11 Fibula – – – – – – 1 – – – Tibiotarsus – – – – – – 2 – – – Astragalus – – 1 – – – – – – – Calcaneus 1 – 11– – – – – – Metacarpal – 15– – 1 – – – – Metatarsal – 29– – – – – – – Carpometacarpal – – – – – – 3 – – – Lateral – – – 2 – – – – – – metapodial 1st phalanx – 12– – – – – – – 3rd phalanx 1 – 11– – – – – – Sacrum – 1 – – – – – – – – Thoracic vert. – – – – – – – – 12 Lumbar vert. – – – – – – – – – 1 Caudal vert. – – – – – – – – – 2 Rib – – – – – – – – 4 10 Skull frag. – 3 – 5 – – – – 2 – Antler frag – – – – 1 – – – – – Long bone frag. – – – – – – – – 15 21 Vertebra frag. – – – – – – – – 77 Rib frag. – – – – – – – – 17 40

Total 8 46 91 33 11 11 2 50 86 04_BGAS125_131-246 16/11/07 15:50 Page 197

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 197

b: mid 2nd Cattle Sheep Pig Galli- Hare Large Medium century AD / Goat form Mammal Mammal

Premaxilla – 1 – – – – – Maxilla 21 1– – – – Mandible – 32– – – – Tooth 5 14 3 – – – – Atlas – 1 – – – – – Scapula 33 – – – 21 Humerus 34 3– – – 1 Radius 24 1– – – – Ulna 21 – – – – – Radius & ulna – – – – 1 – – Pelvis 23 – – – – – Synsacrum – – – 1 – – – Femur – 322– – – Tibia 2 12 1 – – 1– Lateral malleolus 21 – – – – – Astragalus 11 – – – – – Calcaneus 1 – – – – – – Navicular cuboid 1 – – – – – – Cuneiform 1 – – – – – – Sesamoid – – – – – 1– Metacarpal 12 – – – – – Metatarsal 11 – – – – – Metapodial 33 – – – 1– Lateral metapodial – – 1 – – – – 1st phalanx 21 1– – – – 2nd phalanx – 11– – – – 3rd phalanx 21 – – – – – Caudal vert. – – – – – – 1 Rib – – – – – 14 Skull frag. 9 – 2 – – 5 Long bone frag. – – – – – 9 15 Vertebra frag. – – – – – 24 Rib frag. – – – – – 41 20 Total 45 61 18 31100 46

c: late 2nd century AD Large or later Cattle Pig Mammal

Mandible 1 – – Lower premolar – 1 – Metatarsal 1 – – Long bone frag. – – 4 Rib frag. – – 1 Total 21 9 04_BGAS125_131-246 16/11/07 15:50 Page 198

198 TOBY CATCHPOLE

Table 13. Sewage works: estimated animal age according to tooth eruption and wear.

P4 M1 M2 M3 Estimated age

Cattle 1st – early 2nd century (j) f–– 15–26 months

Sheep/Goat 1st – early 2nd century (g) b–– 6–12 months 1st – early 2nd century (h) fb– 12–24 months 1st – early 2nd century fd– 12–24 months 1st – early 2nd century (l) fd– 12–24 months 1st – early 2nd century (n) g – – 12–24 months 1st – early 2nd century gd e 12–24 months 1st – early 2nd century hkge 3–4 years 1st – early 2nd century – – e 3–4 years 1st – early 2nd century ghgd 3–4 years 1st – early 2nd century hgfe 3–4 years 1st – early 2nd century – hg 4–8 years 1st – early 2nd century mj h 8–10 years 1st – early 2nd century lmmj 8–10 years

mid 2nd century (k) gd – 12–24 months mid 2nd century – – e 3–4 years

Pig 1st – early 2nd century fb– 1st – early 2nd century – ce

the cattle died before reaching 24–30 months; that most caprines had been slaughtered before reaching four years; and that pigs were slaughtered whilst immature. In addition, the presence of a radius with the proximal epiphysis unfused is evidence for the death of at least one lamb/kid below four months of age. The only ageing data available for horse comes from a mandibular molar recovered from a ditch; it belonged to an animal aged between seven and three quarters and nine and a quarter at the time of death. Two pig maxillae able to provide sexing information both belonged to males. A few bones show evidence for gnawing by canids (Table 14a). Several bones exhibit evidence for butchery; the majority have cut (rather than chop) marks which were seen on the remains of all the major domestic taxa (Table 14b). A considerable number of bone fragments have been burnt although few burnt bones were identifiable to either taxa or size category (Table 14c). Metrical data is shown in Table 15. Apart from a cattle radius that is slightly smaller, where comparison was possible all measurements fall within the range collected from contemporary sites and held on the Animal Bone Metrical Archive Project (ABMAP) (http://ads.ahds.ac.uk/catalogue/ specColl/abmap/index.cfm).

Mid 2nd century AD The assemblage recovered from mid 2nd-century AD features is slightly smaller in terms of NISP than that dated to the 1st–early 2nd century AD, with a total of 310 identifiable fragments 04_BGAS125_131-246 16/11/07 15:50 Page 199

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 199

Table 14. Sewage works: incidence of taphonomy according to phase (NISP).

a: gnawing 1st–early 2nd century AD mid 2nd century

Cattle – 4 Sheep/goat 41 Pig 31 Large mammal 11

Total 87

b: butchery 1st–early mid 2nd 2nd century AD century Cut Chop Cut & chop Cut

Cattle 12 – 2 Sheep/goat 3 – – 3 Pig 21 1 1 Galliform 2 – – 1 Large mammal 7 – – 8 Medium mammal 3 – – 4

Total 18 3119

c: burning 1st–early mid 2nd late 2nd 2nd century AD century century or later Calcined Charred Calcined Charred part burnt Charred

Cattle – 1 – 11– Sheep/goat – 2 – 2 – – Pig – 2 – – – – Hare – 1 – – – – Large mammal 1 – – 2 – – Medium mammal 25 – 5 – – Small mammal 1 – – – – – Unidentifiable 85 77 60 16 – 1

Total 89 88 60 26 11

recovered (Table 11b). Species representation is similar, although sheep/goat (23%) and cattle (15%) are slightly better represented in this phase at the expense of pig (6%) and horse. This later assemblage does however differ in terms of fragments assigned to mammal size categories with fragments belonging to large mammals twice as numerous as those belonging to medium size mammals (Table 12b). Three bones belonging to galliforms are the only other remains of domestic 04_BGAS125_131-246 16/11/07 15:50 Page 200

200 TOBY CATCHPOLE 1 d . 5 4 1 D 1 f 6 3 5 . . . d 7 e – – 3 6 8 1 B B 2 1 4 3 5 6 . . . 8 6 d D C . . – – 4 3 3 B 7 7 S B 1 2 5 9 2 d C p . . 4 9 2 2 2 i C . . . D – – 1 3 L 1 1 6 8 9 S S D D 1 4 S 4 3 2 3 3 p . . . . . 9 p G C . a – – 7 7 7 1 4 t B 4 S D B 1 1 1 4 4 a d l a c i r P 1 6 2 8 2 8 1 5 5 5 5 3 6 1 p t m p d ...... p P 3 i e m – F L L 9 7 7 9 3 0 9 1 6 1 9 4 7 7 1 B B D L e D 1 2 1 1 2 5 1 8 1 8 3 2 5 5 m B G G p l y a t m t i ) n n E 1 a e ( . 1 5 4 5 3 7 7 7 5 d ...... 5 L L L L : p d p 0 2 – – G . 2 s m 2 9 0 4 3 6 5 2 7 9 1 2 . B B B 8 2 D G G G G L k e 6 3 2 2 5 8 1 2 4 0 1 r r 6 o u s w a e e g a M w s l e l l l a s u a a a S s s t p u s s s r l . u r r r r n a a a r a 5 a a a l s s s r r t e e g c t t t 1 u u u u u u a a a o a a a a a i i i a a i i i i m p t t t t m r e d d d n n l m m b b b b t a e e e e e u l l i i i i a a a l s e e c b a H U U R T T T A M M M F F T M R R S E T t a o m m m m m m r r r r r r g / o o o o o o e e e e e f f f f f f p p p p p p p l l l l l i i i i i i a e e e e e e e l l l l l l t t t t t x l l l l l l t t t t t e e e e e e e a a a a a a a a a a a a h h h h h h h S G G G S S S C C S G G G C S S C C T y y y y y y y y y y y y r r r r r r r r r r r r u u u u u u u u u u u u t t t t t t t t t t t t n n n n n n n n n n n n y y y y y y e e e e e e e e e e e e r r r r r r c c c c c c c c c c c c u u u u u u t t t t t t d d d d d d d d d d d d n n n n n n n n n n n n n n n n n n e e e e e e 2 2 2 2 2 2 2 2 2 2 2 2 c c c c c c y y y y y y y y y y y y l l l l l l l l l l l l d d d d d d r r r r r r r r r r r r n n n n n n a a a a a a a a a a a a e 2 2 2 2 2 2 s e e e e e e e e e e e e a – – – – – – – – – – – – d d d d d d t t t t t t t t t t t t i i i i i i h s s s s s s s s s s s s 1 1 1 1 1 m m 1 1 m m 1 m 1 1 1 m 1 P 04_BGAS125_131-246 16/11/07 15:50 Page 201

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 201

animals present. Wild species are again poorly represented, this time by a radius and ulna belonging to hare retrieved from a probable gully and two bones belonging to woodcock retrieved from a ditch. While the assemblage is too small to examine body part representation in detail, the major domesticates (cattle, caprines and pig) are represented by elements from all parts of the body. As in the case of 1st–early 2nd-century deposits there is clearly a predominance of tibia fragments belonging to sheep/goat. Caprines are more numerous than cattle in all feature types except gullies and other linear features. Consideration of the large and medium size mammal categories, however, suggests that the remains of cattle may be as numerous as caprines in the pits (Tables 11b and 12b). Ageing data is again scarce. There is no evidence for immature cattle: of the seven bones that could provide epiphyseal fusion data all were fused. Two specimens, a mandible and an isolated mandibular molar, were able to provide data for sheep/goat suggesting that at least one animal was between one and two years old at the time of death and another between three and four years. The presence of several bones with unfused epiphyses provides evidence for the presence of immature sheep/goat; a distal humerus attests to the death of a lamb/kid aged less than four months. As in earlier phases there is no evidence for mature pigs. The one pig mandible able to provide sexing information belonged to a male. A few bones display evidence for gnawing (Table 14a). The frequency of butchered bones is highest in this period with nineteen possessing cut marks (Table 14b). A large number of fragments have been burnt. Most are calcined and are unidentifiable whilst several of the charred specimens are identifiable (Table 14c). Metrical data is shown in Table 15. Where comparison was possible all measurements fall within the range collected from contemporary sites and held on ABMAP (http://ads.ahds.ac.uk/ catalogue/specColl/abmap/index.cfm).

Late 2nd century AD or later This phase produced few identifiable bone fragments: two (a mandible and metatarsal) belonging to cattle, one (a premolar) to pig and nine to large mammal (Table 12c) and all coming from pits. Ageing data are absent. None of the bones shows evidence for gnawing or butchery. A single, unidentifiable fragment is charred.

Discussion and Interpretation The relative frequency of the major domesticates (cattle, caprines and pig) represented at Roman sites has for some time been shown to vary according to the type of site and the degree of ‘Romanisation’. Thus high frequencies of sheep/goat are commonly found on native rural sites whilst military and other ‘Romanised’ settlements tend to display higher frequencies of cattle and pig (King 1978; 1991). More recently it has been suggested that differences in the proportion of the major food animals reflect not only changes in dietary choice but also fundamental shifts in animal husbandry strategies, all occurring in response to the socio-economic change that followed the Conquest (Hamshaw-Thomas 2000, 168). At Dymock, whilst caprines are more numerous than cattle in terms of fragment counts, the larger size of cattle suggests that beef was the meat more often eaten. The increased relative frequency of fragments belonging to large mammal in mid 2nd-century deposits hints that the numbers of cattle increased over time. The small sample of ageing data suggests that caprines were valued for both meat and wool. During the 1st–early 2nd century AD a number of caprines were slaughtered before reaching two years. That suggests that the production of prime meat was important. However, an almost equal number were kept until they reached at least four years of age and this is likely to reflect animals 04_BGAS125_131-246 16/11/07 15:50 Page 202

202 TOBY CATCHPOLE

being managed for their wool . King (1991, 17) notes that caprine assemblages from both urban and rural sites tend to be adult-dominated and suggests that reflects the development of a wool economy. This has also been argued by Maltby (1994, 96) who observed that sheep from both Winchester Northern Suburbs and nearby Owslebury were aged over three years old when slaughtered. At Dymock there is no evidence to suggest that either cattle or pigs were kept into adulthood. For pigs the explanation for the lack of such evidence might be their primary role as meat producers. For cattle it might possibly be the small sample size. The possibility that Dymock was a consumer, rather than a producer, site is suggested by the presence of male pigs; most males would have been sent for slaughter as soon as they reached the required meat weight and females kept into adulthood for breeding purposes. The presence of bones belonging to foetal or neonatal animals reveals that sheep/goat at least were being bred in the surrounding hinterland and the presence of elements belonging to all parts of the carcass indicates that cattle, caprines and pig arrived at Dymock on the hoof. This is unsurprising for a site of this period. The mixed nature of the deposits recovered from urban sites such as Silchester (Ingrem n.d.) and Exeter (Maltby 1979, 87) suggests that all parts of the skeleton were distributed along with the meat. At some Roman sites large dumps of primary butchery waste have been found (Maltby 1985) and, although this is not the case at Dymock, the variation in species frequency between the boundary ditch, other linear features and pits does suggest the existence of differential disposal practices related to animal size. It has often been suggested that primary butchery, especially of large animals, is most likely to have taken place on the periphery of settlements with the waste being dumped away from the main settlement areas whilst smaller animals such as caprines and pigs are more likely to have been cooked whole with their bones disposed of in centrally located pits (ibid.). The effects of taphonomy on bone survival should not be forgotten as it is likely to have created considerable bias in the archaeological record. Bone preservation is well known to be density dependant and for this reason the bones of sheep-sized animals are less likely to survive than are those of cattle-sized animals (Lyman, 1994). The burial context is also likely to have a considerable effect on bone survival; pits generally offer protective micro-environments more likely to preserve the bones of small species than conditions pertaining in ditches and shallow features. At Dymock fragments belonging to cattle and large animals tend to be better represented in ditches and other linear features. Although this could be a genuine reflection of size-related disposal practices it could equally represent differential survival of the remains of cattle-sized animals. Horses are generally relatively poorly represented at Roman sites, especially on urban sites and in deposits representing butchery waste (Maltby 1981, 184). This may result from their treatment being different from that of animals whose main purpose was to provide food, both during and after life. Variation in the proportion of horse bones recovered from rural, compared to urban, sites in Britain has been noted by Maltby (1994 89) who suggested that a greater emphasis on the acquisition of beef by the urban population resulted in the retention and disposal of working animals at rural settlements. Wild species are commonly found in small numbers on Roman sites and the presence of roe deer, red deer, hare and woodcock suggests that hunting was an occasional pursuit.

Conclusions The assemblage of animal bone recovered from Dymock, although relatively small in terms of the number of identifiable fragments, in general conforms to the pattern displayed by contemporary sites. The high proportion of caprines and correspondingly low proportions of cattle and pig during the 1st–early 2nd century AD suggest that the site was not particularly ‘Romanised’ in character. 04_BGAS125_131-246 16/11/07 15:50 Page 203

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 203

However, there is some evidence to suggest that, in line with national trends, wool was of considerable importance and that the contribution made by cattle may have increased with time.

Archaeobotanical Evidence by Julie Jones

Methodology Samples for environmental assessment were taken during the excavation from features associated with two of the timber buildings, from pit and ditch fills and deposits associated with the inhumations. The samples, which varied from 1.55 kg/1.5 litres to 28.8 kg/30 litres, were wet sieved in a flotation tank to a minimum mesh size of 250 microns for the float and 500 microns for the residue. The floats and residues were then dried and examined for charred plant remains. An assessment of the charred plant remains (Jones 2000) showed that in general most deposits produced low concentrations of charred material with variable quantities of charcoal. Preservation of the charred cereal grain and chaff was poor, with many grains fragmented, although the weed seeds were in good condition for species identification. Two samples were recommended for full analysis, a wood fill (1045) from Phase 2 pit [1038], which included the greatest concentration and best preserved assemblage of cereals, and the fill (1186) of a Phase 1 pit [1185], which had a fairly rich weed assemblage. Despite the low concentrations of charred remains in the remaining samples the results from all samples are shown in Tables 16 and 17 and are included in the following discussion. Plant nomenclature and habitat information are based on Stace (1997).

Phase 1 Features Structure A (contexts 1119, 1128, 1127, 1219) Four samples were examined from features associated with Structure A. Context (1119), the fill of post pit [1120] and the fill of construction trench [1069], is thought to be the only deposit contemporary with the use of Structure A. A small float (<1ml) from the feature consisted of small charcoal fragments with a single unidentifiable cereal grain plus two weed seeds. The residue included a few large pieces of burnt clay, with several examples showing the impressions of plant material including grain and possible stem fragments. One of the smaller pieces contained the impression of what appeared to be textile. Context (1128), the fill of posthole [1129] in construction trench [1069], may also be contemporary with the construction phase, but contained only charcoal fragments and a single grass (Poaceae) caryopsis. Both (1119) and (1128) were sealed by a deposit of burnt daub, (1083), associated with the destruction of the building. The small float from (1128) only produced fine charcoal fragments. The fill (1127) of a post pipe [1129] from within the wall of Structure A (1127) included burnt clay fragments with only four charred weed seeds. (1219) was a sample of weathered natural from within the interior of Structure A. Apart from a few charcoal fragments only a single barley, one unidentifiable cereal grain, and one seed of cleavers ( Galium aparine ) were recovered.

Structure B (context 1032) Only one sample was recovered from Structure B. This came from context (1032) which formed the burnt fill of post [1033]. A single grass caryopsis ( Poa/Phleum ) and several dock ( Rumex ) seeds were recorded. 04_BGAS125_131-246 16/11/07 15:50 Page 204

204 TOBY CATCHPOLE

Structure G (context 1198) Context (1198) was the lower fill of Structure G [1195]. Three wheat ( Triticum ) and three oat (Avena ) grains were present together with a small assemblage of arable and grassland weeds including narrow-fruited cornsalad ( Valerianella dentata ) and barren brome ( Anisantha sterilis ). The single hazel ( Corylus avellana ) nut fragment is likely to relate to hazel wood used as a fuel that had become incorporated in this deposit.

Pits (contexts 1186, 1118 and 1053) Circular pit [1185], which lay within the area of Structure D, contained an ashy secondary fill (1186), including a small assemblage of cereal grains, such as wheat, hulled barley ( Hordeum ) and oats, with a few spelt wheat ( Triticum spelta ) and other hulled wheat glume bases and several silicified wheat/barley awns. Although the cereal component of the sample was limited, 461 weed seeds were recorded from several habitat groups. Arable weeds, likely to have been associated with the cereal crop, include fat-hen ( Chenopodium album ), orache ( Atriplex ), black bindweed ( Fallopia convolvulus ), cleavers and scentless mayweed ( Tripleurospermum inodorum ). There are, however, also many grasses (Poaceae), fescues ( Festuca ), meadow-grass/cat’s-tail ( Poa/Phleum ) and brome (Bromus ), with other taxa such as yellow rattle ( Rhinanthus minor ), dock, vetches ( Lathyrus/Vicia ), clover/medick ( Trifolium/Medicago ) and ribwort plantain ( Plantago lanceolata ), typical of grassy habitats, with further indications of damp ground taxa such as sedge ( Carex ), spike-rush ( Eleocharis palustris/uniglumis ) and rush ( Juncus ). Context (1118) was the fill of a small circular pit or posthole [1117]. The fill had an ashy nature with fragments of burnt bone and it was originally thought that this represented a cremation. Bones, including some burnt fragments, were fairly numerous and have been identified as amphibian and rodent remains (Ingrem, above). The float included few charcoal fragments, with several wheat and oat grains, a single wheat glume base and weed seeds including dock, sedge and meadow-grass/cat’s-tail. Context (1053), the fill of a large sub-rectangular pit [1054] which cut the backfilled construction trenches of Structure A, only produced small charcoal fragments and a single dock seed.

Phase 2 Features Inhumation s (contexts 1192 and 1166) Samples were taken from fills of two of the adult inhumations. Context (1192) was the fill of grave cut [1194]. The charred remains included a single oat grain, a wheat glume base, plus several weed seeds. The fill (1166) of grave cut [1168] produced five wheat and five oat grains, with 22 wheat glume bases and 7 spelt glume bases. The weed assemblage was fairly limited with the arable weeds brome and redshank ( Persicaria maculosa ), plus grasses and a few vetches ( Lathyrus/Vicia ).

Ditch (context 1084) Context (1084) was the secondary fill of ditch [1020]. It produced abundant charcoal, although cereal remains were sparse with three wheat, five barley and four oat grains plus spelt wheat, glume bases and an oat awn. A small assemblage of weed seeds included the same range of arable weeds as found in the gully fill (1123 below), plus weeds more typical of grassland including buttercup (Ranunculus acris/repens/bulbosus ), dock, ribwort plantain and grasses.

Gully (context 1123) Context (1123) was the fill of gully [1124]. Larger charcoal fragments were associated with a few possible oat grains (cf. Avena ), a single barley grain, and several wheat glume bases showing the 04_BGAS125_131-246 16/11/07 15:50 Page 205

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 205 p t a W w t i H R n n W b P a S G G D D D # D D # H M C C # C H # # # # 4 1 / 7 8 8 5 . 7 0 1 1 7 3 0 0 . 1 1 1 – – – 1 – – – – – 4 – – 2 2 – 1 5 1 0 1 1 . 0 s 3 t / i 5 6 8 p 0 . 1 8 8 8 9 1 3 3 2 8 1 . 1 1 d 1 < – – – 2 1 6 9 1 2 1 3 3 1 2 9 2 5 0 1 1 n a 2 1 / G 4 3 2 5 6 . 5 5 7 0 3 . 0 0 & 0 – – – – – – – 0 – – – – – – – 2 5 1 0 1 1 B 6 , 2 / A 5 8 3 s . 3 9 9 1 e f 5 0 1 . 1 1 r 0 – 1 1 1 – – – 6 – – – – 3 3 – 5 2 5 3 1 1 u t 4 c 1 / u 3 2 r 6 5 . 8 3 3 t 2 3 0 1 . 0 0 S 0 – – – – – – – 0 – – – – – – – 1 1 5 0 1 1 1 8 2 e / s 8 9 7 5 a . 6 1 1 2 5 5 1 . 2 2 h 0 – – – – – – – 2 – – – 1 – – 1 2 1 5 0 1 1 P – s 6 9 7 t / 5 l 8 2 2 8 6 . 1 0 . 1 1 u s 0 – – – – – – – 0 – – – – – – – 5 1 5 0 1 1 e r l 2 9 8 a / 9 2 2 c 9 5 i . 0 . 1 1 0 – – – – – – – 0 – – – – – – – 3 5 2 0 1 1 n a t o b o 0 9 5 e 5 2 1 4 a 7 / 0 1 . 1 1 h 0 – – – – – – – 1 – – – 1 – – – 4 5 < 0 1 1 c r a : s s k e r ) r g o t t i y k ) r l l l w ( i e / o l e g t e m m y w t ( k u g a r e t e f l t d a a a e i r n a t e e z s a w i o p o p e h n l s t e B f r u o S e o / x e n f c S e e r f f t t t l W l r e e r e l l y l m a a a h . o o u t p s e e t p h a a e e c e e e t l 6 t s l t t - o n z m e e t a t a s h h e h r t m m 1 a z z o r a o o o o e u a e a a p a i i e T W C B H L G W S F O B T O W S S S S F C l b s a ) n . e . i E T s t . E a L p a A e A L p b a d E m s l E n e e . u C i m r C m L u A m m / e l E u b I s E A ) u a a l a i . e l m s A e r a s A e L a n d D a g L c n . t c l r r u a E ) ( l t E a f l s a o U O e e e a w l m i e o p C s s c p p s e p c a C ) u P s i b d C d H p s s s a u u i l s v A o / s l l A o n i d d d b a O u g p i N u u L m o p d e s m m m m b x a e s C c c s i l e p u / e r o I N r u u u u U f f n n u l d U s a a o e n a f i c c c c l f r n c i p i i i i m E u u e e d n n T n c i y t i a a N t t t t T e a i r e t r e e u n n e i i i i r r l u r l R h o h r p H h e A i t v r r o r r h E a a e s g n W R R r B R T C U C ( T C U T T C ( C ( A C G H A C 04_BGAS125_131-246 16/11/07 15:50 Page 206

206 TOBY CATCHPOLE o l w S d S o o c - G W W G D H D D d G w C G – G C G G B G G D D C C C C E C – – – – – – – – – – 2 – – – – – – – s g a r f 6 + 0 6 – 1 3 1 2 1 5 1 1 5 1 4 2 3 1 1 – 1 3 – – – – – – – – – – – – – 1 – – – – – – – – – – – 6 – – – – – – 3 – – 1 – – – – – – – – – – – – – – – 4 – – – – – – – – – – – – – – – – – – – – 1 – – – 1 – – – – – 1 – – – – – – – – 1 – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – – 1 – – – – – – l t t e r n h i g n d o g a k r i a n e t w i i c e r e e e l i n r t l h p b r r t d a a h w c e a l t l e c c e t m d i i y i a i s t h i x P T s t s n o p c e k a E M s i R r P t f t l a - S c / / r h r e r b V e e t y r w F t a o e r - o w g i l u e e P e s o v o V e t l l s k k y o l s q v / w s a t s r l l o c c e m r o a s l a r r l i o n e a b a a e r o l i l n i e e l a a e a a c m B S P M S C F H C P G R Y B C D K P L E E A A E . p E . . p . E p . A L . s C s L E . E L a . L L C i E e a L . A a s l o A L A a r L i . I o A i a p m g u C s m E L a a l m f E r t l s E l i a r L e u p a R i o L o s A u a A c u t e s s s u r h i l C c n C i e s n A E c s L v t p a i i n o h i E i i d l l N s t a n A A E e a e n p L A e i o I Y i n v r a u c m r e p E C v c m r E E c u A v r a a . i a N e E N n E r v p a U G t H a f s M a A A e l n r h s E A A r a b E p a e / V A p c p A t l o u C v / g O A P t a m E e / L I H m p c a s s E E a a a p y h s s C o r i i E m o s . A l e u c s t a s T P l G O a l R C f r t u u a U i g e C l C u h A i L n i r i n m C V c x r r i m a . r t I a Y a a c Y t o E A n l O N p c a e m A A y y u t u a M A g n i v o g L a n i A S o G a S B l i L R L n u l h h s f I A R l n I e B i ) ) ) l y l o a m i h i t t n l N . . . t l A r c a n e a G a A d A O U n p I i P r o L h R u O i a A a a o l e o C t L L L ( A R A A M D T ( C L L P L F V P F L S P R O G R V P M P ( R P P S C 04_BGAS125_131-246 16/11/07 15:50 Page 207

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 207 g 0 o n W - i 1 W w R M w G P R o H w G , l f P G M M D D D w G D G C G C – G M – G C o l s / t n a n g a t 1 – – 8 – – – – 5 – – – – s – . s s l e i 1 1 4 h o 6 0 2 1 0 c s t 4 1 8 1 7 1 1 – 5 6 – 2 7 i p d , m s a d d / n t o e 1 – – – – – – – – – 1 – – p w : : P w ; ; h s s l r i 5 a o 1 – 1 – – – 1 – – – – – – s m h : c i M r ; e t w 5 – 1 – – – – – – – – – – a o r h e p s g o d e h h p 1 – – – – – – – – – – – – : : p H e ; ; s m t d a o n t r i a l 4 – – – – – – 1 – – – – B b s a s e h a y r R n g / e : t p f 1 1 – – – – – – – – – – G o o ; : S r / o o h ; t o s l o i m o o / s h m t 2 1 – – – – – – S – – – h a c d e l i a i r h l a a : t n s - . E e s n ’ d g ; r t n d o d o l a i r a e e l t C a . C e i b t d e r / - n d w c o s s u e e : y ’ s n t o t a s n g i a a D m i w t r o ; u o r d M : C s g r h r l f / o s - : D i s s - B p u s W o w G u D d r e w s ; e o l - n e n m ; l o e s D i t i s t t b u e d e e l a r t s / s h g l r n c a h u t k r a c s a e n i r s d i r e c b e g a r G o r u a e i e e c p e c a l s s m r B T G F D M C B S S N R S / : a s : o l / u S n ; d e s p ; o c l e i s c i a t e B a d o m - s r v f . a i z o ) o t t e y . L h l l / r a r / t t s t s u u L / d i s s ( a n c u h m s r / t : t u a i E : t E c s u l p s l a n d s r i o E S t A t e A p p C u ; s r ) m e s l e d . i s A e : E r m E v d a r t d i e e u e L a s p c E E e p c r C l p C m p ( h n o d t p l p s s i a s s : i : p a c A p u C A e e a s A o i i p i r s s u s e h p r l m e v r s r n t l E R s A r t R l k s R i s i a m i a a a a u u a h s t ; u n t o u i n k c C t c C l E e e u r E s h s r x l i u P s l l a r a o c c o u P o e u c s e A / a P m e p v g b o t ) T i N b d a t r l r c n i i o . n c a s e a e o D Y a g n a y a r S : o l O r o e n u : U L n # c w A B P H P – F A N E u B C P ( C T i C A V C J J 04_BGAS125_131-246 16/11/07 15:50 Page 208

208 TOBY CATCHPOLE

characteristic veins of spelt wheat . A small assemblage of arable weeds included brome, cleavers, chickweed ( Cerastium ) and orache. The few grass caryopses plus buttercup and dock may relate to field edge species gathered with the cereal crop or possibly with hay.

Pits (contexts 1066, 1045 and 1217) Burnt material (1066) from the base of pit [1043] was sampled. A small assemblage of cereal remains included wheat grain and glume bases, with oat and brome also present. The seeds are mostly typical grassland weeds with spike-rush suggesting damper ground conditions. Context (1045) formed the burnt fill of pit [1038]. The float produced an abundance of charred cereal grains and, although c.30 per cent of these were so poorly preserved they were classified as ‘cereal indeterminate’, there were over 200 wheat, 50 barley and 20 oat grains. The majority of the wheat was identified as a hulled variety from the slender parallel-sided grains showing a rounded dorsal profile. The presence of well-preserved glume bases and spikelet forks of Triticum spelta confirm the presence of spelt wheat. These show the typical sharp keel with clear lengthwise striations on the glume face and, although this is not so clearly visible on many of the other examples found, these have been identified as chaff from hulled wheat. A small proportion of the wheat grains (<6%) were of a more rounded profile, typical of free-threshing wheat, so, although no accompanying chaff was recovered, it seems likely that two varieties were present. Barley was less abundant. Some of the grains were well enough preserved, with traces of the lemma and palaea visible, to confirm the presence of hulled barley; in some cases straight grains were also recorded, suggesting that both two and six row forms may have been present. A single rachis internode was the only barley chaff. There were also a few oat grains. Several examples of wheat/barley awns and a wheat glume beak were preserved in a silicified form. The bulk of the charred weed assemblage is brome ( Bromus racemosus /hordaceus /secalinus ) caryopses. Rye brome ( Bromus secalinus ) is a typical weed of cereals and marginal and waste ground; it may have grown amongst the crops with the other arable weeds such as corncockle ( Agrostemma githago ) and knotgrass ( Polygonum aviculare ). Smooth/soft brome ( Bromus racemosus/hordaceus ) is more typical of grassy places, as are the other grasses, meadow-grass/cat’s-tail and fescues. Clover/medick, vetch, ribwort plantain and dock are also typical of grassy habitats. There are also indications for damp ground species with spike-rush, sedge and lesser spearwort ( Ranunculus flammula ). The fill (1217) of truncated pit [1218] appeared to have been burnt in situ and has been interpreted as an oven. The small float (<1 ml) only produced fine charcoal fragments and several dock seeds.

Conclusions Unfortunately much of the evidence associated with the timber buildings is from small assemblages of charred plant remains from features such as fills of postholes, construction trenches and gullies. As such it represents secondary deposition of debris scattered around the buildings. Similarly samples from the enclosure ditch and fills associated with inhumations outside the enclosure produced only limited evidence and are likely to represent stray occurrences from activity around the site. There is, however, evidence for wheat and barley from cereal grains with chaff, including some well-preserved spelt wheat spikelet forks and glume bases. Spelt is a hardy cereal which thrives on heavy soils, is ideal for winter sowing and is the form most commonly recovered in the Roman period. The best evidence comes from two pit fills located inside the enclosure. The ashy fill (1186) of circular pit [1185] within Structure D included many charred weed seeds with the cereal 04_BGAS125_131-246 16/11/07 15:50 Page 209

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 209 # t a t i b a H # # # – # # # # # # # # # # # # # # 8 2 / 8 7 9 . 4 1 1 1 . 3 0 1 2 2 – – – – – – – – – 0 – – – 0 – – – – – – 4 2 9 5 1 1 . s e 0 r 2 u / t 3 6 5 8 a . 0 2 4 6 7 e 2 . 9 2 0 0 0 f – – – 1 1 – – 3 4 8 5 3 – 1 – – – 2 2 – 2 1 1 5 1 1 d n a s n 8 5 2 o 4 7 0 0 4 0 3 4 i 4 1 t 0 4 5 1 3 8 0 4 2 0 9 4 . / 1 0 0 0 a 2 1 < – – 3 3 1 1 5 2 4 1 4 3 6 8 2 2 1 0 4 < 5 1 1 m u 8 4 3 h 2 2 / 5 2 2 n 1 i . 8 0 1 1 1 – – – – – – – 6 – 3 3 4 3 – – – 1 – – – 2 2 9 5 1 1 2 e s 6 a 2 h / 0 4 7 P . 0 3 2 8 0 5 . 3 2 0 0 0 – – – – – 1 – – 4 3 1 4 1 2 – – – 5 3 – – 3 2 1 5 1 1 s t l 7 . u 5 s 5 e / r 8 6 2 8 l . 2 6 6 6 a 2 0 2 . 3 0 1 1 1 c i 2 1 – – – – 3 – – 1 2 5 7 – – – – 5 – – 1 2 5 5 1 1 n a t 4 t o 1 a b / e 4 2 5 4 o h . 7 9 9 e 6 . 4 5 1 1 1 a – – – W – – 1 – – 1 1 1 – – – – – – – – 0 1 2 5 1 1 h c r a : t s a k e r h o w W e g t t t y y y g a a a n e a t e e i e e e l l l t a h r h h h r r w a e s ) a a a e e s e h t S B W W B B W r h k e / x e . a h r t t : : l W w d d d d d e t l l 7 e y y y a a u e e e e e t p - t t a a e e e 1 b l l l l l e e t l l l l l e l l l l l t t n t t a h e e h r r r m e e e u u u u u a a o o o l a a a e r a p p b m H B W O T S T H O S B H B H H F W S F C s a u n l i T g ) a p o s d p ) m y s e e ) r d i e r g f ) t e r b i i l m k ) e l l ) c l l ( u m i / t d ) u a ) l e e s e g i p m h o ) ) e ( a a p h d n ( s s n k t t u ( g s r n r k k ) ) ( l l t t i w t r o r r d e e e e e w e e e a i p p p p a a e m p p p p p s s p p o o c a m z r s s s s s ( t d d H s s s s s s s f f a a i o u t u / e l s n n n e s d c d b b t t p p f r i i i i / m m m m d e s s m m m m m m m m e e e t e f f i l l r e e l l d s u u u u i l r u u u u u u u u f f i o a a e e e e a a n o o e e r e i c c c c c c c c f r p i l i i i i i i i i m m e e d d d d h n n c k k l a a t t t t t t t t H T a e e i r i r i r r r r e e c u u i i i i i i i i m l u r l l z z o h o o o h e e . . a i v p p v r r r r r r r r i i a f f r s s s g g h H ( T ( T A C T ( T ( C T ( A C ( T H ( c T H T G H c S C S S 04_BGAS125_131-246 16/11/07 15:50 Page 210

210 TOBY CATCHPOLE – p W w H R # # n o o P - G G G G D D d D D D D – G C G D C D C C C C C C – D M D – – – – 2 – – – – – – – – – – – – – s g a r f 2 – – – 3 – – – – – – – – 1 6 – – – s g a r 2 8 f – 2 1 1 1 – 2 2 – 1 – – – 1 – – 1 – 1 – – – 6 1 – 1 – – 1 1 – – – 2 – – 3 – – – 2 1 – – 6 – – – – 1 – – 2 – 1 – – – – – – 2 1 – – – – – – – – – h c t e l V e / z a a e P – – – 1 – – – – – – – – – H – – – e l o t y n r l e C i i l o / a t k t t e m c w i e n p a r f d a a a e l N l e d l t e i R s h P e p k k o p / s n o c M e c t u f t a n o d S o / r f r c o e e e r a w r r r e c o a g u m h h e k V s e t t e k s ) n c q c v / w t s s l c o s o i r m a t d s l a o n b u i o o r g l i n h o o i e e u e a C D R D M C P R K C G C O B L C r f E p . E t p A E t . p . E s . L u A e E p . A L L n A L a s d E E ( l m L C E e o s E n o . A u u r C i g a n A s a p g E C p a C s i e L t a E A l m s a e l p o L E A A a / I h c P l E A s o u a l a s s C i . t / s m c L e i i A u i E o e A i d L n N D a a a c u A r E L e v p g c e i i l I Y v s a E c c p a C E E f a a c c l a p E U o A O a a a i i l i N a n E r p s n G C s s A H c b M A e m C s A i P a l E g i p d C p / V V A l u u m v O A P A o m s E / / C l l S u A p o a E m i a a C s s O I / i E m o N s b l u u u L m m p T d G O u l C x r u u s C a c c s g C S u A e i i n / u e c C o I N x r r i a U t t t I a Y n n Y l u d o U s A a i l S N s c e A s y y h l t n c s p A n # y i o g E u u e n t o B S T i s L y a R W i h h N s A f T A e n I e B y a e e t m r n n i t t e r r a r t l r S a G r R r p h A H U O A n g t r P e o R E L u O a a r a A a l o o e e D T A A P P R B B R P L P R F L P P G C C C A A C H C U R B U R r R W 04_BGAS125_131-246 16/11/07 15:50 Page 211

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 211 W W w R o P R H S w P G M M – H . s s G G D G G M G C l e i h o c s t i p d , m s a d d – – 2 – – – – – – – / n t o e p w : : P w ; ; h 1 s s l 6 – 2 – 3 5 1 2 – – r i a o s m h : c i M r 1 ; 7 2 0 2 e t w 5 1 1 3 – 8 1 3 – – a o r h e p s g o d e 2 h h 4 – 2 – – 2 1 – – 1 p : : p H ; ; s t d a n t i a 5 6 l b 1 – 5 – – 5 2 – 1 6 s . a s d h a r n n g a l e : d p G o o ; o 0 : r o 2 – 1 – – 4 – – – – w o ; : o s l i m W o / ; s h b t h u a c r e i c 1 3 – – 1 – – – – r h s : : n e l i E e S a m g ; ; t s o o d - r r s e ’ m t . B i b t a l e r i a n e c e a u r : y t C n t t / - s n a s i R s s t / ; ’ / s r d s s t g a r l f o : i r e o o p o g h v D S s i s - D m ; / r s i u t w e r h d : r l h : t e o e c - e l i b e R g t v o u e d a i a s h g ; a l c a t o k m r s e r i s e d e : a o r l o u e l e e p m / t n ; T M F C S S S R C d a s o l e c w i t e a o g s v f . i n o t i y L l / r / t . w s s u / s d i s n c u o u L r : l t u a E : t f n l p s a e d s i o E t A t l p p C n u ; s i w s l e a i s A : E m D c r d a r o d e u e e c l E a E e p c C s C p m n s o d t p s / i / a s s : i p A / p u C A e s a a i i p r s s t u s e p r e v s r u t l E G A r R k s i s n a m e a a a a h s t u m a c u n c C t c C l E e u D s h x l i u P l a u a n / c c o u e u c i A / a P m g b d t o l N b a g r c n i o n c a s r e a G Y a a y a : o l O r o e n o u : t U P s H B c # P F C D E u B P h C C J J G 04_BGAS125_131-246 16/11/07 15:50 Page 212

212 TOBY CATCHPOLE

component restricted to a few wheat and barley grains and accompanying chaff. The arable weed component includes species typical of disturbed ground habitats which are likely to have grown with the cereal crops. Cleavers and black bindweed are both twining species which could easily have become attached to cereal stems and cut with them at harvest. Other species such as fat-hen and orache have no preferences for soil type and will thrive wherever soil disturbance occurs. Phase 2 pit [1038] contained a burnt fill (1045) with a rich assemblage of charred cereal grains, accompanying chaff and weed seeds. Grain morphology suggests that both hulled and free- threshing varieties of wheat were present and well-preserved glume bases and spikelet forks confirm the presence of spelt wheat. There is also evidence for both barley and oats although these were less abundant than the wheat. The oats may represent an additional crop but no floret bases were preserved to confirm whether they were cultivated or wild forms, the latter frequently present as crop weeds. The most common weeds are brome, which grows as a crop weed but is also considered to have been utilised as an additional grain to help bulk out crops. Other common arable weeds which would have been gathered with the crops at harvest include corncockle and knotgrass. Many of the charred weeds from both the above contexts are more commonly thought of today as grassland species. They include dock, vetch and clover/medick as well as grasses brome, fescue, meadow-grass/cat’s-tail, many of these typical of grassy places, rough and cultivated ground. Ribwort plantain, meadow-grass/cat’s-tail and bents ( Agrostis ) are typical of a range of grassland habitats from damp meadows and pastures to rough ground while crested dog’s-tail ( Cynosurus cristatus ) is found in meadows and pastures on a wide range of soils (Fitter, Fitter and Farrer 1987). There are also indications in both samples for areas of damp ground with the presence of rush in (1186), sedges and spike-rush, the latter a plant of damp places frequently found in association with charred cereal assemblages. The common occurrence of these damp ground elements suggests that parts of fields used for crop cultivation in the past may have been poorly drained or that use was made of marginal land (Jones 1978). Some species may have invaded from adjacent unploughed fields or have continued to grow as residual species once the fields were cultivated. In both samples there were a few examples of silicified wheat/barley awns. Silicification is a process shown to occur in high-temperature oxidising conditions, typical of a bonfire burnt down to a heap of glowing charcoal. All the carbon is burnt, leaving only the silica skeleton of remains such as cereal chaff. Silicified remains have been recovered from other sites from contexts such as ovens or kiln floors and corn drier flues (Robinson and Straker 1991) and it may be possible that this material originated from a similar feature. Although there is only fairly limited evidence for cereal crop cultivation, spelt wheat and barley are shown to have occurred in several features. Oats and brome may have occurred as weeds or have been used as a valuable food source to help bulk out the crops. Much of the evidence comes from the abundance of weed seeds, many of them typical of disturbed ground or grassland habitats today and frequently recovered in association with charred cereal remains in archaeological contexts. This suggests that they were growing with the cereal crops, on field margins, or adjacent to boundary ditches which may have been wet and were harvested with the cereals. The presence of the charred remains in secondary contexts makes interpretation of crop- processing activities difficult, but small weed seeds, items of chaff and silicified awns are likely to have originated from cereal processing on the site, for example from cleanings from ovens or kilns or from the remains of fine sievings used in hearths as tinder. It seems likely that crop production was local with fields of wheat and barley cultivated by the inhabitants of the settlement at Dymock. 04_BGAS125_131-246 16/11/07 15:50 Page 213

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 213

Charcoal by Rowena Gale Charcoal was recorded in most of the environmental samples but usually in insufficient quantities to warrant examination. This report presents the analysis of comparatively charcoal-rich samples from three Roman pits, [1185], [1038], and [1043], two of which contained metal-working debris. The study was undertaken to provide environmental data and to assess the economic use of woodland resources for industrial purposes.

Methodology The charcoal was fairly well preserved and reasonably abundant. That from [1185] and [1043] consisted almost entirely of narrow roundwood. The samples were prepared using standard methods (Gale and Cutler 2000). Anatomical structures were examined using incident light on a Nikon Labophot-2 compound microscope at magnifications up to × 400 and matched to prepared reference slides of modern wood. When possible, the maturity of the wood was assessed (i.e., heartwood/sapwood) and stem diameters recorded. It should be noted that during the charring process wood may be reduced in volume by up to 40 per cent.

Results The taxa and number of fragments identified are presented in Table 18. Classification follows that of Flora Europaea (Tutin, Heywood et al. 1964–80). Group names are given when anatomical differences between related genera are too slight to allow secure identification to genus level. These include members of the Pomoideae ( Crataegus, Malus, Pyrus and Sorbus ) and Salicaceae ( Salix and Populus ). When a genus is represented by a single species in the British flora it is named as the most likely origin of the wood, given the provenance and period, but it should be noted that it is rarely possible to name individual species from wood features and that exotic species of trees and shrubs were introduced to Britain from an early period (Godwin 1956; Mitchell 1974). The anatomical structure of the charcoal was consistent with the following taxa or groups of taxa:

Aceraceae. Acer campestre L., field maple Corylaceae. Corylus avellana L., hazel Fagaceae. Quercus sp., oak Rosaceae. Subfamilies: Pomoideae, which includes Crataegus sp., hawthorn; Malus sp., apple; Pyrus sp., pear; Sorbus spp., rowan, service tree and whitebeam. These taxa are anatomically similar; one or more taxa may be represented in the charcoal.

Table 18. Sewage works: charcoal from Roman contexts.

Context Sample Description Acer Corylus Pomoideae Prunus Quercus Salicaceae Ulmus 1045 500 Fill of pit 1038 ––2142h, 32s –– 1066 502 Fill of pit 1043 16r 25r 21r 4r 3r 4r 1r 1186 511 Fill of pit 1185 2r 8r 1r 29r 1h, 3s 2r –

Key: h = heartwood; r = roundwood (diameter <20 mm); s = sapwood (diameter unknown) 04_BGAS125_131-246 16/11/07 15:50 Page 214

214 TOBY CATCHPOLE

Prunoideae, which includes Prunus sp., possibly both P. spinosa L., blackthorn and P. avium L., cherry or gean Salicaceae, which includes Salix sp., willow, and Populus sp., poplar. In most respects these taxa are anatomically similar Ulmaceae. Ulmus sp., elm.

Pit 1185 (context 1186) Charcoal was recovered from the fill (1186) of pit [1185] within the area of Structure D. The function of the pit is unknown but since it also contained cereal grains and bone it seems likely to have been used as a repository for domestic waste. The charcoal consisted of burnt brushwood ranging in diameter from <1 mm–4 mm, mainly from blackthorn ( Prunus spinosa ) but also including hazel ( Corylus avellana ), field maple ( Acer campestre ), oak ( Quercus sp.), willow ( Salix sp.) or poplar (Populus sp.) and the hawthorn/ Sorbus group (Pomoideae). There was no evidence to indicate that it was used in conjunction with largewood fuel, e.g. as kindling. Without the addition of a more substantial fuel, these twiggy bits and pieces would have provided a very short-lived (although probably fairly hot) fire. It is possible that the sample represents the disposal of twiggy brushwood trimmed from roundwood or wider stems or, perhaps, the sweepings from fallen tree debris.

Pit 1038 (context 1045) Charcoal was recovered from the fill of [1038]. The pit also contained burnt clay and a mould fragment associated with metal working. The charcoal was mostly very fragmented. The sample indicated the predominant use of oak ( Quercus sp.) largewood but also included the hawthorn/ Sorbus group (Pomoideae) and Prunus (probably cherry, P. avium ).

Pit 1043 (context 1066) The gravel extraction pit (1066) [1043] contained a large amount of burnt roundwood. Pieces of mould and slag, including tap slag, were also present. The charcoal included hazel, field maple, blackthorn, oak, willow or poplar, the hawthorn/ Sorbus group and elm ( Ulmus sp.). Much of the charcoal was fragmented and, although it was not possible to assess whether it derived from coppice stems, four intact segments of roundwood provided the following (charred) dimensions:

willow/ poplar – diameter 12 mm, 6 growth rings hazel – diameter 12 mm, 4 growth rings oak – diameter 30 mm, 5 growth rings field maple – diameter 15 mm, 13 growth rings.

Environmental Evidence The species identified included field maple, oak, elm, blackthorn, the hawthorn/ Sorbus group, hazel, willow or poplar and, possibly, cherry. This list is probably biased towards the economic use of wood, particularly for firewood and charcoal making, and is therefore unlikely to represent the entire range of woodland species growing in the vicinity. It does, however, indicate the presence of mixed deciduous woodland. While there was insufficient evidence from the samples available to indicate the use of coppice stems, the frequency of narrow roundwood in context (1066), probably debris from metal working, could imply managed woodland as a source of fuel for industrial activities. 04_BGAS125_131-246 16/11/07 15:50 Page 215

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 215

Industrial Fuel Iron-smelting slag and mould fragments in contexts (1045) and (1066), the fills of pits [1038] and [1043], testified to the production of small tools and brooches. Associated charcoal probably represents fuel debris. The reducing atmosphere required for iron smelting can only be obtained through the use of charcoal fuel (Horne 1982). Interestingly, the charcoal from (1045) indicated the extensive use of mature oak obtained from largewood, whereas that from (1066) demonstrated the preferred use of narrow roundwood from a wide range of taxa. These differences may reflect fuel selection for specific processes but could relate to the availability or supply of fuel at any given period. Although the use of oak heartwood is frequently associated with Roman metal working in many parts of Britain (Gale 2003a), evidence of iron-working fuel composed almost exclusively of narrow roundwood was recorded at five other Roman sites on the fringes of the Forest of Dean: at Woolaston (Figueral 1992), Blakeney (Gale 2000) and thrice at Ariconium (Leyall 1923; Gale 2003b). Some aspects of the iron-working practices at Dymock therefore appear to parallel those of places elsewhere in the environs of the Forest of Dean.

Conclusions Environmental evidence from charcoal deposits in three pits associated with the Roman settlement identified deciduous woodland, composed of field maple, oak, elm, blackthorn, the hawthorn/ Sorbus group, hazel, willow or poplar and, possibly, cherry. These woodlands supplied fuel for the local metal-working industries. It is suggested that differences in the character of metal-working fuel associated with pits (1045) [1038] and (1066) [1043], mature oak wood in the former as opposed to narrow roundwood in the latter, may be related either to specific processes or to fuel supply. The use of narrow roundwood for metal working at Dymock correlates with similar evidence from other Roman sites in the Forest of Dean.

DISCUSSION

Summary of Evidence During the later 1st century AD a series of timber structures was constructed within a ditched and gated enclosure on a previously unoccupied site. The structures were made of wattle and daub. Nails of a type generally used for timber cladding were also retrieved (Cool, above). The backfilled construction trenches and postholes contained quantities of daub, much of it burnt, some pieces having impressions left by small timbers. Clear evidence for wattling stakes was recovered from the south-east corner of Structure A. The lack of evidence for ceramic or stone roofing materials suggests that these were of thatch or timber. That Structures A and D were roofed is indicated by drip gullies. Little evidence survived to indicate for what the buildings were used. If Structure A had a timber floor as suggested by the presence of steps (Structure C) at its down slope end it seems highly unlikely that it was used for metal working. Instead it may have been used as living accommodation or storage. The apparently open fronted spaces on the north side of Structure B and the small posthole structures, such as E, may be better candidates for metal-working locations but direct evidence for this was lacking and they may have been for stabling or storage. Structure G was recorded as a possible sunken featured building (SFB). Although this seems unlikely at this date, there are a large number of more complex examples, of 2nd-century date, excavated at Monkton, Kent (Hicks et al. forthcoming). On the evidence available however, it is 04_BGAS125_131-246 16/11/07 15:50 Page 216

216 TOBY CATCHPOLE

difficult to support an interpretation that Structure G was an SFB with conviction. The environmental evidence is entirely in keeping with a normal rural, and thus agricultural, settlement of the date. It indicates a mixed farming economy involving the growing of wheat, barley and possibly oats and brome. Significant areas of grassland and meadow (Jones, above) and of mixed deciduous woodland (Gale, above) were also exploited. The small assemblage of animal bone suggests cattle and sheep (or goats) were bred for food and that the latter were also kept for their wool. Pigs were less important. It is unclear if they were kept or carcasses brought in. In common with most sites of the period, the amount of beef eaten increased through time. Some hunting of wild animals took place (Ingrem, above). Agricultural tools were recovered in the form of a goad and reaping hook (Cool, above) and a high proportion of Severn Valley ware storage jars was present (Timby, above). A major significance of the site lies in the evidence for smelting of iron and casting of copper- alloy objects. This industrial activity was presumably prompted by wider changes in society and economy in the early Roman period and may help explain differences between Dymock and other rural sites of the same date. There are suggestions from the pottery that local industry was not restricted to the manufacture of metals; hand-made Severn Valley ware is ‘exceptionally well represented with several slight seconds’ and there is some evidence to suggest that greywares were being produced locally to satisfy a demand for Roman pottery styles (Timby, above). The nature of household items hints at an establishment of some, if fairly modest, pretensions but the brooches lost on the site indicate styles of dress that were those of the local indigenous population (Cool, above). A possible source of the high-status objects recovered is exchange for the industrial products already discussed (but see below). The coin evidence suggests that monetary exchange became important at Dymock earlier than at most rural sites in the area (Guest, above), perhaps indicating interaction with the Roman military (discussed below). Claudian copies are generally associated with military pay and one unstratified example was retrieved (Guest, above). The enclosure and timber structures were relatively short lived, being removed and backfilled by the early 2nd century. Continuing activity of a less well-defined nature in the mid 2nd century is indicated by inhumation burials, isolated pits and a series of ditches difficult to group into recognisable enclosures. There is evidence for iron smelting in the period AD 120–150 but none to suggest that industry continued beyond 150. The adult burials indicate that the area to the east of the Phase 1 enclosure was considered outside the settlement area by the mid 2nd century. They demonstrate a rite of crouched or stretched inhumation burial, common in rural Gloucestershire at the time and thought to be a continuation of late Iron-Age tradition. They can be compared to those from the larger Hucclecote cemetery (Thomas et. al . 2003). Whether the Dymock burials can be described as being within a cemetery or simply scattered burials near a settlement is difficult to say without further excavation. At Dymock, admittedly from a small sample, there is a similar numerical dominance of female over male as at Hucclecote, 4 being female and 1 male. The only burial (1190) that included anything other than a very simple coffin was female. The implication that a more substantial coffin was required so that the body could be crouched within it (Fig. 11: 1190) indicates the continuing importance of the rite. Proximity and alignment of the graves suggests that the females were buried in pairs. Hobnails found only in the graves of two adjacent females may support the slightly later date of those graves, [1168] and [1194], suggested by the presence of BB1 sherds (Timby, above). All the Dymock burials had their heads on the northward side of their graves. There was a smaller proportion of crouched burials at Dymock than Hucclecote; two of the females were crouched, the male and one female were on their backs and the remaining female was on her side. The putative second phase of enclosure is uncertain in its nature and extent. It must have been very short lived, as the burials cutting the backfilled ditches could not be separated from it in date. 04_BGAS125_131-246 16/11/07 15:50 Page 217

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 217

Evidence for activity after AD 150 was scant, comprising an entirely robbed building and an isolated pit.

Dating Activity on the site had clearly commenced by the early Flavian period (AD 69 onwards: Wild, above). An unstratified rosette brooch (Cool, above: catalogue no. 1) provides the only and uncertain indication of activity in the first half of the 1st century. The pre-Flavian samian and coins recovered during the excavation may indicate activity of that date but the items may not have been new when brought to the site. Unfortunately in this regard much of the earliest material was unstratified and few contexts could be regarded as relating to the construction or use phases of the timber structures and enclosure; the excavated features mainly provided information regarding the destruction of both. The base of the southern enclosure ditch contained a small amount of silt which only produced loosely datable Severn Valley and oxidised sandy wares. Other than that, the great majority of the fills of Phase 1 features were interpreted as the result of a single episode of deliberate destruction and backfilling. Therefore most finds retrieved from these features must, in any attempt to closely date the phase, be seen as residual. A sherd of Les Martres-de-Veyre samian from the backfill (1206) of a Structure B construction slot is of a product thought to have only been imported after AD 100, although it was a copy of an earlier form (Wild, above). This provides a t.p.q. date for the removal of the Phase 1 enclosure and timber structures. There was no sign that the main enclosure ditches had been cleaned out or re-cut, suggesting a short-lived phase of activity. There appears to have been a hiatus from c. AD 100 to 120, although, given the short time involved, that is difficult to prove. A number of mid 2nd-century features included much residual material, dating to the 1st century. Where undatable material is present, such as iron slag, it is impossible to say what was contemporary and what was residual. That metal working may have continued is suggested by the form of pits dated to Phase 2 but conclusive evidence was lacking.

Evidence for Military or ‘Official’ Involvement The layout of the enclosure, the form and construction methods of the structures and the unusually ‘Romanised’ material culture at Dymock at such an early date all require further examination. The high level of decorated samian pieces suggest military contact (Wild, above), a local greyware industry producing non-local forms has been identified and sherds of a fabric greatly resembling that from the Kingsholm military kilns were recovered (Timby, above). Objects such as high-status glassware, plated spoon handles and door keys are also unusual in a rural context of this date. The large number of iron nails indicate that ‘the inhabitants were clearly using Roman building methods earlier than some of their neighbours’ (Cool, above), and the assemblage of 1st-century coins resembles that typically found on military and urban sites in the region (Guest, above). The short- lived nature of the Phase 1 enclosure ditches and timber buildings and their apparent deliberate destruction and backfilling are also suggestive of the military (Timby 2001, 79). Despite this, there is insufficient evidence from the excavation to support an assertion that a military unit was based on the site. Arguing against the presence of a military garrison are the facts that the bulk of the pottery (Timby, above) and the diet of those living at Dymock (Ingrem, above) are typical of rural ‘native’ sites of the period. In addition, not one piece of military equipment has been identified, while the brooches that were being made and/or lost on the site were of civilian not military type. Recent critical re-examinations of the evidence for military origins at many Romano-British ‘small towns’ have found it to be ambiguous at best. It is hard to improve on Jackson’s (2003, 151) recent examination of the evidence for a formal military presence directly 04_BGAS125_131-246 16/11/07 15:50 Page 218

218 TOBY CATCHPOLE

controlling the iron industry at Ariconium , suffice it to say that so far there is a similar lack of convincing evidence for this at Dymock. Even the smallest Roman fortlet was defended by rampart(s) as well as by at least one ditch. The lack of space for a bank and gate posts located between the ditch terminals of the Dymock enclosure provides a clear indication that there was no rampart here. There is evidence for a fence line near the gate but none elsewhere despite the fact that ephemeral features such as drip gullies have survived near the enclosure ditches. It therefore seems that we can discount a military defensive enclosure. Any proposal that the enclosure might be a fort annexe would require evidence for a fort. Whilst Structure A was similar in size to barracks at smaller Roman military establishments (see Wacher 1998, figs. 13, 17 and 21), it was not a single stand-alone structure; it is very likely that Structures A and B were adjacent and stood at right angles to each other. The part of Structure A resembling officer accommodation was also located furthest from the enclosure ditch, at the wrong end for a barrack block. Black’s (1995) survey of the empire-wide background and the British archaeological evidence for the Cursus Publicus (the official transport system) and its network of mansiones and other facilities provides an alternative context in which to consider the Dymock enclosure (M. Corney pers. comm.). Black lists a number of features that are to be expected at a mansio . These include barrack- like standard accommodation; grander first-class accommodation, often with entrance halls and heated rooms; separate staff accommodation and stabling; and large granaries used as fodder stores. Buildings were often grouped around one or more yards. Separate wagon parks might also be present. Many possible mansiones were located in a position well away from the main road but that would have not been possible at Dymock due to the proximity of the river Leadon. Not all elements were necessarily present at every site, and in some cases elements were dispersed around a deliberately founded roadside settlement. Accommodation might be provided in the homes of local residents or in local tabernae , and facilities intended mainly for slower moving pedestrian and wagon traffic might not include all the higher-status elements. An attempt to find similarities between the enclosure and buildings at Dymock and other possible mansiones has highlighted the lack of uniformity in those sites rather than furnished direct parallels. It seems that they were sensibly designed to meet specific needs rather than to a general blueprint. For example at Keays Lane, Carlisle (McCarthy et. al. 1982, 79–82), similarities to Dymock included date, the size of the enclosure ditch, the lack of a rampart and the presence of a barrack- like timber structure with a boarded floor (although the Carlisle building was larger, measuring 31 × 10 m), and the final deliberate dismantling of the compound. Brandon Camp, near Leintwardine, Herefordshire, thought by Frere (1987) to be a campaign base, was re-interpreted as a mansio of Neronian date by Black (1995, 26–7). There a large number of timber-framed structures of differing designs and sizes may have been tabernae but there were also a large granary and a discrete group of timber buildings, possibly comprising first- and standard-class accommodation with separate stores and staff accommodation. Certain elements of the site at Dymock fit the general pattern of early mansiones described by Black. They include the roadside setting; an enclosure lacking a rampart; ‘Romanised’ rectangular timber structures, especially barrack-like structures, built using ground beams; the arrangement of Structures A and B at right angles, perhaps suggestive of a further west wing around a courtyard; and finally the high-quality pottery and glass. There was, however, little clear evidence for the purpose of the excavated structures. If Structure B was stabling, as suggested by its form, the attached Structure A is more likely to have been used for storage or staff accommodation than as accommodation for travellers. The presence of a small early mansio at Dymock is far from completely proven due to the absence of first-class accommodation, granaries or bath house. More complete excavation of the enclosure 04_BGAS125_131-246 16/11/07 15:50 Page 219

EXCAVATIONS AT THE SEWAGE TREATMENT WORKS, DYMOCK 1995 219

may have allowed a more certain interpretation but, even so, the presence of a bath house or heated rooms should have been evident in the form of dispersed ceramic building materials, which were notably lacking. Occasional flue tiles have been reported as stray finds from Dymock but as these are not dated they cannot add to this discussion. The high-status finds from this site might be explained as the result of exchange for the industrial and agricultural products of the area; early coins have certainly been found elsewhere at Dymock but, as Timby has noted, neither of the other two recently excavated contemporary pottery assemblages from Dymock contained the early samian or other unusual imports which so far are limited to the sewage works site and chance finds at the nearby graveyard. Domestic timber strip buildings are not unknown in 1st-century Britain (M. Corney pers. comm.; Perring 1987), and there is no reason to assume that the indigenous population lacked the joinery skills required for their construction (Millett 1990, 69–72) but the fact that they have been recorded in settlements such as Verulamium and London hardly makes them any less unusual when recorded in north-western Gloucestershire. In conclusion, on current evidence, given the parallels with contemporary sites thought to be mansiones , we can tentatively interpret the enclosure and structures at Dymock as a short-lived establishment of the Cursus Publicus although their exact function within that system remains obscure. The results of the excavation presented here are of an incomplete sample of one early enclosure. They point to the desirability of future re-assessment of this discussion once further information becomes available.

Acknowledgements The evaluation and excavation were commissioned by Severn Trent Engineering Ltd. Thanks are owed to Robin Phillips and Len Swift, both of Severn Trent, for help during the excavation. Advice during the setting up of the post-excavation analysis was received from Iain Ferris, archaeological advisor to Severn Trent. Grateful thanks are due to all the specialist contributors to this report. Without their contributions there would be no detailed understanding of the site. Mark Corney kindly agreed to act as internal academic referee and provided extremely helpful comments and suggestions, not the least of which was to investigate the comparative evidence for early mansiones . Jan Wills and the editor provided most helpful comments on the draft text. Thanks are also due to Jon Hoyle, Charles Parry, Jo Vallender and other colleagues at the Gloucestershire County Council Archaeology Service for their advice, and to Robin Jackson, Pete Clark and Andrew Simmonds who kindly provided information from their draft reports. Any errors or misunderstandings remain the authors. Brenda Dickinson confirmed the identification of the samian stamps. Phil Parkes at Cardiff Conservation conserved several of the brooches and coins allowing full identification; Vanessa Fell at Oxford Institute of Archaeology X-rayed the metal finds. The excavation was directed by Sarah Reilly assisted by David Adams, Helen Bailey, John Percival, Mike Sims and John Smith. All the site team deserve particular mention for their skills in identifying almost invisible features and for their perseverance in horrible weather conditions. Advice on sampling strategy was received from Vanessa Straker, English Heritage Regional Science Advisor. The evaluation work was undertaken by Jon Hoyle, Lisa Emmanuel and John Smith.

Archive The finds and archive will be deposited at Dean Heritage Centre, Soudley, under accession number SOYDH 2005.8. 04_BGAS125_131-246 16/11/07 15:50 Page 220

220 ANDREW SIMMONDS

Excavations at land adjacent to the Rectory, Dymock, Gloucestershire, 2002

By ANDREW SIMMONDS

With contributions by Leigh Allen, Kayt Brown, Dana Challinor, Denise Druce, Emma-Jayne Evans, Lynne Keys and Jane Timby

INTRODUCTION

During October 2002 Oxford Archaeology (OA) carried out an excavation on land adjacent to the Rectory at Dymock, at OS Nat. Grid. SQ 7000931234 ( Fig. 2 ). This work was commissioned by Bruton Knowles Property Assets Consultancy on behalf of the diocese of Gloucester in advance of construction of a new cottage. The excavation was secured by means of the attachment of a condition by Forest of Dean District Council to the planning permission for the development.

Geology and Topography The site is located west of St Mary’s church, to the rear of the rectory house ( Fig . 3). It is approximately 50 m above OD and slopes gently westward toward the house, the area around which appears to have been artificially levelled, most likely during the house’s construction in 1953 (Verey and Brooks 2002). The house does not occupy the site of the old parsonage, which stood between the churchyard and the village street on land that now forms part of the village green. The underlying geology is Devonian mudstone, sandstone and associated drift over Lower Old Red Sandstone.

Summary of the Evaluation Results In September 2001 OA carried out an archaeological evaluation comprising four 10 m-long trenches (OAU 2001). This revealed a large ditch aligned north–south and containing Romano- British pottery. Also revealed was a gravel surface at the eastern end of the development area interpreted as part of one of the two Roman roads which pass through the village. There were also a gully and a small pit, neither of which produced datable artefacts.

Excavation Methodology

The main excavation area was a rectangle 17.0 × 9.5 m, forming the footprint of the proposed cottage ( Fig . 3). Additionally, two small areas to be affected by the digging of a pair of soakaways were excavated. The larger soakaway, located to the east of the main area, was 3.5 m square, while the smaller one, to the west, was 2.5 m square. The overburden was removed using a mechanical excavator down to the first significant archaeological horizon. 04_BGAS125_131-246 16/11/07 15:50 Page 221

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 221

All archaeological features were cleaned, excavated and recorded by hand in accordance with standard OA practices (Wilkinson 1992). A watching brief undertaken later during the excavation of drainage and service trenches for the proposed building revealed no further remains as the works in question did not extend deep enough to impact on archaeological horizons.

THE EXCAVATIONS ( Fig. 20 )

General The modern overburden consisted of a layer of loamy topsoil (1) 0.4 m thick and a reddish brown silty clay subsoil (2) up to 0.35 m thick. This overlay a layer of brown silty clay (3) between 0.05 and 0.25 m thick which extended across the entire site. This layer contained animal bone, slag and mid–late 2nd-century Romano-British pottery. The relatively intact and unabraded condition of the pottery indicated that it had not experienced a significant amount of disturbance or redeposition and may thus be in situ . A similar deposit exposed in an excavation by Marches Archaeology at Rose Cottage and Winserdine a short distance to the north-west of the Rectory site was interpreted as a Roman ground surface (Tavener 2001); a similar interpretation would seem appropriate here. No features could be identified cutting into this layer, so it was removed by machine to expose features beneath it. The lower boundary of this layer was poorly defined and this resulted in some over-machining.

Phase 1 The earliest evidence for activity on the site comprised a probable fence-line extending across the area of the excavation from north-east to south-west. Five postholes survived from this alignment (28, 37, 47, 50 and 68: Fig . 20). The best preserved was posthole 47, an oval feature 0.64 m wide which survived to a depth of 0.24 m. Posthole 50 was also oval, measuring 0.47 × 0.33 m, with a depth of 0.15 m ( Fig . 21). Its lower fill (52), a reddish brown silty clay, contained pieces of sandstone, possibly packing material and was overlain by a greyer, more charcoal-rich backfill (51). This layer also produced 2nd-century pottery. Posthole 28 was an elongated oval in shape, suggesting that it may have been subject to some disturbance, possibly during the removal of the post. It measured 1.0 m long by 0.5 m wide and had steeply sloping sides and an irregular base, with a depth of 0.28 m. A sterile primary fill (30) 80 mm thick was overlain by a main fill (29) which yielded 1st-century pottery, animal bone and some slag. The other surviving elements of the alignment, postholes 37 and 68, were less well preserved, having been truncated until only their bases remained ( Fig . 20). This truncation suggests that other elements may have been lost completely, and that the alignment may originally have extended much further than the length of the surviving part. The pottery recovered from postholes 50 and 63 indicated a 2nd-century date for this structure, with some apparently residual pottery of 1st-century date being present in posthole 28. A gully (63) ran at a right angle to the fence-line on its north-west side ( Figs. 20 and 21 ). The gully was exposed for a length of 2 m, continuing to the north-west beyond the limits of the excavation. It was 0.24 m deep with a ‘V’-shaped profile and had been back-filled with a deposit containing a large quantity of bone and some hand-made pottery, including from context 64 a dish with internal and external burnished lattice decoration in a Malvernian rock-tempered ware of pre or early Roman date. The location and alignment of the gully indicate that it marked a boundary associated with that represented by the fence-line. 04_BGAS125_131-246 16/11/07 15:50 Page 222

222 ANDREW SIMMONDS . s e r u t a e f l l a : y r o t c e R . 0 2 . g i F 04_BGAS125_131-246 16/11/07 15:50 Page 223

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 223

Pit 76 was also interpreted as belonging to this phase of occupation as it was cut by a beam slot forming part of Phase 2 structure 88 ( Fig . 20). The ceramic assemblage from pit 31 is slightly earlier than those from the majority of the features excavated, indicating that it too belongs in the earlier phase. The location of pit 41 within structure 88 suggests that the two are not contemporary, as it would present a hazard to the building’s occupants. It is therefore likely that this pit should also be assigned to Phase 1. All three of these features were large circular pits with steeply sloping sides and flat bases ( Fig . 21). They had very similar dimensions, with diameters in the range of 1.20–1.45 m and depths of around 0.5 m. All three pits contained domestic refuse – indicating that they were used as rubbish pits – in the form of pottery and animal bone, some of which exhibited signs of butchery. The later of the two fills in pit 41 (39) also yielded a piece of lead waste and 2nd-century pottery, while pit 76 contained a fragment of copper-alloy sheet and an iron nail. The fills of pit 31 produced pottery including a Malvernian tubby jar from context 33 and fragments of iron. Also dating to this phase was pit 56, a small, flat-based feature 0.65 m across and 0.22 m deep located near the southern corner of the excavation. Its only fill (55) contained a dense concentration of iron-working debris (see Keys, below), most of a Severn Valley ware jar and a few sherds of grog-tempered pottery (see Brown and Timby, below).

Phase 2 At some time in the 2nd century the boundaries marked by the fence-line and gully fell out of use and were replaced by a rectilinear structure (88) on a slightly different alignment. The structure was formed by beam slots aligned at right angles, running from north-west to south-east (72) and from north-east to south-west (86: Fig . 20). These had been partially truncated, resulting in the loss of the southern corner of the building and some of its south-west end. Part of the structure lay beyond the limits of the trench, and consequently its full dimensions could not be established. Within the area exposed in the excavation it measured 9.5 m from north-east to south-west and it was at least 6.5 m wide from north-west to south-east, although it is possible that it was narrower and that the north-west side had been removed by truncation. The beam slots on which the building was founded were 0.3 m wide and up to 0.18 m deep, with vertical sides and a flat base (Fig . 22 ). A pair of postholes (26 and 35) flanking the beam slot on the south-east side of the structure may have held clamping posts which secured horizontal timbers forming the superstructure of the walls. Another possible structural element forming part of the building was represented by posthole 65. This vertical-sided posthole was 0.5 m in diameter and 0.1 m in depth, and was located adjacent to beam slot 86 on it west side. A row of pits of varying sizes was dug along the gable end of the building. Pits 16 and 18 intersected but their fills (17 and 19) were indistinguishable, both consisting of reddish brown silty clay. It was not therefore possible to establish a stratigraphic relationship between them. Both were relatively shallow oval features with bowl-shaped profiles, pit 18 being slightly larger with dimensions of 1.3 m across and 0.3 m deep compared with a width of 1.0 m and a depth of 0.2 m for pit 16 (Fig . 22). Pit 16 contained an assemblage of animal bone and early 2nd-century pottery consistent with use as a rubbish pit, while pit 18 produced similar material, albeit in smaller quantities. Pit 42 was irregular in shape, comprising a circular bowl 0.7 m in diameter and 0.2 m in depth with a 0.5 m-long projection on its north-east side ( Figs . 20 and 22). Its basal fill was a layer of disturbed natural material (49) overlain by a dump of charcoal (43) 30 mm thick. This was sealed by a layer of backfill (44) containing flecks of charcoal and heat-discoloured soil and early 2nd- century pottery. The feature was cut slightly into the top of earlier pit 31. Pits 20 and 84 were located at the south-eastern end of this row of pits, near the south corner of the building. Both features are shallow and concave in shape, and probably represent the 04_BGAS125_131-246 16/11/07 15:50 Page 224

224 ANDREW SIMMONDS

Fig. 21. Rectory: Phase 1 sections. 04_BGAS125_131-246 16/11/07 15:50 Page 225

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 225

truncated bases of substantial pits. Pit 20 was 0.6 m in diameter and 0.13 m deep. Pit 84 measured 0.52 m across and was 0.1 m deep and it produced 1st- and 2nd-century pottery. A small number of other pits in the southern part of the excavation were also attributable to this phase of activity on ceramic grounds. Pit 12 was located 3 m from the south-west end of structure 88. At 60 mm deep and 0.50 m wide it was unusually shallow in relation to its width, and it contained a charcoal-rich fill (13). The base of a stake hole or small posthole (14) located immediately adjacent to feature 12 may be associated with its use. The truncated base of a pit (6) was located toward the south corner of the excavation. It was 0.7 m in diameter but only 90 mm deep, and the pottery and animal bone retrieved from it may indicate a function as a domestic rubbish pit. Features 58, 60 and 62 were identified in section in the south-west face of the trench. Feature 58 was a small pit or posthole 0.47 m wide and 0.18 m deep which produced no finds. Although both considerably larger than that feature, pits 60 and 62 were almost identical. Both were 1.1 m wide with steep sides and a flat base and were 0.35–0.40 m deep. Pottery and animal bone were retrieved from the fills of both features (59, 61). Pits 6, 56, 58, 60 and 62 were sealed by a spread of broken up slag 60 mm thick (54) which extended across the southern end of the excavation ( Fig . 20). No dating evidence was recovered from this layer, but its location outside the end of building 88 suggests that it may be contemporary with the occupation of the building. The fragmentary nature of this material suggests that it does not represent primary deposition, and it may have been deliberately laid down as a surfacing material, although it is not thick enough or compacted enough to be a metalled road surface.

Unphased Features A small number of features could not be allocated to a specific phase due to an absence of either stratigraphic relationships or datable artefacts. It is nevertheless probable that they are broadly contemporary with the Roman activity recorded on the site. Posthole 10 did not form part of any identifiable structure. It had clearly been truncated, with only the lower 70 mm surviving. It was 0.3 m in diameter and contained a single fill (11) which yielded a sherd of Central Gaulish samian ware dating to the first half of the 2nd century. Two pits (22 and 24) were located in the central part of the site, within the footprint of building 88. Both were oval, bowl-shaped pits, and had similar dimensions, pit 22 measuring 0.70 × 0.5 m with a depth of 0.35 m and pit 24 measuring 0.8 × 0.6 m and 0.22 m deep. Neither feature contained any finds to indicate a date or function, and so they cannot be attributed to a specific phase, although their form and the nature of the fills are consistent with the other Romano-British features found on the site. Three small postholes (74, 80 and 82) were uncovered in the eastern part of the site. None of these features contained datable artefacts, and it was not possible to relate them to other structures or features on spatial grounds. Excavation of the soakaway to the west of the main excavation area revealed part of a gully (context 8). It was 0.35 m wide and 0.20 m deep and aligned from north-west to south-east, at a right angle to the principal axis of the building uncovered in the main area of the excavation (Fig. 20). The similar soakaway pit dug to the east of the main area was excavated entirely within a large archaeological feature (context 46: Fig. 20). This is believed to be the large boundary ditch recorded during the evaluation (OA 2003). The gravel surface interpreted in the evaluation as possibly the metalling of a Roman road extended into the eastern corner of the area of the excavation but was later revealed to be a modern garden path. 04_BGAS125_131-246 16/11/07 15:50 Page 226

226 ANDREW SIMMONDS

Fig. 22. Rectory: Phase 2 sections.

THE FINDS

Pottery by Kayt Brown and Jane Timby A small assemblage of 339 sherds of pottery (3,607 g) was recovered from the excavations as detailed in Table 19. The assemblage appears to span the 1st and 2nd centuries AD. The condition of the sherds is generally good, with a number of large and unworn sherds and good preservation of surfaces. Where relevant the fabrics are cross-referenced to the Gloucester City type fabric (TF) series. The assemblage was quantified by sherd count, weight and estimated (rim) vessel equivalents (EVE). The assemblage was recorded directly onto an access database, which, along with the fabric series, forms part of the site archive.

Fabrics and Forms A small quantity of hand-made pre-Roman native ware was recovered comprising two fabrics; a grog-tempered ware (Glos TF 2C) and a Malvernian rock-tempered ware (Glos TF 18; Tomber and Dore 1998: MALREA). The only featured sherds are in the latter ware comprising a Malvernian tubby jar from context 33 and a dish with internal and external burnished lattice decoration from context 64. Such wares feature commonly until and sometimes beyond the 2nd century AD. The dominant fabrics are Severn Valley wares (Glos TF 17, 11B and 23; Tomber and Dore 1998: SVW OX), which account for over half the assemblage by sherd count (66.4%). Forms represented include jars, bowls and dishes, among them a curving-sided dish and a dish with a footring scar. Decoration on these fabrics comprised burnished zones and lattice decoration. 04_BGAS125_131-246 16/11/07 15:50 Page 227

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 227 0 . 4 9 . . 0 0 0 1 7 8 6 0 0 0 0 3 8 0 0 4 0 ...... 0 7 2 % 0 0 0 2 4 5 2 0 0 0 0 6 0 0 0 7 0 0 1 5 1 . s t n e l E a 0 1 8 6 2 0 4 8 v 0 0 0 8 0 0 0 0 3 0 0 0 0 8 V i 1 6 2 1 2 2 2 3 u E q e l e s s e 0 . 1 v . 9 1 2 3 1 2 3 9 3 7 8 7 2 2 1 0 6 5 ...... 0 7 d % 0 0 0 0 1 8 1 0 0 0 0 0 3 0 5 3 5 0 0 6 e 1 t a m i t s . 7 e 1 5 4 3 7 3 0 2 t 3 0 9 6 1 0 4 0 4 8 d 4 7 9 1 8 8 0 0 4 6 3 1 3 4 3 1 2 3 2 1 , , n 2 1 1 1 2 W a 2 3 ) s m a r 0 g . 9 4 . . 6 9 3 5 6 8 6 3 3 3 3 2 3 5 3 4 6 ...... 0 n 5 6 % i 0 0 0 1 0 1 0 0 0 0 0 3 0 1 0 4 0 0 ( 1 6 1 t h g i e w . , 9 5 t o 4 1 5 2 2 3 1 5 6 2 1 1 1 1 2 1 5 1 2 3 n 5 1 1 d 2 3 N u e r o e c n p o d r g m e e a e r l t h f a d - s p e i e e w e d l r e y r r n e s e r n d a a b o p t a a p e t n i w w p e y m n s r h w a i m r r a a i l s a n y m d i e a i e e j e s e y a e r t e o t r r - r m m n s r t d w s - r a e a a e c r a i i s h g o v k n t g n s y h e u w i d w e c a a p h l i s s n r d d w b r s h t i r h o a x a a y d e f u i l a o s n r o r e o o w e k e i d o n o u t a w l w n l s e w c c e l s s e a n i r r o D r k u a s n p s i a k c a e e e l i d a a y t n i f G c t t a i o u t m b d V i i m h e s s u a c x n p o l i o t G l i r i e h c e e r h t a x o e a t t n r e b g c c e e r s c - h c r o . w m c c t s v i u u t a e e c c c g r l s r f W W n e e r c s s s o o u i v a o u e i i i i l l o e t o n n W V V e o r o i i n f m m m F f m G G S S S D S S s C M g D a u q : ? y r A B o 1 5 3 7 1 8 C t 1 7 1 2 1 4 1 1 2 c a i e S F F F F F F F F F r c n n Y R i a D a a T T T T T T T T T t r i i . W I E F r 1 b 9 W X X Y R L L L L L o L L L L m m W a E 1 a a O O W B F G G G G G G G m G G s s G G F e l b a e r T a n l w a s w t n o e r l l o v n a o i a i t k t c p g a n o e o m T U R L I N 04_BGAS125_131-246 16/11/07 15:50 Page 228

228 ANDREW SIMMONDS

Unsourced, but presumably local reduced and oxidised coarse wares make up the bulk of the remaining assemblage. Of particular note is a fine grey, slightly micaceous ware which accounts for 15.9 per cent of the assemblage. This features as jars, including a bead rim jar, bowls and a single beaker rim sherd. Decoration on these coarse ware sherds included rouletting, rustication on the shoulder of a jar sherd, and the use of cordons at the base of neck. The typology would suggest an industry dating to the late 1st–early 2nd century. Other local wares include three possible sherds from the Gloucester kilns (TF 11A and 7), all from flagons, and two sherds with an ironstone temper probably from a local source within the Forest of Dean. Continental imports are limited to two sherds of samian, one, possibly from Les Martres-de- Veyre, a dish (Dragendorff form 18/31R) and the other, South Gaulish, a decorated bowl (Dragendorff form 29). A possible import is a mortarium from context 3. The fabric is very fine, buff to pink in colour, with a sparse scatter of fine quartzite trituration grits. The vessel has a lead repair rivet still in situ .

Metalwork by Leigh Allen The metalwork assemblage comprised a total of four objects. Two iron nails were recovered from fill 17 of pit 16 and fill 77 of pit 76. Context 77 also yielded a piece of copper-alloy sheet (SF 1). The object comprises three fragments of folded sheet metal at least one of which has rough perforation through it. There appears to be a rivet through one fragment and a neatly cut and finished edge on another. It is possible that these are fragments from a sheet metal vessel or vessel repair. A lump of melted lead (SF 2) weighing 91 g was recovered from fill 39 of pit 41.

Metal-Working Waste by Lynne Keys A total of 46.1 kg of material identified as iron slag was recovered from the excavation, as detailed in Table 20. Much of the slag is in a broken condition and so could only be characterised as undiagnostic. Of the material which could be securely identified, tap slag, the most diagnostic by-product of the smelting process, formed the bulk of the assemblage. It is likely that the bulk of the material described as run slag and much of the undiagnostic material is in fact also broken tap slag. Of the two largest individual assemblages, pit 56 was filled by a deposit (55) composed almost entirely of smelting debris while context 3 consisted of similar material, albeit in a more fragmentary state. The presence of hammerscale and other micro-slags in a number of deposits indicates that the assemblage represents the debris from smithing activities as well as smelting. This debris is more likely to result from primary smithing to remove excess slag from the bloom than from the production of finished artefacts. No evidence for furnace structures or smithing hearths was found in the excavation, but pieces of furnace lining were recovered from layer 3 and from deposit 55, the fill of pit 56. Fragments of iron were present in context 33, the second of the three fills of pit 31, and in layer 54. 04_BGAS125_131-246 16/11/07 15:50 Page 229

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 229

Table 20. Rectory: quantification of slag by context (* hammerscale present).

Context Tap Run Dense Micro- Hammer- Undiag- Total (g) slag (g) slag (g) slag (g) slags (g) scale nostic (g) 3 749 120 1,554 2,423 11 150 217 367 13 64 46 110 21 714 88 1,612 2,414 29 440 52 492 31 34 34 32 62 * 16 78 33 468 254 188 910 34 52 * 52 39 86 86 40 * 43 76 *884 44 90 * 12 102 54 62 8708 38 277 * 11,576 20,661 55 10,216 756 776 13 5,476 17,237 61 * 24 24 64 34 50 466 550 73 * 77 * 24 24 87 58 * 58

ENVIRONMENTAL EVIDENCE

Animal Bone by Emma-Jayne Evans

Results A total of 375 (2,487 g) fragments of bone and teeth were recovered during the excavation. The refitting of broken elements reduced this fragment count to 289 pieces. The bones survived in good condition, with the majority scoring 1 according to Lyman’s criteria (Lyman 1996). Despite the generally good condition of the bones, there are many small broken fragments, many with fresh breaks, resulting in only 63 fragments (21.8%) being identifiable to species (Table 21). A single bone was certainly identified as sheep which has been incorporated into the sheep/goat category in Table 21. Cattle appear to be present in slightly larger numbers than sheep/goat. Although this is based on very low numbers, it does follow a pattern commonly observed on Roman sites, in which cattle are often present in greater numbers than sheep/goat. Fusion data gave some limited information regarding the age at death of some animals present at the site. The assemblage from Phase 1 included juvenile cattle, sheep/goat and pig, while juvenile cattle and pig were present in layer 3. Carnivore gnawing was noted on three bones from pit 31, one bone from pit 6, and on two bones from the Roman soil layer (context 3). This may indicate that these bones were left exposed on the surface for some time before finally being covered over. Butchery marks were noted on several bones from each phase. 04_BGAS125_131-246 16/11/07 15:50 Page 230

230 ANDREW SIMMONDS

Discussion The small size of the excavated assemblage does not allow any conclusions about animal husbandry regimes to be made, although the main domestic species of cattle, sheep/goat and pig are all present. The presence of several juveniles and the occurrence of butchery marks indicate that these animal remains are likely to represent household domestic waste.

Charred Plant Remains by Denise Druce A total of fifteen bulk samples were collected from a range of features during the excavation. Following an initial assessment, three samples were selected for further analysis (Table 22).

Table 21. Rectory: total number of bones according to species and phase.

Phase cattle sheep/goat pig horse unidentified Total 1 16 14 51142 178 2661–36 49 Soil layer 3 851–48 62 Total 30 25 71226 289

Results All three samples contained cereal grains, although sample 4, from fill 43 of pit 42, only contained three identifiable grains. Sample 8, from beam slot 72, produced the largest cereal grain assemblage, and this was dominated by Avena sp. (oats) and Triticum sp. (wheat) grains. Many of the grains were of the short-grained variety, resembling free-threshing wheat such as Triticum cf. aestivum (bread wheat). Triticum sp., including Triticum cf. spelta/dicoccum (spelt/emmer wheat), grains were also present in all three samples. All of the samples contained Triticum spelta (spelt wheat) glume bases/spikelet forks, in addition to which samples 8 and 10 also contained Avena sp. awn fragments and limited numbers of culm nodes. Sample 8 also contained Triticum aestivum and Secale cereale (rye) rachis. The weed seed assemblages from all three samples contained taxa typical of grassland/cultivated land, dominated by Poaceae (grass) and Fabaceae (pea family), with some element of wet or damp ground. Additionally, Corylus avellana (hazel) shell and Brassica sp. (cabbage etc) seeds were recorded in small quantities, but even if these plants were collected or cultivated as food their low amounts suggest that neither was likely to have formed a major part of the diet.

Conclusions The charred plant remains contained taxa typical of crop husbandry during the Roman period in Britain. Spelt wheat, considered to be the principle cereal crop during this period, was the dominant crop represented, along with free-threshing wheat such as bread wheat and rye. Free-threshing wheat is not so common in the Roman period, and its presence here and at other sites in Gloucestershire (e.g. Birdlip Quarry: Pelling 1999) and south-east Wales (e.g. Biglis: Hillman 1981 cited in Caseldine 1990) may reflect a regional pattern. The relatively high numbers of Fabaceae seeds has been attributed elsewhere to arable agriculture and decreasing soil fertility (Jones 1978). Weeds indicative of both damp and well-drained ground were present, indicating that a variety of land was being exploited. As Dymock is situated on the edge of the floodplain of the river Leadon 04_BGAS125_131-246 16/11/07 15:50 Page 231

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 231

Table 22. Rectory: the charred plant remains.

Sample 48 10 Context. 43 73 32 Feature Pit 42 Beam slot 72 Pit 31 Sample size (litres) 20 40 40 Cereal Grain Triticum sp. Wheat 16 9 Triticum sp. Short-grained/free-threshing wheat 1 20 – Hordeum vulgare Barley undiff. 1– 1 Avena sp. Oats – 40 1 Secale cereale Rye –4 – Cerealia indet. Indeterminate grains 3 17 11 Total Grain 6 87 22 Cerealia indet. frag. 19 174 48 Cerealia indet. Sprouted embryos 2 17 29

Cereal Chaff Triticum spelta Spelt wheat glume base 8 38 586 Triticum spelta Spelt wheat spikelet fork 1– 4 Triticum sp. Wheat spikelet fork base. 1 31 78 Triticum aestivum Bread wheat rachis –6 – Secale cereale Rye rachis –8 – Avena sp. Oat awn frag. – 28 12 cf. Cerealia indet. Culm node –5 2 Total Chaff 10 116 682 Triticum cf. spelta frag. cf. Spelt wheat glume base frag. – 60 420

Other Charred Edibles Corylus avellana frag. Hazelnut shell frag. 51 – Brassica sp. Cabbages –1 –

Weed Seeds Ranunculus sp. Buttercups –6 – Ranunculas cf . repens Creeping Buttercup – 10 – Papaver sp. Poppies 3– – Chenopodium/Atriplex Goosefoots/Oraches 79 3 Stellaria media Common Chickweed 11 1 Agrostemma githago Corncockle –9 – Cerastium sp . Mouse-ears – 14 – Polygonum undiff. Knotgrasses 1– 1 Polygonum aviculare Knotgrass –3 – Polygonum lapathifolium Pale Persicaria 1– – Fallopia convolvulus Black-bindweed –1 – Rumex acetosa Common Sorrel – 10 1 Rumex acetosella Sheep’s Sorrel 2 12 – Rumex obtusifolius Broad-leaved Dock 83 – 04_BGAS125_131-246 16/11/07 15:50 Page 232

232 ANDREW SIMMONDS

Viola sp. Violets –1 – cf. Hypericum sp. St John’s-worts –6 – Anagallis arvensis Scarlet Pimpernel –6 – Rosaceae Rose family –1 – Fabaceae <4 mm Pea family 18 45 7 Vicia/Lathyrus/Pisum Vetches/Peas/Garden Pea 11 – Trifolium/Medicago/Lotus Clovers/Medicks/Trefoils – 30 – Conium maculatum Hemlock –2 – Plantago sp. Plantains 2 18 – Plantago major Greater Plantain –2 – Rhinanthus minor agg. Yellow-rattle –8 – Galium sp. Bedstraws 1– 2 Asteraceae Daisy family –1 1 Lapsana communis Nipplewort –2 – Leontodon sp. Hawkbits –2 1 Anthemis cotula Stinking Chamomile 2– 1 Chrysanthemum segetum Corn Marigold –8 – Tripleurospermum inodorum Scentless Mayweed –2 – Carex spp. trigynous Sedges- three sided 11 14 1 Carex spp. lenticular Sedges- two sided –4 3 Poaceae <2 mm Grass family 17 327 44 Poaceae 2–4 mm Grass family 5 30 72 Poaceae >4 mm Grass family –– 1 Bromus sp. Bromes 1 10 24 Indet. Indeterminate seeds 11 10 – Indet. Unknown seeds 11 – Total Weed Seeds 93 609 163

it is possible that some low lying areas were being cultivated during the summer months when flooding was less likely to have taken place.

Wood Charcoal by Dana Challinor Two samples were analysed for wood charcoal remains. Sample 1 was from fill 13 of pit 12, while sample 13 was from context 77, the fill of pit 76. A sample from a large irregular pit with a dump of charcoal, slag and burnt soil (feature 42; context 44) was briefly examined to confirm the provisional identifications made in the assessment by Elizabeth Huckerby. The results of the analysis by fragment count are given in Table 23.

Methodology Fragments of charcoal >2 mm were extracted from the flots and identified. The charcoal was fractured and sorted into groups based on the anatomical features observed in transverse section at × 10 and × 20 magnification. Representative fragments from each group were then selected for further examination using a Meiji incident-light microscope at up to × 400 magnification. Identifications were made with reference to Schweingruber (1990), Hather (2000) and modern 04_BGAS125_131-246 16/11/07 15:50 Page 233

EXCAVATIONS AT THE RECTORY, DYMOCK, 2002 233

Table 23. Rectory: results of the charcoal analysis.

Phase AD 2nd/3rd century Cut number 12 76 Context number 13 77 Sample number 1 13 Volume floated (litres) 10 40 % flot identified 25 25 Ulmus sp. elm –1 Quercus sp. oak – 91sh Corylus avellana hazel 12 7 Betula sp. birch –8 Maloideae hawthorn type 90r 13 Hedera helix ivy 3– Indeterminate 12 4 Total 117 124

Key: h = heartwood; r = roundwood; s = sapwood

reference material. Classification and nomenclature follow Stace (1997). The maturity of the wood was assessed where the condition of the wood permitted but the material was too fragmented to examine growth rings.

Results Six taxa were positively identified (Table 23). The taxonomic level of identification varied according to the biogeography and anatomy of the taxa. The preservation of the charcoal was average; it was generally fragmented with most fragments less than 4 mm in size. The indeterminate fragments are likely to represent additional specimens of taxa positively identified. Both oak sapwood and heartwood were identified in sample 13, while sample 1 produced a large quantity of narrow roundwood fragments (<20 mm in diameter). The brief examination of charcoal from context 44 (pit 42) confirmed that a range of taxa was indeed present, including Quercus (oak), Maloideae (hawthorn type), cf. Rhamnus cathartica (purging buckthorn) and Prunus spinosa (blackthorn), but the provisional identification of Ilex aquifolium (holly) was not verified.

Discussion The analysis of the two charcoal samples shows a distinct contrast in the selection of taxa and type of wood for fuel. The assemblage from pit 12 (sample 1) was dominated by narrow roundwood of hawthorn type (Maloideae), whereas the assemblage from pit 76 (sample 13) was chiefly composed of oak, heartwood and sapwood. While this variation may relate to changes in woodland resources or management practices over time, it is more likely to relate to the function of the fire. The processes of iron smelting and smithing would both have required charcoal as fuel (Edlin 1949, 160; Cleere and Crossley 1985), although it is not possible to determine from archaeological charcoal whether it was used as charcoal or as wood fuel. Traditional methods for making charcoal utilise shallow pits with layers of straw/grass or clamps of roundwood to shut out 04_BGAS125_131-246 16/11/07 15:50 Page 234

234 ANDREW SIMMONDS

the air (Edlin 1949, 160). However, the hawthorn type stems from pit 12 were particularly narrow and it is unlikely that they would have provided the prolonged high temperatures required for smelting, although the intense, short-lived heat provided could have supported some smithing activity. Nevertheless, the complete absence of oak from this sample is unusual. Traditionally, oak heartwood forms a significant component of the fuelwood used for smithing (Edlin 1949), and this is also supported by archaeological assemblages from features associated with both iron smelting and smithing. Evidence from other sites suggests a dominance of oak, with a range of other, variable, taxa (e.g. Campbell 1998, 37; Challinor forthcoming; Cleere and Crossley 1985, 37; Figueiral 1992, 189; Gale 1999, 378). Even where there is evidence of alternative fuels being used, indicating pressure on woodland resources, oak remains a component of the assemblage (Murphy 2000, 220). It is possible that the presence of small quantities of other non-oak taxa may be explained by their use as an aid to ignition or as an accidental inclusion. Certainly, the ivy in sample 1 is likely to have been attached to a host tree which was deliberately felled/gathered for use. It is interesting to note that sample 5 from pit 42 also produced a range of taxa, including some additional taxa not identified in the analysed samples. This suggests that, apart from the need for oak for smelting, there was little deliberate selection of wood for charcoal burning.

DISCUSSION

The ceramic evidence indicates that occupation at the Rectory site commenced during the late 1st or early 2nd century with the establishment of linear boundaries in the form of a fence-line extending NE–SW across the area of the excavation and a gully extending at right angles to it toward the north-west. These features were associated with rubbish pits, the fills of which contained pottery and animal bone, some of which exhibited signs of butchery. Although there was no evidence for structures associated with these features within the excavated area the domestic refuse deposited in the pits and in the gully was presumably generated by occupation close by. This phase of activity would be contemporary with occupation recently recorded in excavations c.150 m to the north-west of the current site at Rose Cottage and Winserdine (Tavener 2001) and c.250 m to the east at Dymock Sewage Treatment Works (Catchpole 2000 and above). During the 2nd century the earlier features were superseded by the construction of a rectilinear building (88) associated with a number of pits, some of which were dug along the building’s gable end. The building measured at least 9.5 × 6.5 m and was founded on ground-fast beams. The absence of other building materials from the excavation would indicate that it was constructed from timber or wattle and daub, with a thatched, shingle or turf roof. The large ditch recorded during the evaluation and encountered in the soakaway pit to the east of the main excavation area is aligned parallel to the building’s south-east wall and may have been part of an enclosure ditch within which the building stood. A gully (8) excavated in the western soakaway may be a related boundary or the beam slot of a second building. This is similar to the results of the excavation at the sewage treatment work, where a series of rectangular buildings up to 22.5 m long, associated with pits and burials, was identified within a large ditched enclosure (Catchpole 2000 and above). The evidence from these two sites suggests that the settlement may have been divided into a number of rectilinear plots arrayed alongside the road, each containing an individual dwelling together with out-buildings, pits and other associated features. 04_BGAS125_131-246 16/11/07 15:50 Page 235

ROMAN DYMOCK: ORIGINS AND FUNCTION 235

Acknowledgements Thanks are owed to Neil Percival of Bruton Knowles Property Assets Consultancy which commissioned the work on behalf of its clients, the diocese of Gloucester. The diocese paid for the fieldwork and the post-excavation work including this publication. The fieldwork was managed for OA by Dan Poore and the post-excavation programme by Alan Hardy. The fieldwork was monitored by Charles Parry of Gloucester County Council Archaeological Services who also provided helpful comments on the draft text. The editor is thanked for assistance during the publication of this report. Thanks are owed to all of the specialists for their reports and to Oxford Archaeology’s finds, environmental and archive staff for their work at various stages of the project. Dana Challinor would like to thank Dr. Mark Robinson (Oxford University Museum) for reading a draft of her report. Philippa Bradley edited the report.

Archive The archive will be deposited at Dean Heritage Centre, Soudley, under accession number SOYDH 2002.75.

Dymock: its origins and function

By TOBY CATCHPOLE, ANDREW SIMMONDS and TIM COPELAND

Recent excavations at the Rectory site, at Dymock Sewage Treatment Works and at Rose Cottage and Winserdine (Tavener 2001) have all failed to uncover evidence for either Iron-Age occupation or military activity at Dymock. This leads to the conclusion that the settlement founded in the Roman period, probably during the second half of the 1st century AD, was from the outset civilian in nature. Dymock certainly comes within the category of ‘roadside settlements’ described by Roger Finch Smith (1987) as there is evidence of buildings within 50 metres of the road. Its description as a large rural settlement (McWhirr 1981, 59) is probably based on the spread of finds reported by Gethyn-Jones. The function of this type of settlement is much more difficult to establish. On the evidence of the archaeological interventions that are the subject of these papers, we can now postulate a primarily agricultural settlement with a significant industrial component stretching along a metalled road from at least the Rose Cottage to the cricket pitch sites in the late 1st century. It may well be that official travellers were being housed in the settlement. Black (1995) suggests that roadside settlements were deliberately founded by the Roman administration to provide a workforce and lodgings to aid the functioning of the transport system. Dymock was either deliberately founded or developed due to market availability but as the same evidence could be used to make a case for either scenario it is extremely difficult to determine if one or both applied here. The settlements at Kenchester, Stretton Grandison, Dymock and Worcester (and possibly Gloucester if the postulated southern road from Dymock did exist) are equidistant from each other and this may indicate that these settlements were planted as an act of deliberate policy, perhaps associated with the functioning of the imperial transport and postal system. 04_BGAS125_131-246 16/11/07 15:50 Page 236

236 TOBY CATCHPOLE, ANDREW SIMMONDS AND TIM COPELAND

Economy The animal bones from the sewage works and Rectory sites, along with the charred plant remains indicate that mixed farming was practised during the occupation of the site, with spelt wheat the main crop and cattle the main source of meat. Such a predominantly agricultural economy is typical for a roadside village of this date (Rivet 1975). Iron smelting is now known to have been carried out over a distance of more than 700 m from the Rose Cottage site to the cricket pitch in the 1st and early 2nd century AD. Any attempt to estimate the significance of this industry should bear in mind that only a representative sample of slag was collected at the sewage works site. A recurring aspect of the evidence from the three recent excavations at Dymock is the lack of physical remains of shaft furnaces, leading to a suggestion that smelting was carried out ‘near by’. It is, of course, possible that this is the case, although if similar results are obtained in the future it may well be decided that the furnaces were constructed in such a way as to leave little physical trace. Romano-British shaft furnaces need have been no more than c. 0.5 m in diameter. They were always protected from the weather in ventilated structures of some kind and they would usually have been associated with tapping pits that could be cleared out (Hoyle et al. 2004, 103). The casting of copper-alloy objects required a bowl hearth, with or without a superstructure, to hold a charcoal fire and the crucible in which the metal was melted. Fire hardens the clay in a casting hearth to a depth of only 10–20 mm (Hammer 2003, 16–20) and therefore significant quantities of burnt clay are not to be expected. The unburnt bulk of the superstructure of casting hearths and furnaces may therefore disintegrate and be archaeologically invisible. The shallow bowl-shaped pits and smaller timber structures recorded at the sewage works and Rectory sites and the unexcavated stone structures from the evaluation at Rose Cottage therefore remain as candidates for metal-working locations. The iron-ore sources used at Dymock are uncertain. Ore samples examined from the sewage works were identifiable only as having probably originated from the Forest of Dean area (Dungworth, above). Gethyn-Jones noted a claim (Bick 1978) that iron ore was mined from under Castle Tump, 2 km to the south-east of Dymock, adjacent to the modern road to Newent, where two undated ‘levels’ had been driven into the bank. Other evidence for iron ore in the district has recently been summarised by Hoyle et. al. (2004). It is assumed that copper-alloy ingots were imported to Gloucestershire; two Roman brass ingots have been recovered from Gloucester (Justine Bayley pers. comm.). The evidence currently is that extensive metal working ceased at Dymock in the mid 2nd century. We can only speculate as to the reasons for this. It may have been that the military market went elsewhere, although there was presumably still a need for great quantities of iron at Gloucester and the other emerging towns and is it perhaps more likely that local ore sources were exhausted.

The Later Settlement The absence of material, particularly forms of ceramic evidence, dating to the later part of the Roman period is a recurring feature of the archaeology of Dymock. Remains dating to the later 3rd and 4th century have not been encountered in recent excavations and are scarce among the reports of casual finds made throughout the 20th century. Occupation at the site at Rose Cottage and Winserdine (Tavener 2001) came to an end early in the 2nd century while at both the Rectory and at the sewage works the latest activities date to the late 2nd or early 3rd century. By the mid 2nd century, however, the adult dead were being buried roughly in the centre of this strip at the sewage works site and also at its western end at Rose Cottage. This argues that these areas were not considered part of the settlement at that time 04_BGAS125_131-246 16/11/07 15:50 Page 237

ROMAN DYMOCK: REGIONAL CONTEXT 237

and that the settlement was shrinking to a focus outside the areas so far investigated. Occurring during a time of relative peace and security, such a decline is most likely explained as the result of economic or political factors that have left no archaeological trace. The excavations reported above have little to tell us about the later settlement except that it was not focused on any of the sites. That there was a settlement at Dymock throughout the 3rd and 4th centuries is proven by, amongst other things, the coin finds, both those published and those posted on the Portable Antiquities Scheme website, which indicate on-going coin loss through at least to the house of Valentinian. Local place-name evidence suggests that the Dymock area kept its British identity and language longer than most of Gloucestershire (Smith 1965, 29). Current evidence therefore does not indicate that Roman Dymock was totally abandoned but that it shrank or changed its character and moved, or both. The sewage works site is located on the edge of a low river bluff overlooking the river Leadon. It is sited on a not inconsiderable slope with higher and more level ground to the west at the location of the medieval church and level ground to the east at the cricket pitch. These topographically more favourable sites may or may not have continued in use. The location of a post-medieval ‘stoneberrow’ field to the south-west (Fig . 12; Gwatkin 1992) may suggest a further area worth investigating for signs of the later settlement. The character of the later settlement is not known. Whether it shrank to a smaller village, became a scatter of farms, or developed into the estate centre of a local aristocrat remains to be seen. No indication of a late Roman villa has so far been discovered within the immediate area, the little-understood site at Donnington mentioned in to the introduction to these combined reports being, so far, the only nearby candidate.

Regional Context

By TOBY CATCHPOLE

The picture of Romano-British rural settlement in the Severn Vale has become fairly well established. Recent publication of the excavations at Bishop’s Cleeve (Parry 1999; Enright and Watts 2002), Tewkesbury (Walker et al. 2004), Cheltenham (Catchpole 2002) and Brockworth (Thomas et al . 2003) suggest a pattern more indicative of the continuation of late Iron-Age traditions than of major changes after the Roman conquest. This information has recently been summarised by Neil Holbrook in papers to the 25 years of Gloucestershire Archaeology Conference and the Birmingham Roman Archaeology Conference (2006, 108–14). Following Millett (1990, fig 16), late Iron-Age Dymock was within the tribal area recognised by the Romans as Dobunnic, an area of centralised control where the Romans supported existing power structures rather than imposed their own, an argument which is consistent with the available Dymock evidence. A different pattern is suggested by Peter Guest’s recent work at a 1st-century site at Lyonshall, Herefordshire, 45 km to the north-west of Dymock, where Iron-Age farming traditions continue similarly but with virtually no evidence of Romanised material culture to indicate changes after the Roman conquest (Peter Guest pers. comm.). The major difference between Dymock and the recently published sites in the Severn Vale is the evidence for metal working. A recent summary of the Forest of Dean iron industry in the Roman period suggested that a group of centralised production centres could be identified in the 04_BGAS125_131-246 16/11/07 15:50 Page 238

238 TOBY CATCHPOLE

1st two centuries AD, after which they declined (Hoyle et al. 2004, 12), perhaps to be replaced by smaller scale and intermittent production at villa estates and towns. It is also possible that production shifted from centres north of the Forest to sites such as Woolaston (Fulford and Allen 1992) and Blakeney (Barber and Holbrook 2000), utilising the Severn fringes of the Forest and the easier transportation to developed markets that access to the river allowed. The influence of Worcester is difficult to assess as the chronology of the iron industry there is not fully understood. Current evidence, however, suggests a predominantly later Roman industry and that large-scale smelting did not commence until several decades after it had apparently ceased at Dymock (Burnham and Wacher 1990, 232–4; Dalwood and Edwards 2004, 41, 45). Ariconium is the largest of the known centralised centres of the early Romano-British iron industry in the Forest of Dean area. The evidence for the settlement has recently been collected and assessed (Jackson 2003) and there is a strong case that Dymock is similar in many respects, if on a much smaller scale. Jackson argues that indigenous élites continued to control the area, including the iron industry, and that the settlement operated a mixed economy of which iron working was only a part. As at Dymock, there is evidence from Ariconium for the production of copper-alloy brooches alongside iron working. Dymock is not alone as an outlier from the main Forest of Dean centres of iron production. Newent, 5.5 km to the south-east, is known to have ore sources. Ore was mined at Oxenhall near the town in the 19th century and waste cinders from iron production were re-worked from at least the 17th century. The May Hill area, south-west of Newent, is also known as an ore source (Hoyle et al. 2004, 89). Unfortunately the only Roman iron-working site so far excavated at Newent was carried out under salvage conditions after a site strip (Erskine 1996) but slag dumps and pits, ditches and gullies containing much slag and Roman pottery were recorded. Presumably much more remains to be investigated in the area. A potentially interesting parallel to Dymock is Margidunum , on the Fosse Way near Bingham, Nottinghamshire, where a series of 1st-century ditched enclosures containing timber buildings and slag filled pits was excavated producing high-quality pottery and metal work (Burnham and Wacher 1990). Both Dymock and Margidunum are insufficiently well understood to do more than point out a possibility of a recurring pattern at this stage. We can therefore see Dymock as one of a group of settlements involved in varying scales of iron production in the Forest of Dean and the surrounding area in the early Roman period. These sites show signs of the trappings of Romanised wealth in their material culture at an early date but, although we now have some evidence that Dymock may have housed a minor establishment of the Cursus Publicus , the wider picture on current evidence is that local control over the trade brought high-status items to the indigenous élites, who nevertheless continued their own traditions in many aspects of their lives.

Future Research Directions

By TOBY CATCHPOLE

Much remains to be clarified, not only regarding the early Roman period, about which these excavations have much to say, but also regarding the later phases of activity in the 2nd century 04_BGAS125_131-246 16/11/07 15:50 Page 239

ROMAN DYMOCK: FUTURE RESEARCH DIRECTIONS 239

and beyond, about which they have provided only limited information. There is much to discover about the full extent of industrial activity, the reasons why it commenced and ceased, and the true nature of any military/official involvement. Was Dymock a deliberately founded roadside settlement of the type suggested by Black (1995)? Was there sufficient activity at the settlement from the mid 2nd century onwards for Romano-British Dymock to continue to be classified as a small town at that period? A further subject which might merit investigation is the evidence for re-organisation or other disruption at the end of the 1st century suggested by the possible hiatus in activity from c. AD 100 to 120 at the sewage works site. A similar hiatus noted at Deansway in Worcester (Halwood and Edwards 2004, 43) may indicate a regional pattern or a chance similarity between the two sites. The other excavations at Dymock suggest some re-organisation at this time, although not conclusively. With the archive from the 1960s Malvern Research Group now available, the contents may add significant detail to our knowledge of Roman Dymock, particularly if clear evidence for activity underlying the Roman road can be examined and understood. There is significant metal-detecting activity in the area surrounding the village, and the reporting of finds via the Portable Antiquities Scheme and the pointing out of the possible presence of mould fragments to known detectorists might produce further useful information as to the products of local industry in the Roman period. It should also be borne in mind that the subsoil at Dymock is not suited to the easy identification of cut features, such as those indicating timber buildings, on aerial photographs or in rapidly excavated and investigated evaluation trenches and there is a case for more geophysical survey as an evaluative technique, particularly as this has proved successful on the Lias clays of the Severn Vale to the east (Holbrook 2006).

ILLUSTRATION ACKNOWLEDGEMENTS

Figures in this set of papers were produced by Toby Catchpole except for nos. 13–15 of the sewage works pottery by Jane Timby, nos, 17–19 of the sewage works registered finds by Jo Richards, and nos. 20–22 the plans and sections of the Rectory site by Julia Moxham.

BIBLIOGRAPHY

Allen, D., 1961. A study of the Dobunnic coinage, in E.M. Clifford, Bagendon, a Belgic Oppidum (Cambridge), 75–149. Allason-Jones, L., 1991. ‘Objects of copper-alloy’, in N. Holbrook and P.T. Bidwell, Roman Finds from Exeter (Exeter Archaeol. Rep. 4), 242–63.

Barber, A.J., and Holbrook, N., 2000. ‘A Roman Iron-smelting site at Blakeney, Glos. excavations at Millend Lane 1997’. Trans. B.G.A.S. 118 , 33–60. Bass, W. M., 1995. Human Osteology: a laboratory and field manual (4th edition, Missouri Archaeol. Soc.). Baxter, I., 1999. ‘Animal Bone Assessment Report, Glos. 15285, Dymock Sewage Treatment Works, Dymock, Glos.’ (unpublished typescript rep.). Bayley, J., 1985. ‘Interim report on some clay moulds and associated finds from Compton Dando, Somerset’ (Ancient Monuments Laboratory Rep. 4639). Bayley, J., 1988. ‘Non-ferrous metal working: continuity and change’, in E.A. Slater and J.O. Tate (eds.), Science and Archaeology, Glasgow 1987 (BAR British Series 196 , Oxford), 193–208. 04_BGAS125_131-246 16/11/07 15:50 Page 240

240 ROMAN DYMOCK: BIBLIOGRAPHY

Bayley, J., and Butcher, S., 2004. Roman brooches in Britain: a technological and typological study based on the Richborough collection (Soc. Antiq. London Research Committee Rep. 68 ). Beltrán Lloris, M., 1970. Las Anforas Romanas en España (Zaragosa). BGS 1988. British Geological Survey map 1:50,000, England and Wales Sheet 216, Tewkesbury. Solid and Drift Geology . Bick, D.E., 1978. ‘Dymock, Castle Tump’, in B. Rawes (ed.), ‘Archaeol. in Glos. 2, 1976–7’, Trans. B.G.A.S. 96 , 85. Biddle, M., and Cook, L., 1990. ‘Buttons’, in M. Biddle, Object and Economy in Medieval Winchester (Winchester Studies 7.ii , Oxford), 571–81. Black, E.W., 1995. Cursus Publicus The infrastructure of government in Roman Britain (BAR British Series 241 , Oxford). Boessneck, J., 1969. ‘Osteological Differences between Sheep (Ovis aries Linné) and Goat (Capra hircus Linné)’, in D. Brothwell, E. Higgs and G. Clark (eds.), Science in Archaeology (2nd edition, London), 331–58. Boon, G.C., 1982. Report on the Excavations at Usk 1965–1976: The Coins, Inscriptions and Graffiti (University of Wales Press, Cardiff). Borrill, H., 1981. ‘Casket burials’, in C. Partridge, Skeleton Green: a late iron age and Romano-British site (Britannia Monograph 2), 304–21. Bowden, M., 2005. The Malvern Hills: An ancient landscape (London). Brodribb, A.C.C., Hands, A.R., and Walker, D.R., 1973. Excavations at Shakenoak Farm, near Wilcote, Oxfordshire IV (Oxford). Brothwell, D., 1981. Digging up Bones (Oxford University Press). Buikstra, J., and Ubelaker, D., 1994 . Standards for Data Collection from Human Skeletal Remains (Arkansas Archaeol. Rep. Research Series 44 ). Burnham, B.C., and Wacher, J., 1990. The ‘Small Towns’ of Roman Britain (University of California Press). Buteux, V., 1996. ‘Archaeological Assessment of Stretton Grandison Roman Settlement, Hereford and Worcester’ (Central Marches Historic Towns Survey, typescript rep.).

Campbell, G., 1998. ‘The charred plant remains’, in A. Boyle and R. Early, Excavations at Springhead Roman Town, Southfleet, Kent (Oxford Archaeol. Unit Occasional Paper 1), 36–9. Caseldine, A., 1990. Environmental Archaeology in Wales (Saint David’s University College Lampeter). Catchpole, T., 2000. ‘Archaeological Excavations at Dymock Sewage Treatment Works: Post-Excavation Assessment’ (Glos. co. council typescript rep.). Catchpole, T., 2002. ‘Excavations at West Drive, Cheltenham, Gloucestershire, 1997–9’, Trans.B.G.A.S. 120 , 89–101. Challinor, D., forthcoming. ‘The wood charcoal’, in P. Booth, A. Bingham and S. Lawrence, The Roman Roadside Settlement at Westhawk Farm, Ashford, Kent: excavations 1998–9 (Oxford Archaeol. Mono- graph 1). Children,G., and Nash,G., 1994. Prehistoric Sites of Herefordshire (Almeley). Cleere, H., and Crossley, D., 1985. The Iron Industry of the Weald (Leicester University Press). Cool, H.E.M., 1991. ‘Roman metal hair pins from southern Britain’, Archaeol. Jnl. 147 , (1990), 148–82. Cool, H.E.M., 2004. ‘Some notes on spoons and mortaria’, in B. Croxford, H. Eckardt, J. Meake and J. Weekes (eds.), TRAC 2003: Proceedings of the Thirteenth Annual Theoretical Roman Archaeol. Conference Leicester 2003 (Oxford), 28–35. Cool, H.E.M., 2007. ‘The small finds in their regional context’, in D. Miles, S. Palmer, A. Smith and G.P. Jones, Iron Age and Roman settlement in the Upper Thames Valley: excavations at Claydon Pike and other sites within the Cotswold Water Park (Oxford Archaeol. Thames Valley Landscapes Monograph 26 ), 342–50. Cool, H.E.M., and Baxter, M.J., 1999. ‘ “Peeling the onion”; an approach to comparing vessel glass assem - blages’, Jnl. Roman Archaeol. 12 , 72–100. Cool, H.E.M., and Booth, P., 2007. ‘Small finds for Pepper Hill Roman Cemetery’ (CTRL specialist report series in ADS 2007: CTRL digital archive, Archaeology Data Service, http://ads.ahds.ac.uk/ catalogue/projArch/ctrl) 04_BGAS125_131-246 16/11/07 15:50 Page 241

ROMAN DYMOCK: BIBLIOGRAPHY 241

Crummy, N., 1983. The Roman Small Finds from Excavations in Colchester 1971–9 (Colchester Archaeol. Rep. 2). Crummy, N., and Eckardt, H., 2004. ‘Regional identities and technologies of self: nail-cleaners in Roman Britain’, Archaeol. Jnl. 160 (2003), 44–69.

Dalwood, H., and Edwards, R., 2004. Excavations at Deansway, Worcester 1988–89: Romano-British small town to late medieval city (CBA Research Rep. 139 ). Darvill,T., 2003. ‘The Land of the Dobunni’, in M. Ecclestone, K.S. Gardner, N. Holbrook and A. Smith (eds.), The Land of the Dobunni (Gloucester), 2–11. Derham, K., 2001. ‘An archaeological evaluation of the proposed car park on land behind the Old Forge Garage, Dymock’ (Glos. co. council unpublished rep.). Driesch, A. von den, 1976. A Guide to the Measurement of Animal Bones from Archaeological Sites (Peabody Museum Bulletin 1, Harvard Univ., Cambridge, Mass.).

Edlin, H.L., 1949. Woodland crafts in Britain: an account of the traditional uses of trees and timbers in the British countryside (London). Emmanuel, L., 1994. ‘Dymock Sewage Treatment Works, Dymock: An Archaeological Evaluation’ (Glos. co. council typescript rep.). Enright, D., and Watts, M., 2002. A Romano-British and Medieval Settlement at Stoke Road, Bishop’s Cleeve, Gloucestershire (Bristol and Glos. Archaeol. Rep. 1, Cotswold Archaeol.). Erskine, J., 1996. ‘Newent Business Park, Gloucester Road, Newent, Glos. An Archaeological Excavation September 1994’ (Glos. co. council typescript rep.). Evans, J., 2001. ‘Material approaches to the identification of different Romano-British site types’, in S. James and M. Millett, Britons and Romans: Advancing an Archaeological Agenda (CBA Research Rep. 125 ), 26–35.

Feugère, M., 1985. Les Fibules en Gaule Méridionale de la Conquête à la Fin du Ve Siècle après J.-C . (Revue Archéologique de Narbonnaise Supplément 12 , Paris). Figueiral, I., 1992. ‘The fuels’, in Fulford and Allen, 159–208. Finberg, H.P.R., 1957. ‘The Genesis of the Gloucestershire Towns’, Gloucestershire Studies (Leicester University Press). Finch Smith, R., 1987. Roadside Settlements in Lowland Britain: A Gazetteer and Study of their Origins, Growth and Decline, Property Boundaries and Cemeteries (BAR British Series 157 , Oxford). Fitter, R., Fitter, A., and Farrer, A., 1987. Grasses, Sedges, Rushes and Ferns of Britain and Northern Europe (Collins Pocket Guides). Fowler, E., 1960. ‘The origins and development of the penannular brooch in Europe’, Proc. Prehistoric Soc. 26 , 149–77. Frere, S.S., 1987. ‘Brandon Camp, Herefordshire’, Britannia 18 , 49–92. Frere, S.S., and St Joseph, J.K., 1974. ‘The Roman Fortress at Longthorpe’, Britannia 5, 1–129. Fulford, M.G., and Allen, J.R.L., 1992. ‘Iron-making at the Chesters villa, Woolaston, Glos.: survey and excavations 1987–91’, Britannia 23 , 159–215. Fulford, M., 2003. ‘The Canton of the Dobunni’, in M. Ecclestone, K.S. Gardner, N. Holbrook and A. Smith (eds.), The Land of the Dobunni (Gloucester), 17–23.

Gale, R., 1999. ‘Charcoal [from Pomeroy Wood]’, in A.P. Fitzpatrick, C.A. Butterworth and J. Grove (eds.), Prehistoric and Roman Sites in East Devon: The A30 Honiton to Exeter Improvement DBFO, 1996–9. Volume 2: Romano-British Sites (Wessex Archaeol. Rep. 16 , Salisbury), 372–82. Gale, R., 2000. ‘The fuels’, in Barber and Holbrook, 51–4. Gale, R., 2003a. ‘Wood-based industrial fuels and their environmental impact in lowland Britain’, in P. Murphy and P. Wiltshire (eds.), The Environmental Archaeology of Industry (Symposia of the Assoc. for Environmental Archaeol. 20 , Oxford), 30–47. Gale, R., 2003b. ‘The identification of charcoal from Bridgewater’s 1963 excavation’, in R. Jackson, 151–6, 181. 04_BGAS125_131-246 16/11/07 15:50 Page 242

242 ROMAN DYMOCK: BIBLIOGRAPHY

Gale, R., and Cutler, D., 2000. Plants in Archaeology (Westbury and Royal Botanic Gardens, Kew). Gethyn-Jones, J.E., 1966. Dymock Down the Ages (Gloucester). Gethyn-Jones, J.E., 1991. ‘Roman Dymock–a personal review’, Trans B.G.A.S. 109 , 91–8. Getty, R. (ed.), 1975. Sisson and Grossman’s The Anatomy of the Domestic Animals (5th edition). Godwin, H., 1956. The History of the British Flora (Cambridge University Press). Grant, A., 1982. ‘The use of tooth wear as a guide to the age of domestic ungulates’, in W. Wilson, C. Grigson and S. Payne (eds.), Ageing and Sexing Animal Bones from Archaeological Sites (BAR British Series 109 ), 91–108. Gwatkin, G., 1992. ‘Dymock 1847: copy of tithe map’.

Hammer, F., 2003. Industry in north-west Roman Southwark. Excavations 1984–8 (Museum of London Archaeol. Services Monograph 17) . Hamshaw-Thomas, J., 2000. ‘When in Britain do as the Britons: dietary identity in early Roman Britain’, in P. Rowley-Conwy (ed.), Animal Bones, Human Societies (Oxford), 166–9. Hattatt, R., 1985. Iron Age and Roman Brooches (Oxford). Hattatt, R., 1987. Brooches of Antiquity (Oxford). Hawkes, C.F.C., and Hull, M.R., 1947. Camulodunum. First report on the excavations at Colchester 1930–1939 (Soc. Antiq. London). Hicks, A., Bendrey, R., Bird, J., Cool, H., Crummy, N., Dickinson, B., Harrison, L., MacDonald, P., Mackreth, D., Manning, W., Pelling, R., Riddler, I., Savage, A., and Taylor, M., forthcoming. ‘The Roman Settlement’, in P. Clark (ed.), Monkton–Mount Pleasant: Prehistoric, Roman and medieval discover - ies on the Isle of Thanet 1994–1995. Hillson, S., 1996. Dental anthropology (Cambridge University Press). Holbrook, N., 2006. ‘The Roman Period’, in N. Holbrook and J. Ju ica (eds.), Twenty-Five Years of Archaeology in Gloucestershire (Bristol and Glos. Archaeol. Rep. 3, Cotswold Archaeol.). Holman, J.A., 1985. ‘Herpetofauna of the Late Pleistocene fissures near Ightham, Kent’, Herpetological Jnl. 1, 26–32. Horne, L., 1982. ‘Fuel for the metal worker’, Expedition 25 (University of Pennsylvania Museum), 6–13. Hoyle, J., Butler, L., Tait, G., and Wootton, D., 2004. ‘The Forest of Dean, Glos. The Scowles and Associated Iron Industry Survey’ (Glos. co. council typescript rep.: English Heritage ALSF Project 3342).

Iliffe, J., 1932. ‘Excavations at Alchester 1928’, Antiq. Jnl. 12 , 35–67. Ingrem, C., n.d. ‘Animal bone from the Late Romano-British pits and wells at Insula IX, Silchester’ (Centre for Applied Archaeol. Analyses, University of Southampton, unpublished rep.). Ireland, C., 1983. ‘The Roman pottery’, in C. Heighway, The East and North Gates, of Gloucester (Western Archaeol Trust). Isings, C., 1957. Roman Glass from Dated Finds (Groningen).

Jackson, R., 2003. ‘Ariconium, Herefs.: An assessment and synthesis of the evidence’ (draft rep.). Jones, J., 2000. ‘Assessment of bulk samples from Roman features at Dymock Sewage Treatment Works, Dymock’ (Glos. co. council archaeol. service rep.). Jones, M., 1978. ‘The Plant Remains’, in M. Parrington (ed.), The excavation of an Iron-Age Settlement, Bronze-Age Ring Ditches and Roman Features at Ashville Trading Estate, Abingdon, Oxon. 1974–6 (CBA Research Rep. 28 ), 93–110. Jones, G.D.B., and Mattingly, D.J., 1990. An Atlas of Roman Britain (Oxford). Ju ica, J., forthcoming. ‘Dymock’, in A History of the County of Gloucester XII: Newent and May Hill (Victoria History of England).

King, A., 1978. ‘A comparative survey of bone assemblages from Roman sites in Britain’, Bulletin of the Inst. of Archaeol. 15 , 207–32. King, A., 1991. ‘Food Production and Consumption – Meat’, in R.F.J. Jones (ed.), Britain in the Roman Period: Recent Trends (University of Sheffield), 15–20. 04_BGAS125_131-246 16/11/07 15:50 Page 243

ROMAN DYMOCK: BIBLIOGRAPHY 243

Knorr, R., 1919. Töpfer und Fabriken verzierter Terra Sigillata des ersten Jahrhunderts (Stuttgart). Knorr, R., 1952. Terra-Sigillata-Gefässe des ersten Jahrhunderts mit Töpfernamen (Stuttgart).

Lawson, G., 1990. ‘Jew’s harp’, in M. Biddle, Object and Economy in Medieval Winchester (Winchester Studies 7.ii , Oxford), 724–5. Leech R., 1981. Historic Towns in Gloucestershire (Committee for Rescue Archaeol. in Avon, Glos. and Somerset Survey 3). Legge, A.J., 1992. Excavations at Grimes Graves, Norfolk 1972–1976. Fasicule 4. Animals, Environment and the Bronze Age Economy (British Museum Press). Levine, M.A., 1982. ‘The Use of Crown Height Measurements and Eruption-Wear Sequences To Age Horse Teeth’, in B. Wilson, C. Grigson and S. Payne (eds.), Ageing and Sexing Animal Bones from Archaeological Sites (BAR British Series 109 , Oxford), 223–50. Leyell, A.H., 1923. ‘Charcoal identifications’, in G.H. Jack, ‘Excavations on the site of Ariconium, a Romano-British smelting town in the parish of Weston-under-Penyard, South Herefordshire’, Trans. Woolhope Naturalists’ Field Club , 31. Lyman, R.L., 1994. Vertebrate Taphonomy (Cambridge University Press).

Mackreth, D. F., 1982. ‘Two brooches from Stonea, Cambs. and Bicester, Oxon, and the origin of the Aesica brooch’, Britannia 13 , 310–15. Mackreth, D., 1998. ‘Copper-alloy brooches (Site 1, Site 2 and field-walking)’, in J.R. Timby, Excavations at Kingscote and Wycomb, Gloucestershire: a Roman estate centre and small town in the Cotswolds with notes on related settlements (Cotswold Archaeol. Trust), 113–49. Mackreth, D., 2000. ‘The Romano-British brooches’, in P. Ellis (ed.), The Roman Baths and Macellum at Wroxeter: Excavations by Graham Webster 1955–85 (English Heritage Archaeol. Rep. 9), 144–59. Maltby, M., 1979. Faunal Studies on Urban Sites: the Animal Bones from Exeter 1971–1975 (University of Sheffield, Exeter Archaeol. Report 2). Maltby, M., 1981. ‘Iron Age, Romano-British and Anglo-Saxon animal husbandry – A review of the faunal evidence’, in Jones, G., The Environment of Man: the Iron Age to the Anglo Saxon period (BAR British Series 87 ), 155–203. Maltby, M., 1985. ‘Patterns in faunal assemblage variability’, in G. Barker and C. Gamble (eds.), Beyond Domestication in Prehistoric Europe (London, Academic Press), 22–74. Maltby, M., 1994. ‘The meat supply in Roman Dorchester and Winchester’, in A.R. Hall and H.K. Kenward, Urban-rural Connexions: Perspectives from Environmental Archaeology .(Symposia of the Assoc. for Environmental Archaeol. 12 , Oxford),85–102. Malvern Research Group, 1958. ‘Dymock’ in ‘West Midlands Annual Archaeol. News Sheet’ (unpublished newletter held in Glos, SMR file) 1, 5. Manning, W.H., 1985a. ‘The iron objects’, in L.F. Pitts and J.K. St Joseph, Inchtuthil: the Roman legionary fortress excavations 1952–65 (Britannia Monograph 6), 289–99. Manning, W.H., 1985b. Catalogue of the Romano-British Iron Tools, Fittings and Weapons in the British Museum (London). Manning, W.H., Price, J., and Webster, J., 1995. Report on the Excavations at Usk 1965–1976 VII: The Roman Small Finds (University of Wales Press, Cardiff). Margary, I.D., 1957. Roman Roads in Britain II, North of the Fosse Way–Bristol Channel (London) Maxwell, A., 2004. ‘Roman Dymock’ (University of Bristol BA dissertation). Mayet, F., Schmitt, A., and Tavares da Silva, C., 1996. Les Amphores du Sado (Paris). Mays, S., 1998. The Archaeology of Human Bones (London). Mays, S., 1995. ‘Killing the unwanted child’, Brit. Archaeol. 2, 8–9. McCarthy, M.R., Padley, T.G., and Henig, M., 1982. ‘Excavations and Finds from the Lanes, Carlisle’, Britannia 13 , 79–89. McWhirr, A., 1981. Roman Gloucestershire (Gloucester). Millett, M., 1990. The Romanization of Britain: An Essay in Archaeological Interpretation (Cambridge University Press). 04_BGAS125_131-246 16/11/07 15:50 Page 244

244 ROMAN DYMOCK: BIBLIOGRAPHY

Mitchell, A., 1974. Trees of Britain and Northern Europe (Collins Field Guide). Molleson, T. and Cox, M., 1993. The Spitalfields Project: Volume 2, The middling sort (CBA Research Rep. 86 , London). Murphy, P., 2001. ‘Review of Wood and Macroscopic Wood Charcoal from Archaeological Sites in the West and East Midlands Regions and the East of England’ (Centre for Archaeology Rep. (unpublished) 23/2001, English Heritage).

Newman,C., and Newman, R., 1990. ‘Watching brief and salvage work at the new cricket pavilion, Dymock’, Glevensis 24 , 22–6.

OA 2003. ‘Land Adjacent to the Rectory, Dymock, Gloucestershire: Archaeological Interim Report’ (Oxford Archaeol. unpublished rep.). OAU 2001. ‘Land adjacent to the Rectory, Dymock, Gloucestershire. Archaeological Evaluation Report’ (Oxford Archaeol. Unit unpublished rep.). Oswald, F., 1936–7. Index of Figure Types on Terra Sigillata ‘Samian Ware’ (Liverpool University Press).

Parry, C., 1994. ‘Land adjacent to Winserdine and Rose Cottage Dymock, Glos. Archaeological evaluation 1993’ (Glos. co. council typescript rep.). Parry, C., 1999. ‘Iron-Age, Romano-British and Medieval Occupation at Bishop’s Cleeve, Gloucestershire: excavations at Gilder’s Paddock 1989 and 1990–1’, Trans. B.G.A.S. 117 , 89–118. Payne, S., 1973. ‘Kill-off patterns in sheep and goats: the mandibles from Asvan Kale’, Anatolian Studies 22 (1973), 281–303. Payne, S., 1985. ‘Morphological Distinctions between the Mandibular Teeth of Young Sheep (Ovis), and Goats (Capra)’, Jnl. Archaeol. Science 12 , 139–47. Peacock, D.P.S., 1968. ‘A petrological study of certain Iron Age pottery from western England’, Proc. Prehist. Soc . 34 , 415–19. Peacock, D.P.S., and Williams, D.F., 1986. Amphorae and the Roman Economy (London). Pelling, R., 1999. ‘Charred and Waterlogged Plant remains’ in A. Mudd, R. J. Williams and A. Lupton, Excavations alongside Roman Ermin Street, Gloucestershire and Wiltshire. Volume 2: Medieval and and post- medieval activity, finds and environmental evidence (Oxford Archaeol. Unit), 479–90. Perring, D., 1987. ‘Domestic Buildings in Romano-British towns’, in J. Schofield and R. Leech (eds.), Urban Archaeology in Britain (CBA Research Rep. 61 ), 147–55. Philpott, R., 1991. Burial Practices in Roman Britain (BAR British Series 219 , Oxford). Price, J., and Cottam, S., 1998. Romano-British Glass Vessels: A Handbook (CBA Practical Handbook in Archaeol. 14 ).

Ratkai, S., 2001. ‘Assessment of the pottery and other ceramic materials’, in N. Tavener. Reece, R., 1985. ‘Coins’, in H.R. Hurst, Kingsholm (Glouc. Archaeol. Rep. 1), 22–3. Reece, R., 1998. ‘The excavated coins considered as groups’, in N. Holbrook (ed.), Cirencester: The Roman Town Defences, Public Buildings and Shops (Cirencester Excavations V, Cirencester), 247–62. Rigby, V., 1982. ‘The Pottery’, in J. Wacher and A. McWhirr, Early Roman Occupation of Cirencester (Cirencester Excavations I, Cirencester Excavation Committee.), 118–98. Rivet, A.L.F., and Smith, C., 1979. The Place-Names of Roman Britain (London). Robinson, M., and Straker, V., 1991. ‘Silica skeletons of macroscopic plant remains from ash’, in J.M. Renfrew (ed.), New Light on Early Farming. Recent Developments in Palaeoethnobotany (Edinburgh University Press), 3–13. Rogers, G.B., 1974. Poteries Sigillées de la Gaule Centrale I: les motifs non figurés (Gallia Supplement 28 , Paris).

Seager-Smith, R., 2001. ‘The coarse pottery’, in A.S. Anderson, J.S. Wacher and A.P. Fitzpatrick, The Romano-British ‘Small Town’ at Wanborough, Wiltshire (Britannia Monograph 19 ), 232–300. Serjeantson, D., 1996. ‘The Animal Bones’, in S. Needham and T. Spence, Runnymede Bridge Research Excavations Vol. 2: Refuse and Disposal at Area 16 East Runnymede (British Museum Press), 194–223. 04_BGAS125_131-246 16/11/07 15:50 Page 245

ROMAN DYMOCK: BIBLIOGRAPHY 245

Smith, A.H., 1964. The Place-Names of Gloucestershire Part III (English Place-Name Soc. 40 ). Smith, A.H., 1965. The Place-Names of Gloucestershire Part IV (English Place-Name Soc. 41 ). Stace, C., 1997. New Flora of the British Isles (Cambridge University Press). Stanfield, J.A., and Simpson, G., 1958. Central Gaulish Potter s (London). Stanford, S.C., 1991. The Archaeology of the Welsh Marches (Ludlow). Stead, I.M., and Rigby, V., 1989. Verulamium: the King Harry Lane site (English Heritage Archaeol. Rep. 12 ).

Tavener, N., 2001. ‘Land adjacent to Rose Cottage and “Winserdene”, Dymock, Gloucestershire’ (Marches Archaeol. unpublished rep. 216). Thomas, A., Holbrook, N., and Bateman C., 2003. Later Prehistoric and Romano-British Settlement at Hucclecote, Glos. (Bristol and Glos. Archaeol. Rep. 2., Cotswold Archaeol.) Timby, J.R., 1990. ‘Severn Valley wares: a reassessment’, Britannia 21 , 243–51. Timby, J.R., 2001. ‘A reappraisal of Savernake Ware’, in P. Ellis (ed.), Roman Wiltshire and After: Papers in Honour of Ken Annable (Wilts. Archaeol. and Nat. Hist. Soc). Tomber, R.S., and Dore, J., 1998. The National Roman Fabric Reference Collection: a Handbook (London). Trotter, R., and Gleser, G.C., 1952. ‘Estimation of stature from long bones of American Whites and Negroes’, Amer. Jnl. Phys. Anthrop . new series 10 , 463–514. Tutin, T.G., Heywood, V.H., et al . (eds.), 1964–80. Flora Europaea 1–5 (Cambridge University Press).

Ubelaker, D.H., 1978. Human Skeletal Remains (Washington).

Wacher, J., 1998. Roman Britain (Sutton Publishing). Walker, D.R., 1988. ‘The Roman Coins’, in B. Cunliffe (ed.), The Temple of Sulis Minerva at Bath 2: The finds from the sacred spring (Oxford University Committee for Archaeol.), 281–358. Walker, G., Thomas A., and Bateman, C., 2004. ‘Bronze-Age and Romano-British Sites South-East of Tewkesbury: evaluations and excavations 1991–7’, Trans. B.G.A.S. 122 , 29–94. Walton Rogers, P., 1997. Textile Production at 16–22 Coppergate (The Archaeol. of York 17.11 ). Waters, P.L., 1960. ‘Dymock’, in ‘West Midlands Annual Archaeol. News Sheet’ (unpublished newsletter held in Glos. SMR file) 3, 7. Waters, P.L., 1965. ‘Dymock’, in ibid. 8, 11. Waters, P.L., 1966. ‘Dymock’, in ibid. 9, 9–10. Waters, P.L., 1967. ‘Dymock’, in ibid. 10 , 17. Waters, P.L., 1969. ‘Dymock’, Glevensis 4, 9. Waters, P.L., 1970. ‘Dymock’, in ‘West Midlands Annual Archaeol. News Sheet’ 13 , 38. Waters, P.L., 1971. ‘Dymock’, in ibid. 14 , 10. Waters, P.L., 1972. ‘Dymock’, in ibid. 15 . Webb, P.A., and Suchey, J.M., 1985. ‘Epiphyseal union of the anterior iliac crest and medial clavicle in a modern multiracial sample of American males and females’, Amer. Jnl. Phys. Anthrop. 68 , 457–66. Webster, G., 1993. Rome against Caratacus: The Roman Campaigns in Britain (Routledge). Wild, F., 1985. ‘Samian Ware’, in H.R. Hurst, Kingsholm (Glouc. Archaeol. Rep. 1), 56–67. Worssam, B.C., Ellison, R.A. and Moorlock, B.S.P., 1989. The Geology of the Country around Tewkesbury (British Geological Survey, London).

Young, C.J., 2000. Oxfordshire Roman Pottery (BAR British Series 43 , Oxford).