Nea Farm, Phase GP9, Somerley, ,

An Archaeological Excavation Draft Publication Report

for Tarmac Southern Ltd

by Andy Taylor and Richard Tabor

Thames Valley Archaeological Services Ltd

Site Code SOM11/72

March 2017

Summary

Site name: Nea Farm, Somerley, Ringwood, Hampshire Phase GP9

Grid reference: SU 1310 0870

Site activity: Excavation

Date and duration of project: 1st August–10th November 2011

Project manager: Steve Ford

Site supervisor: Andy Taylor

Site code: SOM 11/72

Area of site: 5.5ha

Summary of results: The fieldwork revealed an extensive complex of archaeological deposits spanning several periods. The earliest datable feature was an Early Bronze Age Beaker cremation burial. Three pits may be datable by Bronze Age pottery in their fills and other pits with undiagnostic prehistoric pottery may have been of the same period. By the Iron Age in the 4th to 3rd century BC a track linking the plateau to the Avon Valley had become the focus for settlement and a field system which remained in use with modification into the Late Iron Age. Many undated pits are likely to be of this period. The evidence of the Roman pottery suggests that occupation of the area ceased at or prior to the Conquest but resumed during the mid to late Roman period. Roman activity was represented by pits and postholes as well as ditches and gullies of badly truncated fields and enclosures as well as ovens, two stone-built corn-dryers and a minimum of three graves. After a further hiatus a field system and settlement were laid out during the late 11th to 12th century, probably lasting until the 13th or 14th century, by which time there are signs of contraction and less intensive land use. Five timber buildings dated only broadly to the medieval period were also recorded.

This report may be copied for bona fide research or planning purposes without the explicit permission of the copyright holder

Report edited/checked by: Steve Ford 29.03.17 Steve Preston23.03.17

i Thames Valley Archaeological Services Ltd, 47–49 De Beauvoir Road, Reading RG1 5NR Tel. (0118) 926 0552; Fax (0118) 926 0553; email [email protected]; website: www.tvas.co.uk

Nea Farm, Phase GP9, Somerley, Ringwood, Hampshire An Archaeological Excavation

by Andy Taylor and Richard Tabor with contributions by Ceri Falys, Steve Ford, Rosalind McKenna, Danielle Milbank, Susan Porter, Jane Timby and David Williams Report 11/72b

Introduction

An archaeological excavation was carried out by Thames Valley Archaeological Services at Nea Farm,

Somerley, Ringwood, Hampshire (SU 1240 0860) (Fig. 1). The work was commissioned by Mr Andy Cadell of

Tarmac Southern Limited, Stancombe Quarry, Stancombe Lane, Flax Bourton, Bristol, BS48 3QD. Planning consent had been obtained from Hampshire County Council to extract gravel from the area as part of an ongoing extraction programme. The excavation was carried out to a specification approved by Ms Hannah Fluck, Senior

Archaeologist with Hampshire County Council. The site code is SOM 11/72. This is one of a long series of archaeological investigations on this quarry site (see below). The phase of fieldwork to which this report refers took place between 1st August and 10th November 2011. The archive is currently held by Thames Valley

Archaeological Services in Reading, and will be deposited in due course with Hampshire Cultural Trust.

Site location, topography and geology Nea Farm, Somerley, lies to the north-west of Ringwood on the eastern margins of Ringwood Forest (Fig. 1).

The geology consists of plateau gravel capped with brickearth underlying slightly acid, free-draining, loamy soils of low fertility (BGS 1990; CSAI 2017). The site lies at a height of 45m to 48m above Ordnance Datum

(aOD) on the eastern edge of the plateau overlooking the River Avon which meanders from north to south 800m to the east. From the plateau edge the river’s wooded valley side falls steeply to 20m aOD over a distance of less than 300m. A tributary stream lies 400m to the north of the site (Fig. 2). The current phase of the extraction site comprised two areas totalling 5.5 hectares, separated by one of the estate trackways.

Archaeological background The phased extraction of sand and gravel has been preceded by a number of archaeological investigations initiated as part of the planning process (Ford 1992; 2001a; 2001b; Cass 2008; Anthony 2002; Anthony and

Ford 2003; Ford and Hall 1993; Oram 2004; Pine 2003; Pine and Bennett 2010; Platt 2010 and 2011; Smith

1996; Taylor 2008 and 2009; Weaver 1995) (Fig. 2). A preliminary study included minimally invasive fieldwork comprising fieldwalking, test pitting and geophysical survey. Later works comprised open area excavations.

1 The preliminary fieldwork (Weaver 1995) is particularly relevant for this phase of fieldwork (GP8) as it partly covered this specific part of the quarry complex. A small area of archaeological potential was defined in what would become the northern extent of the area covered in this report (at that time the southern part of an area referred to as ‘Field 4’), on the basis of fieldwalking and a dense concentration of Iron Age or Roman features in a cluster of six evaluation trenches. Two smaller areas of potential within the then ‘Field 4’ were excavated (Smith 1996) but the present area was not scheduled for extraction until now so the third area of potential was not explored until the current phase of works. The southern part of the current area had not been evaluated and so its potential was unknown.

To the south and west of this phase of extraction a complex of medieval boundary features were identified in excavation phase GP3–GP5 and phase 5 areas (Cass 2008; Oram 2004; Taylor 2008 and 2009), of which

Phase 5 lay immediately adjacent to the current area, to the south. Prehistoric features included an early Bronze

Age Beaker with barbed and tanged arrowhead from GP3 and, closer to the south-west in extraction area GP5, three pits. A radiocarbon determination of 3517–3393 cal BC (KIA39673) indicated an Early Neolithic date for one, whilst Beaker pottery from a second was consistent with an Early Bronze Age radiocarbon determination of

2134–2078 cal BC (KIA39674); and the third was of Middle Bronze Age or Iron Age date.

Nea Farm, though, is chiefly notable for its Upper Palaeolithic site (SOM 01/41: Barton et al. 2009) which lies c.150m to the west.

The Excavation (Figs 3–6)

Two entire areas for extraction in this phase were stripped (Fig. 3) of topsoil and overburden under constant archaeological supervision using a 360º type machine fitted with a toothless grading bucket. About a quarter of a hectare in the north-east corner had previously been truncated, accounting for the absence of features in what would surely have been an archaeologically dense area (Fig. 3).

For many features in this phase of the work, there is little or no dating evidence either in the form of associated artefacts or of stratigraphy. Of these, some have morphological characteristics which are likely to exclude some periods or, in the cases of some ditches and gullies, may share a similar orientation with one or more other linear features.

All archaeological features and deposits are listed in Appendix 1. The evidence for dating is given; this may be datable artefacts, stratigraphic relationships or a perceived association with other datable features. On a site occupied over a long period it is inevitable that some finds from earlier periods are incorporated into later deposits. Conversely, later activity, particularly cultivation, may introduce later finds into earlier deposits. The

2 greatest problem for dating on the present site is that the majority of discrete features, pits and post holes, contained no datable finds. In the phased plans below undatable features are represented in grey for those periods during which their stratigraphic positions would allow them to have been in use. Where features have been allocated to a particular phase with a moderately high or high level of confidence they are marked in solid red; in instances where features are deemed likely to have belonged to one of two periods they are shown in both phase plans in green.

There was variation in the size, shape and fill character of the many discrete cut features which has influenced interpretation of them. Where possible datable pits have been classified in terms of the character and rate of their fills, bearing in mind the issues raised by Lock (1995) but with a simplified classificatory system based on deposit formation. Group 1 comprises pits where one or more fills are judged to be entirely of slow to moderate formation. The implication is that natural process of erosion have washed or blown in the surrounding soil or natural geology. Group 2 are pits with one deposit which was made rapidly. This includes naturally induced collapsing of the pit’s sides but also the introduction of a single deposit by human agency. This may be a simple backfilling of the entire pit by natural gravel, soil, a mixture of the two, or a deposit notably rich in humus and/or burnt material. In some instances it may be the introduction of one or more substantial artefacts

(i.e. too large or too many to have been introduced by natural agency) in a pit otherwise filled by gradual natural processes. Probably the most common rapid deposit-type in pits of this group is localized introduction of humic material on the base, in the middle or more rarely on the top of a sequence of gradual erosion deposits. Group 3 at its simplest may include a single humic and a single collapse deposit with the remaining infilling occurring by gradual natural agency. A pit may include exclusively multiple rapid fills but more commonly they are interspersed with gradual erosion deposits.

Earlier Prehistoric

Characteristic worked flint demonstrated episodes of activity on the site during the Upper Palaeolithic,

Mesolithic and Early Neolithic periods but the material was derived entirely from either unstratified or residual contexts (Ford, below).

Phase 1, Bronze Age (Figs 7 and 8)

Few conclusions can be drawn concerning the distribution of pre-Iron Age features as there is no clear clustering. It is possible that a west to east oriented droveway had Bronze Age origins as a route across the

3 plateau giving access to the Avon valley and that it played a part in the siting of the Iron Age settlement. This

appears to have developed in a triangular space defined by the upper edge of the valley side and boundaries

based on the droveway to the north and on what may have been a newly constructed fence and ditch line on the

west side. There is a relative dearth of features north and west of these boundaries both in the area stripped and

further afield in the same directions in evaluation trenches (Fig. 2).

The only well-dated Bronze Age features were a disturbed cremation pit (39) dated by Beaker pottery (Pl.

1), pit 36 in the central northern area (Fig. 5), and pit 402 in the extreme north-east of the site (Fig. 4), both

dated by significant amounts of pottery (Table 1). These were all small, heavily truncated features with

diameters of between 0.25m and 0.60m and maximum depths of 0.18m. A much larger pit, 41, is more

problematic by virtue of its size of at least 1.10m in diameter and depth of 0.58m, as well as the dark brown

sandy loam initial fill (93). The overlying fill (98) comprised similar loam but with more sand and was probably

of slower formation. The lower fill included a large plain wall sherd, probably from a Bronze Age jar, but a

single small Iron Age sherd was recovered in the fill above it (98), which was cut by pit 43 which included

much of a Middle Iron Age bowl (Fig. 23; 16). The slab of jar was in fresh condition, implying that it was close

to, or in, its place of primary deposition but in all other respects the pit size and fills are analogous to those of

the Iron Age elsewhere on the site. It may be that the Iron Age sherd in fill 98 really derives from pit 43. Small

amounts of undifferentiated prehistoric pottery were found in pit 113, with dimensions fitting well with the

better established Bronze Age features and which was cut by a later Iron Age pit 112, and in pit 245. The size of

the latter is comparable with 41.

Table 1: Summary of Bronze Age pits

Cut Fills Form Length Width Depth Dating evidence Other finds (m) /diameter (m) (m) 36 88, 89 Concave bowl 0.52+ 0.18 Bronze Age pottery* 39 91, 95, 96 Flat bowl 0.28 0.10 Beaker pottery* Burnt bone 41 93 Bowl or cylinder 1.10+ 0.58 Bronze Age pottery 113 173 Flat bowl 0.64 0.63 0.17 Prehistoric pottery 115 175 Concave bowl 0.65 0.13 Prehistoric pottery 245 392, 393, 394, 395 Flat cylinder 1.02 0.68 Prehistoric pottery loomweight 402 552 Flat bowl 0.58 0.56 0.18 Bronze Age pottery 830 1256 Concave bowl 0.91 0.14 Prehistoric pottery 1424 1989, 1990, 1991, 1992 Concave bowl 1.60 0.97 Prehistoric pottery (* Includes illustrated sherds)

The ditches forming S1 could perhaps have origins in this period, but they are more probably part of phase 2.

Phase 2, Middle Iron Age (Figs 8-12)

The earliest widely distributed pottery from the site is datable to the 4th and 3rd centuries BC. Most of this

material was recovered from pits, some of which had sherds of this date in the lower fills but which contained

4 sherds in later styles in the upper fills. In some instances it was clear that the upper fills were within re-cuts into an earlier pit. There were also sporadic substantial pottery deposits in linear features.

Most of the linear features allocated to this phase are only tentatively dated. They have been allocated to one of two system groups based on orientation, relationships and the few datable finds. The dating of S1 depends on the total of 14 earlier Middle Iron Age sherds from west to east oriented gully 2151, and 15 broadly

Iron Age sherds in short ditch 2193. A length of 24m of gully 2151 lay within the excavated area, including a terminus at its eastern end. It was around 0.80m wide and varied in depth from as little as 0.08m rising to 0.30m deep. There was no discernible relationship with parallel gully 2150 but it was cut by late Roman ditch 2152. It seems likely that 2151 represents a droveway, S1, possibly in association with the sinuous gully 2199, which produced some burnt flint and no pottery but was cut by ditch 2252. Approximately 90m of 2199 was exposed, its variation in width from 0.21m to 0.62m and depth from 0.07m to 0.30m across 9 slots probably accounted for by erosion.

One short ditch, 2193, was cut by later Iron Age gullies 2192 and 2195 (Fig. 4) and contained 15sherds of pottery only broadly datable to the Iron Age. There were several linear features which lacked finds and for which no relationships with datable features could be established. A few of these had orientations which were comparable with those of datable features or complexes and a few others appeared to have been replaced by later linear ditches. This last group includes the broadly west to east oriented gullies 2197, 2198, 2252 and 2254 and the south east to north west oriented 2261 and 2265, all in the central and western areas of the site’s north.

These features have been treated as traces of a system of access or fields, S1, which appears to have remained in use during later phases.

Three slots across gully 2197 were up to 0.70m wide and varied in depth from 0.07m to 0.31m. The position of a short length of gully with two surviving termini, 2198, suggests that it may have been associated with it. The gully was around 0.27m wide and 0.07m deep. Gully 2250 may have been a continuation of 2197. It had a terminus at its western end but the eastern end was interrupted by a treebole. It was between 0.40m and

0.53m wide and between 0.05m and 0.11m deep. To its west 35m long ditch 2252 varied in width from 0.50m to

0.93m and in depth from 0.07m to 0.27m. It was cut by gully 2199. Around 18m south of 2250 another 8m long gully with two surviving termini, 2254, was around 0.40m wide and up to 0.10m deep. It is possible that as a group the gullies made up a bounded 20m wide track or droveway. Neighbouring lengths of gullies may represent different phases within the long term organization of space and movement on the site. Some of the gaps between sections of gullies are undoubtedly due to truncation but some may also be access points along the drove. The north-westward turn at the west end of 2252 is the most obvious example.

5 The western edge of the central area of the site was defined by slender linear features, which may have bounded the group of pits which over time extended to the east side of the site’s north and may also have been a branch from the possible droveway. A 6m long gully 2261 with termini at both ends was in excess of 0.50m wide and up to 0.17m deep. It was cut by gully 2262. The system was extended south by a 15m long, 0.50m wide and up to 0.12m deep slightly curving gully, 2265, continuations of which may have been gullies 1402 and

1403. These, in themselves unprepossessing, gullies might be marking the most important boundary of the site, continuing south as 2266-7, 2281, west of which is nothing until the medieval period. Unfortunately these features are essentially dated. If they did belong with 2266–7, they would be middle to late Iron Age.

Both termini of approximately 60m long north north west to south south east ditch, 2157, lay outside the limits of excavation. Over nine slots it varied in width from 1.10m to 2.00m and in depth from 0.08m to 0.45m.

The phasing of the ditch is of particular importance for the northern part of the site as it is the most substantial pre-Modern boundary in that area and it has several identifiable relationships. It was considered to cut gully

2156 and pit 43 but was cut by later Iron Age ditch 2155, Roman gully 2158 and modern trackway 2160. A short section of ditch which bifurcated to the north of 2155, west of 2157, has been treated as part of the former.

To the east of 2157, ditch 2154 slanted across 2155, cutting it before appearing to terminate slightly to the west of slots 213 and 212. It did not intersect with 2157. The relationship of 2157 with pit 43 is of particular significance as the pit can be dated firmly by substantial joining fragments from a single later Iron Age vessel whereas only eight relatively small sherds of earlier and later Iron Age pottery were recovered from 2157 itself, distributed over three of the nine slots. All of the earlier material was from slot 40, the point at which the ditch cut the pit. Pit 43 itself had cur an earlier pit, 41, dated by a large slab of apparently Bronze Age plain wall sherd. It seems likely that the earlier sherds in slot 40 were residual, possibly derived from the re-cutting of 41.

The relationship with 2158 is not straightforward: 2158 cut 2157 but also terminates at that point, suggesting a degree of contemporaneity.

Ditch 2154 cut an inhumation with entirely Roman characteristics, 218 and also cut ditch 2155. Ditch

2155 included the full profile of a later Iron Age jar in fresh condition in slot 201 and was cut by ditch 2154 which included substantial sherds of similar date.

An enigmatic group of linear gullies, S2, formed either a blocked-off track or a very narrow enclosure.

The southern end of 14m long, south–west to north–east arcing gully 2186 terminated just short of, and may have been associated with, pit 543, which was dated to the earlier Middle Iron Age by ten sherds of moderate size. A handful of small sherds from a single slot across 2186 were dated broadly as prehistoric. The gully was consistently around 0.40m wide but the depth varied from 0.12–0.30m. It cut two short gullies, 2189 and 2190,

6 neither of which produced finds but both of which were in broad alignment with 2156. The gullies were contained within the lines of 2186 and the elbow-bend-shaped gully 2188, and all three were cut by a straight linear ditch 2187 which was well dated by later Iron Age pottery. At its southern end 2188 was butted by the upper fills of 2187. It was around 0.35m wide and 0.10m to 0.14m deep. Four sherds of moderate-sized, undifferentiated prehistoric pottery were recovered from it.

Three curvilinear gullies, 2153, 2172 and 2179, represent between 20% and 40% of possible circular features. None had potential diameters exceeding 10m and they would be of small scale for barrows hence they may have been associated with roundhouses. None of them contained any finds but 2153 and 2172 were perceived to have been cut by later features. 2153 was exposed as a 5.5m length of up to 0.25m wide, 0.10m deep curvilinear gully with a terminus at its east end. It was cut by pit 26 which contained two sherds of later

Iron Age pottery. A length slightly in excess of 8m of gully 2172 survived. It had a width of up to 0.70m and depth varying in three slots across it from 0.11m to 0.40m, with a terminus at its southern end. It was cut by an undated pit, 940, and by a gully, 941, from which two sherds of Roman pottery were recovered. There was no dating evidence for 2179 other than its form. Over four slots it varied in width from 0.38m to 0.58m and in depth from 0.11m to 0.17m. There is no evidence for post holes within the arcs, but this might be due either to horizontal truncation or to wattle and daub walls supported by stakes, a well-attested Iron Age practice which was not an established method of construction during the Bronze Age. Any of these three features might equally be allocated to either of the Iron Age phases.

Pits

The earlier Middle Iron Age pits tend to be shallower and to contain fewer complex fills than those of later in the period. The simplest pits were often scoop-like and showed no trace of deliberate deposition and filling appeared to have been a gradual natural process following abandonment, lacking any traces of sudden or high energy episodes of erosion. The fills appeared typically as yellowish to mid brown soil rather than re-deposited natural, presumably reflecting a cultivated or well-trodden local environment. Examples of this form include

206, 514, 614, 640, 900, 904, 1422, 1433 and 1446. The latter, which was cut by later Iron Age pit 1445, included a quern fragment, despite its perceived gradual fill formation. Deeper pits in this group, such as 543 and 1347, have traces of re-deposited natural interleaved with soil and, in the case of pit 917, fragments from a substantial portion of a pot, as well a sherd from at least one other vessel.

Table 2. Earlier Middle Iron Age pits with exclusively slow to moderate fill formation Cut Fills Form Diameter Depth Dating evidence Other finds (m) (m)

7 Cut Fills Form Diameter Depth Dating evidence Other finds (m) (m) 206 279 Concave bowl 0.60+ 0.15 Stratigraphy 514 751 Flat bowl 2.40 0.45 Earlier Middle Iron Age pottery 2 flint flakes, daub 543 791, 792 Flat bowl 2.30 0.31 Earlier Middle Age pottery flint flake 614 880 Flat bowl 1.05 0.17 Earlier Middle Iron Age pottery fired clay 640 993 Concave bowl 0.60+ 0.15 Earlier Middle Iron Age pottery 900 1281 Concave bowl 0.45+ 0.08 Stratigraphy 904 1285 Flat bowl 0.68 0.10 Stratigraphy 917 1299 Flat bowl 1.54 0.36 Earlier Middle Age pottery* fired clay 1347 1879 Concave bowl 1.72 1.03 Earlier Middle Iron Age pottery, stratigraphy flint spall 1422 1986 Concave oval bowl 1.22 0.30 Iron Age pottery 1433 2073 Concave bowl 0.97 0.27 Stratigraphy 1446 3054 Flat bowl 1.20 0.22 Earlier Middle Iron Age pottery, stratigraphy quern (* Includes illustrated sherds)

In some instances there is a single rapid episode of infilling which may be due either to high energy erosion,

typically flooding, collapse or deliberate backfilling, in either case followed by a period of more gradual filling

by natural processes. An initial deposit around the sides on the base of pit 1039 was probably due to erosion but

in the cases of pits 611 and 620 rapid fills appear to have entered the pit from one side, either due to the collapse

of the edge or to the washing in of, or deliberate infilling using, the upcast heap. Thereafter the filling was a

gradual process. Pit 631 also included a primary erosional deposit (971) but it was covered with several thick

lenses of hearth or oven waste (991, 992, 973), which were in turn sealed by humic soil of possibly slower

formation (972). In contrast, the lower and middle fills of pit 229 formed moderately slowly through natural

erosion, capturing six large sherds of earlier Middle Iron Age pottery. The lower fills were covered by a single

humic rapid deposit, including two sherds of later Iron Age pottery introduced in the residual hollow, either after

the lower fills had stabilized or after they had subsided. Pit 1348 had a humic deposit on its base which was

sealed by slow to moderately formed eroded soil and natural. The greater parts of two pits, 1431 and 2548, were

filled in rapidly with sand and gravel respectively. However, whereas the fill of the former sealed a stoney basal

deposit the latter was a single fill. Both of these pits were re-cut during the later Iron Age.

Table 3. Earlier Middle Iron Age pits with one episode of rapid deposition

Cut Fills Form Diameter Depth Dating evidence Other finds (m) (m) 33 85 Flat bowl 0.70 0.26 Iron Age pottery 229 358, 359 Concave bowl 1.39 0.54 Earlier Middle Iron Age pottery Fired clay ‘block’ 611 876, 877 Flat cylinder 1.39 0.54 Earlier Middle Iron Age pottery 620 950, 951 Concave bowl 1.32 0.62 Earlier Middle Iron Age pottery flint flake 631 971, 972, 973, 991, 992 Concave bowl 2.18 0.60m Earlier Middle Iron Age pottery fired clay 1348 1880, 1881 Concave bowl 1.20 0.38 Earlier Middle Iron Age pottery 1431 2067, 2068, 2069 Concave bowl 0.48 0.37 Stratigraphy 2548 2054 Concave beehive 1.30 0.51 Stratigraphy During this phase multiple rapid fills are sparse. An elongated pit, 1409, had a humic deposit made directly onto

its base. Around the side the deposit was covered by collapsed gravel over which a further deposit incorporating

plentiful charcoal had been had been introduced, sealed more gradually by soil. A similar deposit on the base of

pit 1420 was capped by a thick gravelly layer (1977), presumably a deliberate seal introduced immediately. The

fill of a cut 2547 into (1977) included later Iron Age pottery. In contrast the base of 1429 was covered with

8 primary silt 2063 sealed by gravel which had collapsed from the sides (2062). After a further period of gradual

natural silting (2061) similar soil mixed with large natural stones (2060) was dumped into the pit, followed by

an organically rich deposit (2059). Pits 1420 and 1429 were very similar in scale and in the regularity of the

circular plans.

Table 4. Earlier Middle Iron Age pits with more than one episode of rapid deposition Cut Fills Form Length Width Depth Dating evidence Other finds (m) (m) (m) 1409 1897, 1898, 1899, 1950, 1951, 1952 Concave bowl 1.58 0.41 Earlier Middle Age pottery Loomweight/oven furniture 1420 1974, 1975, 1976, 1977 Concave cone 1.95 1.00 Stratigraphy 1429 2059, 2060, 2061, 2062, 2063 Concave cone 2.19 0.99 Prehistoric

Post holes

Only two unrelated post holes could be allocated to this phase with supporting evidence and in neither case was

the evidence conclusive.

Table 5. Earlier Middle Iron Age post holes

Cut Fills Shape in plan Diameter Depth Dating evidence (m) (m) 344 496 Concave bowl 0.44 0.11 Earlier Middle Age pottery 1631 4150 Concave, regular 0.56+ 0.14 Stratigraphy

Phase 3, Later Iron Age (Fig. 13)

The onset of the later Iron Age phase is characterized by the introduction sauce-pan pot style vessels of the type

classified as PB1.1 at where evidence from radiocarbon dating suggests a long currency extending

from around 310BC to 50BC (Brown 2000, 90). Other broadly contemporary vessel forms from the site such as

rounded profiled, high-shouldered jars (JC2), S-profiled jars and bowls (BC2) with upright or out-curved rims

have a similar duration (Brown 2000, 86-7, 89). Typical of the latter forms are the substantial fragments from at

least three vessels in pit 116, one of which had decoration entirely characteristic of the phase. Pottery of this

phase is more widespread and there are several features which are datable by significant concentrations of finds.

As a consequence phasing of linear features and pits is more reliable than in the earlier phases.

The northern droveway identified as a key to the growth of settlement on the eastern edge of the plateau

appears to have been more clearly defined as a track by ditches 2250, 2150, 2154, 2155, 2156, 2174, 2177 and

2178. Ditch 2178 included three sherds of later Iron Age pottery and was cut, possibly extended by 2177 which

contained 76 later Iron Age sherds, many of them substantial, and a single Roman sherd. 2177 was cut by

Roman ditch 2176. Two ditches extending southwards from the droveway, 1132 and 2159, are probably

residues of a field system, S3, in a coaxial relationship with it. However, the substantial linear ditch, 2157, at its

fullest extent, cuts across the track. Given its contradictory stratigraphic relationships with different segments of

9 the track it may be that different sections of it were in use at different times, although this would raise the possibility that the ditch was not a single entity, despite its apparently continuous line. Further south, on the western edge of the exposed area, ditches 2262, 2266, 2267 and 2270 were probably elements of the same system.

The only sherd from three segments of gully 2156 was deemed to be of later Iron Age date. The gully width and depth were fairly consistent at respectively between 0.40m and 0.50m and from 0.09m to 0.12m.

Crucially the western terminus of gully 2156 extended slightly beyond the western edge of 2157, implying that the gully would have formed part of a long intermittent boundary or southern limit of the track, uninterrupted by any boundary perpendicular to it. Thus, even if the perceived relationship between 2157 and 2156 was the result of de-silting of the former the original cutting of 2157 would still have post-dated that of 2156. However, the single sherd from 2156 implies that it would have been a late feature in the development of the droveway.

An enclosure or remaining fragment of a field system, S5, lay south of S3 and on a different orientation to it. An arcing ditch 2277 appeared to be respected by the north-west corner of a near-square enclosure made up of ditches 2280, 2288 and 2291. 52 pottery sherds, of a larger size than is typical of the site, from 2277, 2280, and 2288 were all of middle to Late Iron Age date, as were two smaller sherds from 2291.

A third system, S4, may be a continuation of S5 but with an altered orientation influenced by an indentation of the contour at the eastern edge of the plateau, to which it is parallel. The change of orientation is defined by the acutely angled relationship between ditches 2177 of S5 and the up to 1.10m wide, 0.47m wide

2187 of S4. The south westerly orientation is continued by ditches 2181 and the slightly curving concentric 2164 and 2165. Ditches 2192, 2194, 2195 and 2185 are in perpendicular relationships to that orientation. The ring gully arcs, 2179 and 2172, may be indicative of structures associated with family plots on either side of ditches

2181 and 2165. The narrowing of the track caused by the eastward extension of and terminus of S3 ditch 2177 coincides with breaks in ditches on either side of 2187 so that it is likely that the gaps were gates through which livestock were sorted. There were no relationships between elements of S4 and S3 but 2187 cut earlier Middle

Iron Age gullies 2188 and 2189 and 2192 cut gully 2196. Gully 2181 was cut by an otherwise undated post hole

904, 2185 was cut by pit 634 and 2192 was cut by pit 409. Middle to Late Iron Age pottery was in relative abundance with 586 sherds distributed across 2164, 2165, 2181, 2185, 2187, 2192 and 2196. Three small

Roman sherds may be regarded as intrusive. A further ditch, 2194 contained three sherds of later Iron Age pottery and cut respectively 2196 and 2193 and was cut by Roman ditch 2195. Whilst this demonstrates that

2194 was a later feature it appeared terminate on pit 509/510 which was well-dated by 47 larger than average sherds of exclusively Middle to Late Iron Age pottery. The ditch may have been a late modification of S4.

10

Pits

Pits listed in Table 6, and re-cut 2547 were filled with one or two slowly to moderately formed soil erosion

deposits although 248, 509/510, 527, 601, 630 and 1040 were much deeper cuts. The material in 25 and 26 was

finer and may have included wind-blown particles. Pits 422, 545, 630 and 1523 appeared to have filled

gradually with eroded soil mixed with natural but in a series of identifiable episodes suggesting formation over a

considerable period, although the lower fill (3086) of 1523 included a substantial amount of Middle Iron Age

pottery. The pottery from 545 gives support for this interpretation. The basal fill produced four substantial later

Iron Age sherds but the latest fill included Iron Age and Roman sherds. Substantial fragments from a jar profile

in 422 were comparable with Late Iron Age vessel profiles recovered from the Danebury area, although

decorative motifs would allow a slightly earlier date (Fig. X, 16; Brown 2000, 87-8) which would overlap with a

saucepan pot from 630 (Fig. 23: 16; Brown 2000, 90). None of these pits contained any other finds than pottery.

Pit 635 probably belongs to this class but it was badly disturbed by bioturbation. Pit 1800 is very much an

outlier from the period. Its dating depends upon a single small Iron Age sherd and the fact that it is cut by a

second phase Medieval gully. It might equally be treated as a post hole from the first Medieval phase.

Table 6. Later Iron Age pits with exclusively slow to moderate fill formation

Cut Fills Shape In plan Length Width or Depth Dating evidence (m) diameter (m) (m) 25 77 Concave bowl 0.78+ 0.21 Later Iron Age or Roman pottery 26 78 Flat bowl 1.30+. 0.12 Later Iron Age pottery 112 172 Concave, weakly conical 1.47 0.64 Later Iron Age pottery 124 188 Concave bowl 0.80 0.30 Later Iron Age pottery 125 189 Flat, elongated 1.82 0.39 Stratigraphy 242 386 Flat bowl 0.95 0.40 Later Iron Age pottery, fired clay 248 391 Flat, elongated 1.50 0.13 Later Iron Age pottery 422 596, 597, 598, 599 Flat open cone 2.20 0.88 Later Iron Age pottery, fired clay 509 / 510 692, 693 Flat open cone 1.90 0.60 Late Iron Age pottery *, stratigraphy, fired clay 545 795, 796, 797, 798 Flat open cone 1.65 1.03 Later Iron Age pottery 601 801, 802 Flat open cone 1.45 0.95 Later Iron Age pottery 615 955 Concave bowl 0.86 0.17 Later Iron Age pottery 630 975, 976, 977, 978, Flat cylinder 1.40 0.74 Later Iron Age pottery * 979 635 984 Flat bowl 1.40 0.30 Later Iron Age pottery, fired clay 641 994 Flat dish 1.50 0.18 Later Iron Age pottery 733 1098 Flat dish 1.07 0.11 Later Iron Age pottery 901 1282 Concave bowl 0.85 0.25 Iron Age pottery, stratigraphy 1034 1481 Flat dish 0.93 0.12 Later Iron Age pottery 1040 1490 Concave bowl 1.01 0.36 Iron Age pottery 1305 1779 Concave bowl 0.80 0.20 Later Iron Age pottery 1423 1987, 1988 Concave bowl 0.85 0.45 Middle Iron Age pottery 1523 3084, 3085, 3086 Flat open oval cone 1.70 1.52 0.37 Middle Iron Age pottery 1800 3260 Concave bowl 0.73 0.35 Iron Age pottery, stratigraphy 1933 3458 Concave bowl 0.64 0.14 Iron Age pottery, stratigraphy 2547 1978, 1979 Flat, convex-sided 1.70 0.68 Later Iron Age pottery, stratigraphy (* Includes illustrated sherds)

The simplest examples of rapid infilling were single humic deposits in pit 727, which included 29 later Iron Age

pottery sherds, and 1140. The fill sequence in a near circular, steep-sided pit 107 comprised an initial thick

11 brown soil deposit which was of probably moderate formation rate capped rapidly by re-deposited natural

gravel, itself sealed by eroded brown sandy soil. Pit 633 had a thin, charcoal-rich, similarly humic basal deposit

but the mixed soil and natural deposits sealing it were of generally gradual formation by erosion. More typically

the shallow pit 129 had a rapidly formed sandy erosion deposit around its sides sealed by gradually accumulated

eroded soil. A 0.22m deep gravel deposit on the base of 1430 may have been deliberate infill but this was also

covered by gradually eroded soil.

Two thick moderate gravelly erosion deposits filling the lower half of pit 803 were covered by rapidly

introduced ashy material which was then sealed by slowly accumulated eroded soil.

Pit 43 was one of several later Iron Age re-cuts into earlier pits which included substantial portions of

vessels. It cut into the south-west side of the possibly Bronze Age pit 41. Its single rapid humic fill included and

sealed a nearly complete decorated later Iron Age vessel (Fig. 23; 16). This vessel had carbonized food residues

adhering to it which returned a radiocarbon date of 210-54 cal BC (UBA 22372). A substantial amount pottery,

including several large sherds, indicated a similar pattern for pit 827 which had been cut into gravel-filled pit

828. The gravelly soil (1171) filling the lower 80% of pit 731 had been re-cut and filled with a humic soil (1170)

which also contained substantial fragments of later Iron Age pottery. The fill of the re-cut was sealed by eroded

soil (1169). At the time of excavation the feature was interpreted as a posthole with a 0.45m diameter pipe. This

remains possible, although at twice the diameter the pit would be disproportionately large for the post. Pits 116

and 148 appeared to have been filled by soil erosion in much the same fashion as 25 and 26 but large joining

pottery sherds imply at least one episode of rapid, presumably deliberate, infilling in each.

The base of pit 916 appeared to have been blackened by fire and its lower fill (1298) included several large

cuboid blocks of fired clay (Pl. 2). It is unclear whether they product of firing or whether they were oven or

furnace spacers which had collapsed in situ. The deposit was sealed by gradually accumulated eroded soil

(1297). Several substantial burnt flints were incorporated into both fills. A deposit similar to (1298), including

both fired and unfired clay (1995) under a charcoal-rich lens (1994), had been placed over gradual low and

middle fills of eroded material in pit 1426 (Pl. 3).

Table 7. Later Iron Age pits with one episode of rapid deposition

Cut Fills Shape in plan Length Width/ Depth Dating evidence (m) Diameter (m) (m) 43 97, 98 Flat bowl 1.32 0.47 Later Iron Age pottery*; radiocarbon date 210–54 cal BC 107 165, 166, 167 Concave cylinder 1.38 0.48 Later Iron Age pottery 116 176 Concave bowl 0.75 0.17 Later Iron Age pottery * 129 181,182 Flat bowl 1.23 0.31 Later Iron Age pottery, fired clay 145 263, 264 Concave bowl 1.05 0.33 Later Iron Age pottery, stratigraphy 147 268, 269 Flat bowl 0.90+ 0.42 Later Iron Age pottery, stratigraphy 148 262 Flat bowl 1.50 0.30 Later Iron Age pottery *, fired clay 633 980, 981, 982 Concave cylinder 1.07 0.70 Later Iron Age pottery

12 727 1093 Flat oval bowl 1.40 1.26 0.41 Later Iron Age pottery, fired clay 803 1178, 1179, 1180, 1181, 1182 Concave bowl 1.72 0.82 Iron Age pottery, fired clay 827 1253 Concave bowl 0.65 0.35 Later Iron Age pottery, stratigraphy 916 1297, 1298 Flat bowl 1.46 0.48 Later Iron Age pottery, fired clay 1140 1655 Flat dish 0.90 0.12 Iron Age pottery, fired clay 1426 1993, 1994, 1995, 1996, 1997, 1998 Flat bowl 1.28 0.56 Later Iron Age pottery 1430 2064, 2065, 2066 Flat open cone 1.35 0.60 Later Iron Age pottery (in top fill) (* Includes illustrated sherds)

Pits with more than one episode of rapid filling: The northern side of Pit 711 had collapsed forming a rapid deposit (1075) which was covered by a gradual

deposit of eroded soil (1074). This was entirely sealed by a thin, rapidly formed, charcoal-rich humic layer from

which a 0.20m, 0.05m wide downward tapering projection is visible in section (1073). The projection might be

due to burrowing but its profile is suggestive of a stake which may have burnt in situ with the resulting charcoal

being spread over the underlying layer. However, there is no clear evidence for in situ burning. The deposit was

covered by a gradually formed eroded soil deposit (1072). Following a phase of rapid erosion of the sides

forming soily gravel (1962) on the base of pit 1412 there were several episodes of slower accumulation of soil

and natural sand before most of the upper third of the pit was filled in with humic soil (1958). Pit 1417 had a

similar basal fill which was covered rapidly by a charcoal-rich humic deposit sealed by a moderately formed

erosion product of soily sand. Filling of the pit was completed by the introduction of a soily gravel cap. Pit 1421

had similar lower deposits but the humic deposit (1983) was capped with re-deposited natural (1982). More

humic material had been dumped on the cap before gradual erosion completed the filling of the pit. The pit was

unusual in having significant amounts of substantial pottery sherds in all the middle and lower fills, most notably

within the basal erosion deposit and the cap, the latter producing 44 large sherds. Pit 1427 was a re-cut into pit

2548 which a rapid basal fill made up of material collapsed from the gravelly fill (2054) of the earlier pit. It was

covered by a deposit of slow formation (2052) which was sealed by a sandwich (2051) of yellow clay and a

charcoal rich lens. The final fill was of slow formation cut by pit or post hole 2549.

More common was the deposition of humic soil on the base of the pit followed by a rapid erosion deposit.

In pit 1323 this was followed by slow to moderate accumulation of eroded of mixed eroded soil and natural.

The initial fill of 135 was a charcoal rich, homogenous loamy soil sealed by a layer of large stones burnt to

a red hue (252). Both were clearly deposited rapidly, probably in a single episode. Silty loam partially covering

the stones was probably an erosion deposit (251) which formed more gradually but was sealed by a cap of sand

and small to medium angular stones (250), clearly a second phase of rapid deposition. The final fill, the only one

from which pottery was recovered, may have been within a re-cut (199). The lower two fills of pit 527 were

comparable but then sealed rapidly by a deep humic deposit.

13 There was no clear relationship between adjacent pits 207 and 208 but although their fill patterns were similar it is unlikely that their excavation was closely contemporary. Both appeared to have had a basal humic deposit partially sealed by collapse of natural at the sides, although in the case of 208 the earlier deposit (290) was of material which had been washed in from the fill of gully 2159, which it cut. The humic basal deposit

(288), butted against the collapse, included substantial fragments of pottery. In pit 207 pottery attributed to the collapsed material (285) on its east side may have been part of the basal deposit but most of the pottery from the pit was within a rapidly introduced deposit (282) which sealed the collapse. The pottery included much of a

Durotrigian bowl which is datable to the century leading up to the Claudian invasion hence the episode is of a distinctly late Iron Age date. A second ‘pair’ of adjacent large pits, 603 and 604, had broadly similar fill patterns and in the case of 603 the middle deposits, 883) and 884), included later Iron Age pottery whilst the slowly former upper fill, 881, included mainly Roman pottery. The fill of a probable re-cut into an erosion deposit making up the upper middle fill of 604 included pottery which was deemed either Late Iron Age or Roman. The edge of a rapid humic basal deposit in pit 1039 was also partly covered by natural collapsed from the sides in turn covered with a second humic deposit. Both humic fills included substantial sherds of later Iron Age pottery and there were 8 large fragments of fired clay in the basal deposit. Soil had accumulated gradually in the upper part of the pit.

In some instances an erosion deposit of soil, sometimes mixed with natural gravel or sand had formed gradually at the bottom pits and was covered with a humic deposit. In pit 331 the deposit was capped with over

0.50m of gravel. Pit 1445, cutting the earlier pit 1446, had a basal eroded soil deposit covered by gravel collapsed from the sides which was then sealed by eroded soil. This was covered by a rapid deposit of stony soil including substantial fragments of Late Iron Age and possible some earlier Middle Iron Age pottery.

In pit 1331 several local rapid deposits were interspersed with episodes of soil erosion. A final rapid deposit (1859) included substantial fragments of a decorated globular Middle Iron Age pot of the same type as that from the base of pit 43 (Brown 2000, 87, fig. 3.24, DA334).

Several pits in this group had characteristics in common with single rapid fill pits 916 and, in particular,

1426. In 1428 a rapid infilling of fairly clean gravel was covered by residues of burning, comprising fire- reddened soil on the south east side and an ashy deposit including fired clay on the north west side. The upper pit was filled with interleaving rapid deposits of soil, sand and gravel. The base of pit 605 was covered by a fairly clean rapidly formed gravelly sand erosion deposit (969) which was sealed by a more soily gravel deposit covered by sandy soil. This deposit may have been re-cut before the rapid deposition of a charcoal-flecked humic layer with interleaving sandy clay (866) which was covered by a more slowly formed soily deposit. In pit

14 1635 large lumps of fired clay (4157) on the base were sealed by a humic basal deposit (4156) covered by a mixed gritty and gravelly fire-reddened soil (4155). The overlying humic soil was probably deposited rapidly also. Pit 1432, re-cutting pits 12431 and 1433, also included a layer of fire-reddened soil (2071). It was deposited over eroded soil and ‘closed’ by a central deposit of fire-reduced large stones. Substantial wall sherds lying on the base of pit 1820 were covered by soily gravel sealed by a gradually formed soil. A deposit of fired clay (3290) partly covered by charcoal was set in the centre of the soil which was covered gradually. The upper fill was cut by Roman ditch 2284.

Table 8. Later Iron Age pits with more than one episode of rapid deposition

Cut Fills Shape in plan Diameter Depth Dating evidence Other finds (m) (m) 135 199, 250-2 Flat bowl 1.02 0.39 Later Iron Age pottery Loomweight/oven furniture, flint flake 146 265-7 Flat bowl 2.00 0.43 Later Iron Age pottery, stratigraphy 207 280-5 Flat bowl 2.16 0.84 Late Iron Age pottery * fired clay 208 286-90 Flat bowl 1.10 0.56 Later Iron Age pottery, stratigraphy fired clay 331 553-7 Flat cylinder 1.13 0.90 Later Iron Age pottery 527 769-72 Flat cylinder 1.25 0.80 Later Iron Age pottery fired clay 549 852-4 Concave open cone 2.20 0.85 Later Iron Age pottery 603 881-9 Flat cylinder 2.15 1.40 Late Iron Age */ Roman pottery fired clay 604 890-9 Flat cylinder 2.35 1.30 Late Iron Age / Early Roman pottery fired clay 605 864-9 Concave bowl 2.10 0.95 Later Iron Age pottery fired clay 711 1072-5 Concave beehive? 1.30 1.01 Later Iron Age pottery Loomweight/oven furniture, quern, animal bone 1039 1486-9 Flat cylinder 1.50 0.59 Later Iron Age pottery fired clay 1323 1870-3 Flat cylinder 1.70 0.86 Later Iron Age pottery flint flake, fired clay 1331 1855-9 Flat bowl 1.26 0.41 Later Iron Age pottery * Loomweight/oven furniture, flint flake, animal bone 1401 1884-8 Concave cylinder 1.09 0.70 Middle and Later Iron Age pottery 1412 1956-62 Concave cylinder 1.69 0.77 Iron Age pottery 1417 1968-71 Concave irregular 1.84 0.80 Iron Age pottery, stratigraphy fired clay 1421 1980-5 Flat beehive? 1.14 0.73 Later Iron Age pottery fired clay 1427 20513,4261 Concave cylinder 1.00 0.45 Later Iron Age pottery 1428 2055-8 Concave bowl 1.30 0.55 on Age pottery 1432 2070-2 Flat convex-sided 1.26 0.39 Later Iron Age / Early Roman pottery 1445 2099, 3050-3 Flat open cone 2.50 1.03 Later Iron Age pottery * animal bone 1635 4154-7 Concave cylinder 1.10 0.57 Later Iron Age pottery * fired clay 1820 3289-91, Flat bowl 1.30 1.00 Later Iron Age pottery, stratigraphy fired clay (* Includes illustrated sherds)

Post holes

There is very little discernible patterning in the distribution of post holes of this phase and where it does occur the attribution is doubtful. Close to the southern boundary of the stripped area 2346 is in the south east corner of a group of four post holes forming a convincing rectangle in plan. However, 2346 is much deeper than the other group members. Furthermore dating is dependent on a very small Iron Age pottery sherd. A very substantial post hole 731 retained a 0.43m diameter pipe distinct from the surrounding gravel packing.

Cut Fills Shape in plan Diameter Depth Dating evidence Other finds (m) (m) 30 82 Flat bowl 0.26 0.07 Later Age pottery 31 83 Flat conical 0.63 0.34 Prehistoric pottery fired clay 109 169 Concave bowl 0.44 0.23 Later Iron Age pottery 132 196 Concave bowl 0.43 0.06 Later Age pottery 617 958, 959 Concave bowl 0.50 0.18 Later Age pottery / Early Roman pottery 625 966 0.25 0.05 Later Age pottery 731 1169, 1170, 1171 Concave cone 0.90 0.37 Later Iron Age pottery

15 1022 1469 Concave bowl 0.32 0.06 Iron Age pottery 1332 1879 Concave bowl 0.27+ 0.13 Iron Age pottery, stratigraphy fired clay 1716 3181 Concave bowl 0.39 0.06 Later Age pottery 1826 3278, 3279 Concave bowl, overcut on SSW side 0.67 0.34 Later Age pottery 2033 3564 Concave bowl 0.40 0.10 Later Iron Age pottery 2345 3898 Concave cylinder 0.34 0.22 Association 2346 3899 Concave cylinder 0.50 0.54 Iron Age pottery 2347 3950 Concave bowl 0.27+ 0.17 Association 2348 3951 Concave bowl 0.48 0.14 Association 2349 3952 Conical 0.27 0.16 Association Table 9. Later Iron Age post holes

Phase 4, Later Roman (Figs. 14-16)

The range of pottery provides scant evidence for continuity of settlement from the Iron Age into the Roman

occupation as the diagnostic sherds are datable to the latter part of the period (Timby, below). The general

distribution of Roman features shows little patterning apart from pairing of ditches, probably indicative of

tracks, a cluster of graves in the central north area and two corn-dryers and their related features. A few ditches

and gullies appear to be traces of field systems but the orientations of most of the linear features show little

coherence. This may reflect variations in local topography and possibly the organic continuous development of

land use over a century or more.

The location of a flaring droveway, S6, which at its east end is south of the later Iron Age track, reinforces

the impression of an hiatus in organized use of the space, although it may also link back to an earlier route

further to the west. Modification over time was demonstrated by the cutting of ditches 2257 and 2255 (possibly

originally a single ditch) by 2256. These ditches formed the southern limit of the droveway which was bounded

by 2252 and 2259 to the north. 2252 terminated on 2251 which formed a discontinuous boundary with 2253.

The latter cut gully 2199 but no other relationships with features outside S6 were demonstrated. The dating of

the system depends on 8 medium to small sherds of Roman pottery. Other possible elements of the system

include ditch 2152, which cut earlier Middle Iron Age gully 2151 but produced no finds, and gullies 2163, 2162

and 2366, a possible continuation of 2162. 2162 may have had a longer period of active use as it cut the fills of

2163. Thirteen Roman sherds were recovered from 2162 and two from 2163 along with two prehistoric sherds.

A second track, S7, defined by a pair of ditches 2182 and 2183 had respectively four and three Roman

sherds, as well as later Iron Age sherds. No relationships were established with either ditch but they are set

within the area of densest activity of the period. To its north-east ditch 2195 was in effect a re-cutting of Iron

Age ditch 2194 which bisected 2193. Its easternmost slot, 434, was the most productive of Roman pottery by

weight (620g) from the site. However, at the intersection with 2193 it produced exclusively later Iron Age

pottery.

16 There are several ditches or gullies within or marking the periphery of this space. The north arm of right- angled ditch 2170 is of a broadly similar orientation to that of 2162. It produced a single moderately-sized

Roman sherd and it cut the undated curvilinear gully 2168. Its relationship with gully 2171 could not be determined but the latter had a similar orientation to that of Medieval phase 2 and cut a pit or treebole, 805, which contained two moderately sized late Roman sherds. To its south gully 2167 contained a single substantial

Late Roman sherd and had an orientation similar to that of the back of corn dryer 2. The orientation of a short but substantial ditch segment with a terminus at its north-east end, 942, may also have been based on that of the corn dryer and it contained two substantial Roman sherds. The peripheral WSW–ENE oriented ditches 2158 and

2176, cutting respectively Iron Age ditches 2157 and 2177, both included substantial Roman sherds whilst 2191, a ditch sharing its orientation with that of corn dryer 1, contained a single moderate sized Roman sherd and cut undated ditch 405.

The approximately 25m diameter of an enigmatic curvilinear gully, 2184, on the eastern edge of the stripped area is too large to have been a domestic structure of the period. It contained two Roman sherds and cut a hollow, 719, which in turn cut pit 720, which included five substantial Roman sherds. It also cut undated pits

730 and 845, but its fills were cut by a post hole, 708, which included two small Roman sherds, and an undated pit, 725. Gully 2184 may have remained in use for a significant period as its slot 718 cut the fills of a roughly concentric arcing gully, 717.

Towards the centre of the stripped area there were no perceptible relationships between 2278 and 2279 but both contained fairly substantial Roman sherds, albeit few of them.

Pits

The Roman pits are of a consistent uniform simplicity in comparison to many of those of the Iron Age, particularly the later part of that phase. The stark contrast adds further support for the view that there was an hiatus following the Iron Age occupation as had there been continuity some persistence of earlier fill patterns might be expected.

Later Roman pits of exclusively slow to moderate fill formation: Bowl and lozenge-shaped pits 602, 905 and 948 contained single slow to moderately formed loamy fills including isolated Roman sherds. In contrast the intercutting 1126 and 1128, 1533 and 1915 had similar fills which produced respectively 16, 29, 26 and 2 substantial sherds. 602 cut Iron Age pit 604 and 948 cut the northern end of corn dryer 2, pit 1126 was cut by an undated gully, 1125 and pit 1533 by post medieval or modern ditch 2271. To the south of the corn dryer, pit 943 was filled with a series of slow to moderately formed upper deposits broken by a very local gravelly deposit which covered grey silts probably formed from matter

17 decaying in a damp environment. The pit is likely to have served as a well. Shallow, amorphous or lozenge-

shaped pits 615, 805, 812 and 1012 containing Roman pottery were probably treeboles. To the north of the corn

dryer a spread, 1264, filling a broad shallow depression, sealed the upper fills of shallow bowl or dish-shaped

pits, 1838-43 and a probable post hole, 1841, which were exposed in section. Two of the pits produced two

Roman sherds but there was no other dating evidence. The extent of the spread would allow several more such

pits to have been present and it seems likely that the area was reserved for a particular process, possibly relating

to the treatment of cereal crops, or possibly as an area for waste disposal. The post hole appeared to have

packing and a pipe and would have provided substantial support for any structure covering the area.

Table 10. Later Roman pits of exclusively slow to moderate fill formation

Cut Fills Shape in plan Diameter Depth Dating evidence Other finds (m) (m) 144 260, 261 Irregular bowl 1.60 0.27 Roman pottery 602 863 Concave bowl 1.60 0.44 Roman pottery 616 956-7 Flat amorphous 2.20 0.33 Roman pottery 805 1173 Flat dish 1.00 0.15 Roman pottery 812 1185 Flat amorphous 1.80 0.17 Roman pottery 838 1265 Concave bowl 0.74 0.19 Association 839 1266 Concave dish 0.82 0.11 Association 840 1267 Concave bowl 0.55 0.15 Roman pottery 842 1270-2 Flat bowl 1.08 0.23 Roman pottery 843 1273 Concave dish 0.75+ 0.15 Association 906 1287 Concave bowl 0.95 0.25 Roman pottery 943 1381-7, 1661-3 Bipartite cone and cylinder 2.60 1.67 Roman pottery fired clay 948 1456 Concave bowl 0.66 0.24 Roman pottery 1012 1455 Flat lozenge 1.60+ 0.24 Roman pottery 1126 1582 Flat lozenge 0.79+ 0.22 Later Roman pottery quern 1128 1583 Flat lozenge 0.60+ 0.20 Later Roman pottery quern 1533 3098 Concave bowl 0.98+ 0.28 Later Roman pottery 1542 4060 Flat dish 1.23 0.20 Roman pottery 1915 3378 Concave bowl 1.13 0.33 Later Roman pottery Later Roman pits with one episode of rapid deposition: The initial deposit in pit 144 was of rapidly eroded gravel. On the south side a lens of a subsequent more slowly

formed fill remained but otherwise the greater part of the pit had been truncated first by ditch terminus 143 and

then by pit 142. The base and middle fill of pit 720 comprised coarse gravel (1086) sealed by slowly formed

loam (1085). It contained 5 substantial Roman sherds and was cut by a broad, shallow hollow, 719, which in

turn was cut by the curvilinear gully 2184. Pit 720 was poorly defined in plan and but its steep sides would have

been consistent with either a single or a grouped post setting.

Table 11. Later Roman pits with one episode of rapid deposition

Cut Fills Shape in plan Diameter Depth Dating evidence (m) (m) 639 Oval concave bowl Association 720 1085-6 Gully segment? 0.75 0.40 Roman pottery Post holes There is only limited patterning visible in the distribution of post holes dated to the period. Large post pits, 646

and 708, were packed with large stones around central pipes and each contained exclusively Roman sherds. The

two post holes appear to form a continuation of a line formed by gully 748 which was 0.32m deep at its terminal

18 end and included two sherds of Roman pottery. Post hole 708 cut the fills of curvilinear feature 2184. Gully 748

may have been a slot for upright timbers although there was no supportive evidence such as packing stones. As

noted above, the possible packed post hole, 841, may have provided support for a cover over an activity area.

In the central area of the site several pits post holes lacking any direct dating evidence have broadly similar

profiles to the neighbouring post hole including Roman pottery, 1534. They might have formed a fence line (F1)

extending south eastwards from pit 1533, although the suggestion is tentative. Much the same can be said of two

post holes further to the south, 1824 and 1835, whereas further south again 2305 is an outlier of what is

probably a group of Medieval postholes.

Table 12 Later Roman post holes

Cut Fills Shape in plan Diameter Depth Dating evidence Other finds (m) (m) 105 163 Irregular bowl 0.55 0.13 Roman pottery 142 257 Concave bowl 0.59 0.15 Stratigraphy 646 1067, 1068, 1069 Concave bowl 1.15 0.40 708 1071 Flat bowl 0.63 0.32 Roman pottery, stratigraphy ?worked stone 808 1176 Concave bowl 0.62 0.19 Roman pottery 841 1268, 1269 Concave bowl 0.53 0.21 Association 1534 3099 Concave bowl 0.29 0.10 Later Roman pottery 1824 3352 Concave cylinder 0.44 0.30 Roman pottery 1835 3298 Concave cone 0.28 0.15 Roman pottery 2305 3798 Concave bowl 0.50 0.22 Later Roman pottery

Corn dryers (Figs. 15 and 16)

The droveway suggests that stock management was part of the local economy but two well-made corn dryers

and a handful of quern fragments demonstrate the importance of cereal production. At least two ovens likely to

be of Roman date testify to further food processing.

The two corn dryers were found in the north eastern area of the site set approximately 50m apart. Both

were very similar T-forms, constructed using large, angular, rough-hewn sandstones within purposely excavated

pits. At Catsgore, Somerset, and Alington Avenue, Dorset, the cross of the ‘T’ and in the case of the latter, axial

ancillary chambers were described as flues (Davies et al. 2002, 74; Leech 1982, fig. 55). The descriptions of the

corn dryers follow the nomenclature used for Balksbury Camp, Hampshire, where the cross of ‘T’ was described

as the ‘drying compartment’, the proximal end of the trunk of the ‘T’ was described as the ‘stokehole’ and the

distal end as the ‘flue’ (Wainwright and Davies 1995, 26).

Corn dryer 1 (Fig. 15) This had been constructed in cut 822. The drystone walls, 1165, of the drying compartment at the north-west

end survived to a height of 0.42m in irregular courses determined by the size and shape of the stones. The

combined flue and stokehole had been heavily degraded, surviving in the main as a single course on its south-

west side and only sporadically as a single course on its north-east side (Fig. 15; Pl. 4). The grey brown silty

sand fills of the drying chamber (1195, 1196 and 1197), included large coarse stones collapsed from the walls.

19 Most of those closest to the base had been burnt to a heavily oxidized red hue but those above showed much less

evidence of exposure to fire. More finely hewn, broader slabs with little trace of burning were sandwiched

between the burnt and unburnt coarse rocks. The deep upper fill of stokehole/flue was a continuation of (1197),

the dominant fill in the drying chamber. The upper part of it was grey silty sand, giving way a thicker deposit of

yellow, slightly clayey sand. At the distal end stones were few and rarely burnt but at the proximal end the

stones were much more frequent, larger and frequently burnt to a red hue, in particular those close to the base.

The deposit sealed a thin charcoal-rich seam which lay directly over the cut. Two quern fragments lay beyond

the east end of the flue.

Corn dryer 2 (Fig. 17) This had been constructed in cut 1048, the drying compartment forming its south-west end. The drystone walls,

1499, of much of the drying compartment and stokehole survived to heights of up to 0.54m in irregular courses

determined by the size and shape of the stones (Fig. 17; Pl. 5). Elements included a transverse wall set apart

from the distal end of the flue, an arrangement which is analogous with a more complex corn dryer at Alington

Avenue (Davies et al. 2002, 74-7). The main body of fills in the drying compartment and the stoke hole

comprised up to 0.45m of yellow clayey sand including a few medium to medium large sub-angular stones,

often burnt red, set over a gritty silty sand containing frequent charcoal lumps and flecks. The flue had either

been of a more open construction or its side walls had been removed. The 0.15m deep upper fill was a

continuation of that of the stokehole but there were 10mm to 15mm thick intermediate layers comprising gritty

to gravelly brownish grey loam including charcoal and occasional sharp fragments of red brick and tile. The

deposit over which it lay was very similar to the upper fill. It sealed a 0.08 thick charcoal-rich basal deposit.

The stokehole walls of both corn dryers were burnt deeply red as were the opposing middle sections of the

drying compartments. The chambers of the corn dryers would have been sealed by paving slabs in the manner of

those at Catsgore, Somerset, which were raised slightly above the floor level (Leech 1982, 65-67; figs. 55-8). At

that site the dryers were set under rectangular chambers within stone buildings. Some slabs remained to cover

one end of the drying compartment of a dryer at Balksbury (Wainwright and Davies 1995, 26; fig. 35). The

upper fills of the stokehole included mortar which was presumed to derive from a fallen superstructure. The corn

would have been laid on the sealing slabs to dry, using a hypocaust system.

Table 13. Later Roman corn dryers

Cut Fills Shape Length Width Depth Dating evidence (m) (m) (m) 822 1165 (wall), 1195-7, 1363, 1459 T-form 3.16 2.68 0.42 Late Roman pottery 1048 1499 (wall), 1552, 1597-8, 1659-60 T-form 5.06 3.0 0.54 Roman pottery

20 Ovens

Oven 1047 was constructed within a cut into natural geology and the drying compartment wall of corn dryer 2

(Fig. 16). Its full plan could not be ascertained as it had been cut through the middle by a possibly Medieval

ditch, 2173. The fire pit at the southern end was lined with gravelly yellow clay (1596) the inner surface of

which had fired to oxidized red and which incorporated some medium large sub-angular stones. The fill within

the lining comprised greyish brown gritty to gravelly silty sand with an increased incidence of charcoal on its

southern side. To the north a dark loam including charcoal and burnt stone (1595) filling a cut was interpreted as

rake-out from the oven.

The layout of a second oven, 1117, is typical of linear Roman field ovens with one bulbous end (Pl. 6) and

despite a lack of datable finds it may be dated to this period with high confidence. The stokehole sloped

downwards from south to north at about 10o for a distance of 0.42m before dropping steeply and levelling over

0.40m before merging into a 0.72m diameter fire pit which reached a maximum depth of 0.23m on its north side

(Fig. 16). The sides of the stokehole’s northern end and that of the entire fire pit were fire reddened and there a

0.05m to 0.08m deep charcoal-rich deposit (1586) along the bases of both elements. The surviving lower part of

a 0.12m wide flue breached the western side of the fire pit (Pl. 7). Several substantial quern fragments lay

directly over deposit 1586 and were sealed by a mass of grey gravelly, sandy, clay which may have been

collapsed superstructure although it lacked the oxidized colouring of the sides. Three such ovens, G, H and J

were found amongst other ovens within Late Roman building 562 at Balksbury, Hampshire. The two best

preserved examples had 0.30m deep fire pits and all three had stokeholes sloping into their pits (Wainwright and

Davies 1995, 25–6).

A short gully-segment-like pit, 619, had a basal fill of gritty loam over which lay a charcoal-rich lens

sealed by up to 0.20m of clean sandy clay, some oxidized. The clay is probably the collapsed lining of an oven

within the pit. There is no evidence for an elaborate superstructure or repeated use, and there was no evidence

for in situ burning of the sides.

Table 14. Later Roman ovens

Cut Fills Shape Length Width Depth Dating evidence (m) (m) (m) 619 961, 962, 964 Uncertain Roman pottery 1047 1594, 1595, 1596 Figure-of-eight 1.89 0.64 Stratigraphy 1117 1586, 1587, 1588 Lollipop 1.52 0.72 0.23 Form

Graves The acidity of the soil precluded the survival of bone but several features had characteristics which are typical of

graves. Three oblong pits, 100, 218 and 417 (Fig. 14), contained nails and two of these, 100 and 218 (Pl. 9), in

21 addition contained hobnails from boots, retaining the form of the sole. Several other pits had rectangular forms

and dimensions suited to use as graves. Pit 100 contained exclusively Iron Age pottery but sherds from 4, 149,

214 and 218 were all Roman. Pit 4 was perceived to be cut by ditch 2151 and its exclusively Roman pottery

contrasted with the entirely Iron Age pottery from the ditch.

The grave lengths and forms suggest that all the burials were extended rather than flexed and that three of

the bodies were in wooden coffins. Whilst the use of coffins, especially in association with boots, is generally a

later Roman phenomenon extended inhumation in graves also occurs in the Late Iron Age, perhaps most

famously and most securely dated in the so-called War Cemetery at Maiden Castle, Dorset (Whimster 1979,

356-62). Other examples of possibly the same date from the region have been found at Dorchester and

Osmington (Whimster 1979 343, 347) and, closer to Somerley, below Battery Hill, Winchester, where a tanged

knife was present (Whimster 321-2). If all the graves are held to be of late Roman date, convention would have

it that graves 149, 214 and 218 were Christian, based on their broadly west to east orientations and the others

pagan. The iron blade in 214 might be regarded as undermining the Christian attribution but it is not

unprecedented. A slight central ridge along the long axis of 215 suggests that two people had been interred side

by side (Pl. 9).

Table 15 Later Roman Grave pits

Cut Fills Shape and orientation Length Width Depth Dating evidence, finds (m) (m) (m) 4 55 Flat lozenge, concave sided. South - north 1.60 0.90 0.25 Later Roman pottery, fired clay 100 158 Flat, sub-rectangular. South - north 1.90 0.85 0.40 Later Iron Age pottery, nails, hobnails 149 270 Flat lozenge, straight sided. West south 1.90 0.85 0.18 Roman pottery west - east north east 214 296 Concave, sub-rectangular. West south 1.80 0.70 0.20 Roman pottery, iron blade west- east north east 215 297 Flat rectangular. South west – north east 1.25 0.80 0.05 Form 218 377, 378, 379, Flat, sub-rectangular. West - east 2.40 1.20 0.60 Roman pottery, nails, hobnails, fired 380, 381 clay 417 579 Concave, subrectangular. South south 2.20 0.82 0.22 Nails east – north north west 938 1379 Concave, subrectangular. South west – 1.60 0.62 0.15 Form north east

Phase 5 Medieval (Figs 17-21)

The Medieval period is defined by two distinct extensive but related systems and a minimal later pattern which

dissolved into the Post-medieval and modern landscape. The first two phases are characterized by field systems

constructed around a small group of buildings which probably combined domestic and agricultural use. The

pottery and sparsely distributed ceramic building materials do not allow close dating but key stratigraphic

relationships demonstrate a succession of coherent systems. Unfortunately, this is not the case for the structures

as the single feature with a discernible relationship between buildings B1 and B4 might have belonged to either

building, whilst B2 could fit into either or both phases based on its position and orientation. However, as will be

22 shown below, it is unlikely that certain important elements of B1 would have co-existed with some of those of

B4. It has been assumed that the initial Medieval phase followed a period of low intensity land use reflected in a dearth of artefacts. Thus few finds are likely to have been captured during the first phase of ditch-digging and construction. As soon as domestic activity commenced detritus entering open ditches and pits may include durable materials such as pottery or tile. Such detritus may have been captured during later phases of construction or ditch digging hence the later a building the more likely it is to include artefacts, some probably residual. The pottery from buildings B1-4 is all likely to date from the 11th-12th centuries AD (Timby, below).

Imported glazed wares dating to the last decade of the 11th century have been found at Winchester (Pearce et al.

1985, 128) but vessels were being glazed in London by the end of the 11th century and the process became widespread in the ensuing decades. For the purposes of the Medieval phasing on the present site it has been assumed in most instances that features including glazed material or tile are later than those from which only unglazed pottery has been recovered. Obviously the assumption is weakened by the very limited amount of material recovered generally and from the structures in particular.

Medieval phase 5a (Fig. 17)

The first Medieval phase is represented by three sides of a large field with some partially sub-dividing ditches and tracks all of which shared their orientation with a series of timber-built structures, to form a single coherent system, S9. The outline of the large field is given by ditches 2275, apparently extended by 2274, 2386, 2293,

2298 and 2296. Ditch 2273 to the north of 2274 was parallel with it, hence has been treated as contemporary.

The moderate amounts of pottery, including 15 substantial sherds, were exclusively from ditch 2293. Green glazed tile of possibly 13th to 14th century found in 2293 and 2296 suggests that the system had a longer life than the buildings first associated with it (Milbank, below). The system has a very consistent orientation with strongly coherent elements. Although stratigraphic relationships were determined only rarely, relationships with later features support the phasing. Ditch 2275 was cut by Phase 5b ditch 2289 and post hole 1830. To the west of the southern end of ditch 2293 and at 45o to it, ditch 2351 contained substantial Medieval sherds and was cut by Phase 5b ditch 2352. A broad ditch 2360 on the east side of the terminus contained two large Medieval sherds. The gap between 2293 and 2360 may have been a gateway. Gully 2287 appears to have formed a second gateway with the terminus of 2386 and was probably used for the sorting or movement of livestock. The fills of

2287 included two substantial Medieval sherds and it was cut by Medieval phase 3 ditch 2284. Gully 2297 produced no pottery but it shared its orientation with boundary 2296.

A line of three post holes west of 2351 was parallel with ditches 2275, 2297 and 2296 and may have formed a fence line, F2, but no direct dating evidence was recovered. A single Medieval sherd was recovered

23 from post hole 2310, one of three possibly forming a short fence, F4, which would have guided movement between B4 and B2. It is very likely that some of the pits and post holes north of the two buildings were contemporary with them. A distinct dearth of features in the crook of the ‘L’ formed by the buildings may mark where there may have been a yard. Lines of substantial posts on the western and northern sides may have bounded the space leaving an opening at the northern corner which would have provided access to the larger system via an opening at the terminus of 2298. Three substantial Medieval sherds were found in the easternmost of six post holes continuing the line of ditch 2275, forming a fence, F3, which is likely to have been contemporary with the ditch. A kink in the ditch accommodated the westernmost post, implying that the fence was already in place when the ditch was dug. A post hole 1904, slightly north of the ditch, included a single

Medieval sherd.

Timber structures (Fig. 18)

A sequence of medieval timber buildings can be recognized (four or five in at least two phases), four of which appear to share similar construction techniques, although very different ground plans. The building sequence appears to be B4 and B2; followed by B3 and B1; with B5 existing separately. The buildings are described in terms of both wall lines (A–J) and individual cuts (numbered).

Direct stratigraphic information concerning relative phasing of the Medieval buildings was limited to a group of intercutting pits or post holes, 2312-5, at a point where the line of B1’s west wall intersected with (and cut) B4’s north wall. In the group the latest cut, 2313, is slightly south of the line of B4’s north wall but is on the line of B1’s putative west wall. A further post might be expected between 2314 in the 2m gap between 2314 and wall trench terminus 2208. It is highly unlikely that wall J (of B4) and the southern section of B1’s wall F would have co-existed, so it is more likely that wall J replaced that section and that the surviving middle section of J was reinforced by postholes 2045 and 2043. However, it has been assumed the B4 was a discrete building pre- dating B1 based on the relationship between postholes 2313 and 2314 and the absence of tile from the features which comprised it.

The west wall of B4 is represented by post trenches J and K but the other walls depended on earth-fast posts. The trenches were interpreted as sill beam trenches during excavation but post holes cutting both the trench fills and their bases suggest that a post-in-trench construction is more likely (this is more clearly demonstrated for B1 in the next phase). Trench J was truncated (by B1) at a length of 2.20m but the full length of K survived at 2.7m. Had the two trenches been of the same length there would have been a 0.80m wide doorway with a door hung from an inner post, 2325. There was no evidence for an opposing entrance on the west side. There was marked variation in the depth of the two trenches (Table 16). The external dimensions of

24 the building would have been 4.50m by 6.20m. In other respects there were notable patterns. The important load-bearing mid posts at each end, 2314 and 2321, were of comparable scale, although 2314 may have replaced the shallower 2315, as were three corner posts at either end of the west wall and one south of trench K. The off- centre end posts which would have borne lighter loads were of smaller scale. The internal supports are problematic. It is not unprecedented for a single line of aisle posts to be off-centre (Gardiner 2000, fig. 3) but the apparently stepped line on either side of the centre is unusual. However, it is possible that a third post was erased by B1’s trench E which would have given a space of 1.20m between the end walls and the first aisle posts at either end. Post 2312 would then have provided additional support at the north end, the east side of which may have been a 2m wide entrance.

Table 16. Building B4, dimensions of post pits

Element Cut Diameter (m) Depth (m) Corner posts 2300 0.46 0.10 2301 0.49 0.17 2311 0.55 0.14 Mid end posts 2314 0.53+ 0.32 2315 0.24+ 0.13 2321 0.52 0.60 Aisle posts 2316 0.26+ 0.09 2317 0.28+ 0.20 2324 0.41 0.23 Aisle post? 2312 0.30+ 0.08 Mid west wall 2234 0.40 0.07 2249 0.42 0.10 Other end wall 2309 0.28 0.07 posts 2224 0.22 0.06 Door post 2325 0.50 0.33 Post-trenches J 0.50 0.06 - 0.12 K 0.50 0.20 - 0.27 B2 comprised two parallel trenches, G and H, sharing their orientations with B1 and a short trench, I, which adjoined the east end of G. The fully-exposed G was 6.3m long with a maximum of 0.70m, tapering towards both termini and depths varying from 0.04m to 0.15m. Trench I was 1m long and 0.40m wide with a depth of

0.06m. The southern trench, H, was set slightly further east than G. A length of 5.3m was exposed from the western terminus eastwards but the eastern terminus was beyond the limit of excavation. Its width varied from

0.40m to 0.45m and the depth from 0.10m to 0.19m. The external width of the structure as represented by the trenches was 3.70m. At the time of excavation one post hole was identified cutting G and a further two cutting

H. The relationship of post hole 2243 to slot 2242 was unclear but the west terminus of G cut a pit which in turn cut a post hole. Two sharp downward steps in the base of G and one in I strongly suggest three more cuts along the courses of those trenches. One of two post holes forming a parallel line with H within its south-west end contained two Medieval sherds and a single sherd was recovered from H. It is possible that the east terminus of I accommodated a cross-entrance, the opposing terminus of which was truncated by the broad Medieval ditch

2358 although the resulting structure would have been disproportionately narrow.

25 Medieval phase 5b (Fig. 19)

In this phase it appears that the large westward looking enclosure is abandoned in favour of an equally large east-facing enclosure, although the trackway leading west is maintained, and the buildings are replaced in more or less the same place.

Ditches 2286 and 2289 form the north and west sides of a large enclosure extending east beyond the excavated area. Its south side was probably defined by ditches 2352 and 2360, with an entrance gap between these, or it may have been open on this side. Ditch 2294 formed an internal subdivision.

Several ditches converge to form gates, trackways and races indicating that the space was a particular focus for the sorting and movement of animals. Substantial Medieval sherds were recovered from both ditches 2352 and 2353 which formed a possible track or race close to the southern terminus of 2286. Ditch 2352 is in effect a continuation of the broad ditch 2360, which probably survived from the previous phase, and cut ditch 2351, indicating that the latter did not survive. Gaps between 2352 and the southern terminus of 2286, the termini of

2352 and 2360 and between the southern terminus of 2294 and 2360 may have been gateways.

Post holes 2038 and 2201 contained Medieval pottery and together with other undated postholes in the same area they may have formed a livestock holding pen. A 1.08m wide, 0.50m deep pit, 2149, lay between the two dated post holes and contained a single Medieval sherd. It was cut by post hole 2200 of phase 5c building,

B5.

Further north pairs, of undated post holes either side of a probable gate between 2286 and 2289 may have controlled passage through it.

Two ditches, 2299 and 2350, in a near perpendicular relationship to 2296 and parallel to each other contained no dating evidence but defined a track of this, the following phase or both. This track is related to trackway 1, identified during an earlier phase of archaeological work when tile and the latest Medieval pottery from the site were recovered gully 1001 (Timby 2004, 7; Oram 2004, 8, fig. 3). The track formed part of an extended field system to the south of the current site.

The fence F2 may have been of this phase or have remained in use from phase 1. It is close to the orientation of 2352 and a further post hole could be added to it towards the east end of the ditch.

Four nearby pits contained a few substantial Medieval sherds and have been allocated to this phase based on their proximity to F5. For similar reasons other pits and a post hole containing Medieval pottery, 2336

(proximity to F5), 2401, 2508, and 2510 (proximity to B3) have been allocated to this phase.

A gully 2400 and neighbouring pit 2401, which it cut, both contained small sherds of Medieval pottery. An

8m long, 0.70m wide, 0.20m deep gully, 2527, containing a single Medieval sherd was adjacent to 0.90m wide,

26 0.13m deep gully 2528. There was no discernible relationship between the two but it is likely that one re-cut the other.

Timber structures (Fig. 18)

Two further buildings were added in this phase, again defined wholly or in part by wall trenches, and with B1 almost certainly replacing B4: it is not clear if B2 survived both phases but the presence of a ditch (2358) seems to imply the demolition of B2.

Trenches on three sides defined an east south east oriented aisled structure, B1, with external dimensions of

8.4m by 5.4m (Pl. 10) which was constructed in the relatively open space of the possible yard. Trenches on both sides of the long axis, A, B, D, E, and at the eastern end, C, gave the outline plan and a sixth trench west of the building’s centre, F, created a discrete chamber within it of 3m by 5.4m at the west end. As for B4, the west wall appears to have been supported on free-standing posts with no trench. Unusually, the two breaks in the trenches in the two long walls (0.80m wide between A and B; 0.70m wide between E and D) were asymmetrical and also at slightly different distances from the eastern wall, so that the cross-entry would have been on a slant. This would have rendered the hall area small in relation to the service area. The trenches had straight, steeply sloping sides and fairly flat or slightly undulating bases. The axial trenches generally varied in width from 0.40m to

0.60m, although there was some narrowing at termini (Table 17). The trench at the east end tended to be narrower with a width range of 0.35m to 0.45m and the trench dividing the hall from the chamber was narrower still, ranging from 0.30m to 0.35m. The outer trench depths ranged from 0.28m to 0.10m. It is notable that the variation coincides with a general reduction in depth from north to south rather than with particular trenches, suggesting either that depths were determined by the lie of the land at the time of construction or that they have been eroded. There was no evidence for a hearth, at least one of which normally would be within the hall area but this may be due to the same erosion. A total of only 19 Medieval sherds (112g) and four tile sherds (90g) restricted to slot 2108 were recovered from the constituents of B1.

Table 17. Building B1, dimensions of post-trenches

Beam trench Beam trench length Cut Width (m) Max depth (m) A 4.30 2205 0.60 0.28 2203 0.60 0.21 2204 0.60 0.23 2030 0.40 0.24 2031 0.33 0.25 B 3.00 2106 0.40 0.22 2108 0.47 0.23 2110 0.55 0.14 2111 0.55 0.15 C 4.60 2113 0.40 0.09 2137 0.35 0.13 2138 0.40 0.13 2139 0.45 0.27 2140 0.40 0.25 D 1.95 2126 0.50 0.10

27 2129 0.50 0.22 E 4.80 2210 0.55 0.17 2213 0.60 0.15 2230 0.55 0.13 2228 0.50 0.19 F 6.40 2212 0.30 0.17 2144 0.35 0.27 2044 0.30 0.12 2029 0.30 0.05 2330 0.30 0.12 There were six large, broadly symmetrically arranged post pits in the interior. Their sizes appear to imply they

were for more than simply screen walls, but whether they supported a roof that was not supported on the

external walls, or an upper storey, is unclear. The diameters of the surviving aisle post pipes ranged from 0.36m

in 2032 to 0.49m in 2100, although it is likely that the post in 2147 would have been smaller (Table 18). Posts

2049 and 2147 may have been replaced or reinforced respectively by 2048 and 2148. Cut 2047 seems to have

been excavated into the fills of a tree throw which may have rendered that post particularly susceptible to decay.

The other earth-fast posts comprised three from inside corners of the building and three forming the supports for

the west wall, at least two of which, 2312 and one of 2221 and 2220 were replaced or reinforced respectively by

2313 and one of 2220 and 2221. All of these posts were small, although those from the wall tended to be slightly

more substantial. Post 2313 is of particular significance as it cuts both 2314 of B4 and 2312, so that B1 post

dates B4.

Table 18. Building B1, dimensions of aisle post pipes and post pits, corner posts and west wall posts

Element Cut Cut diameter Depth Pipe diameter Aisle 2032 0.71 0.50 0.36 2049 0.77 0.57 N/A 2100 0.77 0.48 0.49 2105 0.57 0.52 N/A 2147 0.60+ 0.40 N/A 2209 0.68 0.57 0.45 Corner 2115 0.12 0.07 N/A 2125 0.27 0.07 N/A 2232 0.12 0.05 N/A West end 2312 0.30 0.08 N/A 2313 0.35 0.19 N/A 2302 0.26 0.13 N/A 2220 0.14 0.09 N/A 2221 0.16 0.10 N/A Other post holes were identified at irregular intervals within the wall trenches. They present some difficulties for

the characterization of the construction process. The advantage of a sill beam is that it provides a relatively level

surface which delays the basal rotting of upright timbers. As such they lengthen the life-expectancy of a

building. The basal ends of the upright would be narrowed to form a tenon which would be inserted into a

socket or mortise in the sill. Eight of the nine posts sockets identified in the trenches would not only have fully

penetrated the sill but also the natural below it (Table 19). Furthermore, post settings at trench termini, 2107,

2112, 2141 and 2129 were either on the very edge of the trench or only slightly within it so that there would

28 have been little or no timber socket. It is more probable that trenches were for upright posts rather than sill beams.

Table 19. Building B1, dimensions post holes cutting post-trenches

Beam trench Post cut Diameter (m) Depth (m) Depth above (+)/below(-) beam base (m) B 2107 0.36 0.25 -0.03 2109 0.32+ 0.14 -0.04 2112 0.47 0.19 +0.04 C 2141 0.35 0.27 -0.03 D 2127 0.33 0.13 -0.02 2130 0.50 0.25 -0.03 E 2211 0.28 0.18 -0.01 2229/31 0.33 0.21 -0.04 F 2145 0.46 0.31 -0.03 Two post holes, 2334 and 2329, and a gully or post-trench, 2403, continued the line of building B1’s post trench F and may have been the foundation of a substantial fence, F5, projecting from the north of the building and also aligned with ditch 2286.

An 8m long, 0.70m wide, 0.20m deep ditch 2358 parallel to and east of the east end of building B1 contained three moderate Medieval pottery and five large tile sherds. The ditch showed traces of re-cutting at its north end and may have existed through the life of both phases of buildings, although it would have been uncomfortably close to B2 and could not have accompanied the latter if the building did extend further east as posited, so the ditch has been taken to be newly created when B1 was constructed.

A second posited building, B3, would have lain mostly outside the excavation to the east and is represented solely by the junction of two post-trenches forming a north corner of what is likely to have been a structure of similar orientation to B1, but if so, only one wall and the beginnings of a second were within the excavated area.

Five Medieval sherds were recovered from a trench slot and one from post hole 2517 which cut the fills of the west trench. Two further post holes, 2525 and 2513 also cut the trench fills and the north trench cut the west trench. The very variable depth of the trenches from 0.13m to 0.30m over short distances excludes the possibility that they were beam slots. Any of three post holes in the vicinity may have formed a gate with gully

2400, with 2401 the most setting for a gate post. Both the latter features included Medieval pottery.

No relationship was established between 2400 and 2358 but the latter included both Medieval pottery and five substantial tile fragments.

The ‘mature later medieval domestic plan’ typically comprised a hall and services divided by a cross-entry with a chamber on the other side of the hall, often concealed behind a screen or dividing wall. Three early sill- beamed ground plans have been distinguished: 1) un-aisled hall with chamber and services; 2) one or two aisled hall with chamber and services; and 3) smaller structures with cross-entrances lacking clear signs of other internal subdivision (Gardiner 2000, 170-4). The use of sill beams allowed the introduction of cruck-built structures by the later 13th century (Dyer 1986, 36; Currie and Fletcher 1972). However, an aisled plan is

29 usually indicative of truss rather than cruck construction. The timber framework of a probably 13th-century

building incorporated into a later house at Boxted, Essex, is of a similar scale at 8m by 6.8m, to B1 although all

the uprights, including those of its two aisles, were founded on sill beams (Hewett 1973, 132). It has been

suggested that there may have been a standard architectural plan incorporating a discrete chamber set apart from

the main part (hall) of the building (Gardiner 2000, 177). The apparent lack of sill beams for a structure which

resembles a type 2 ground plan might imply that B1 is an earlier example of it or it may be a matter of regional

or local variation. At the Mid-Saxon village of Chalton, south east Hampshire, longer post-trench buildings with

central cross-entrances succeeded a building occupying the same footprint which had a frame set in discrete post

holes (Addyman et al. 1972, 17).

Fragments of tile may were found in features of the second and third Medieval phases. They would appear

to derive from buildings of higher status than those at Somerley and none of the identified buildings are likely to

have been covered or floored by them (Milbank, below). If they were in associated with the buildings it may

have been for a specific and limited purpose. Peg tiles were used in a sequence of hearths associated with a

probably 13th century flint-walled out house at (Harding and Light 2003, 141).

Medieval phase 5c (Figs 20 and 21)

Ditch 2284 of a pair with 2283, defining track S8, included a smattering of Roman pottery but cut gully 2287

which contained Medieval pottery. Slot 1818 of 2284 also contained undated ceramic building material. The

track is on an entirely different orientation to those of the preceding and subsequent phases but rectangular

structure B5 had a similar orientation to the earlier buildings. It also does not share their post-in-trench

construction (Fig. 21). It had external dimensions of 5.80m by 4.50m and was made of six posts, some of which

may have been replaced or reinforced over time. Post holes 2102 and 2104 in the middle of the southern wall

both cut Phase 5b gully 2294 and 2200 cut 2149, a pit containing a Medieval sherd. A Medieval sherd was also

recovered from 2200 itself. All the post holes were very substantial and had clearly defined large post pipes. 23

substantial Medieval sherds and three large 13th-14th century tile fragments were found in pit 1730, west of the

building (just beyond the area shown in Fig. 21). Some pits allocated to Phase 5b might equally belong to Phase

5c.

Table 20. Building B5, dimensions of post holes

Cut Fills Width (m) Depth (m) Post pipe diameter (m) Dating evidence 1748 3258-9 0.99 0.64 Visible in photo, c. 0.60 Medieval pottery 1804/5 3268-9 1.14m 0.70m 0.42 2102 3658-60 0.93 0.67 0.55 Stratigraphy 2120 3691-2 0.93 0.57 0.57 Roman pottery 2134 3696-7 0.87 0.59 0.45 Association 2200 3750-2 0.90 0.48 0.37 Medieval pottery, stratigraphy

30

Table 21. The Medieval phase 5c pits

Cut Fills Width (m) Depth (m) Dating evidence 1730 3194 1.47 0.25 Medieval pottery, tile Despite the very substantial construction of the small structure the final Medieval phase represents a diminishing

in the intensity of land-use and of settled occupation of the space.

Post-medieval (Fig. 22)

Ditch 2271 is the only substantial Post-medieval feature and has an orientation perpendicular to a modern track

(2260). It contained both Iron Age and Roman pottery but was dated by eight 15th-17th century sherds. Two

fence lines, F6 and F7, share its orientation but lacked finds.

The Finds

The Pottery by Jane Timby

This phase of the archaeological work resulted in the recovery of 4368 sherds of pottery weighing just over

62kg and with 39.50 estimated vessel equivalents (EVE) (Orton et al. 1993). The material is of mixed date with

sherds of earlier prehistoric, later Prehistoric, Roman, medieval and post-medieval currency present. Much of

the pottery, particularly the prehistoric and Roman sherds, is in poor condition with worn edges and surfaces

although the overall average sherd weight of 13-14g for the later prehistoric and quite high for Roman material.

The assemblage was spread across some 450 contexts from a total 381 cut features which mainly comprise pits,

postholes, ditches, gullies and corn-dryers. The quantity of material per feature varied greatly with 64% of the

contexts producing just 1–5 sherds and 73% with 10 or fewer sherds. This combined with the extreme longevity

of the use of the site and the complexity of the archaeology has made it particularly difficult to date, with a high

degree of confidence, a significant number of the features on the basis of the pottery alone.

The assemblage was sorted into broad fabric groups based on inclusions present, the frequency and grade

of the inclusions and the firing colour. The prehistoric assemblage was sorted into fabrics following the PCRG

(1997) guidelines where letters denote the main inclusions present, for example, FL for flint; GR for grog etc.

For the Roman material known regional or traded Roman wares were coded following the system advocated for

the National Roman reference collection (Tomber and Dore 1998). More local Roman wares were grouped and

coded in a similar manner. A summary of the defined fabrics can be found in Appendix 2. Sherds showing fresh

breaks made during or subsequent to excavation were counted as one where joins could be made.

31 The sorted assemblage was quantified by sherd count and weight for each recorded context. Rims were measured for diameters and percentage present to provide an estimated vessel equivalent (EVE). The resulting data was entered on to a Microsoft Excel spread sheet a copy of which is deposited with the site archive. In the following report the assemblage is discussed chronologically.

Early Prehistoric (Table 22) Forty-two sherds were designated as early prehistoric, several of which came from one vessel. This comprised

21 very fragmented sherds from a grog-tempered Beaker with a partially black sooted interior and cord- impressed decoration from pit 39 (91) (Fig. 23: 1).

Also present are two small grog-tempered crumbs re-deposited in a posthole from group 2355 (B1) which are likely to come from this or another similar vessel. Three coarse calcined flint-tempered bodysherds from urn- type vessels were the only finds from pit 402 and gully 411 and a further two similar sherds came from later pit

828. Fourteen sherds from a shell-tempered vessel (Fig. 23: 2) with a plain undifferentiated rim from pit 36 may also be of Bronze Age date. Further coarse shell-tempered sherds found in later levels may also date to this phase of use.

Later Prehistoric (Table 22) Some 3321 sherds, weighing just under 47kg, date to the early, middle and later Iron Age. There is a very diverse range of fabrics including sandy wares, flint-tempered, grog-tempered, calcareous and organic with various mixtures of inclusions in many cases. The vessels are all handmade. Overall the sandy wares dominate at around 45% by sherd count of the later prehistoric assemblage followed by calcareous (shelly) wares at 23.7%, flint-tempered at 14.6% and grog at 12.7%.

It has not proved possible to segregate all the features on the basis of the pottery alone either due to small numbers of sherds, lack of diagnostic pieces or the redeposition of material. The site is very complex and has seen a long history of use but certain groups can be isolated as characteristic of certain phases of activity. The earlier Iron Age assemblage is, in particular, characterized by pottery recovered from a number of pits and possibly three ditches (2151, 2157 and 2174). There are four sherds of haematite-slipped fine ware from pits

640, 641; ditch terminus / pit 339 and ditch 2151. The example from 339 (Fig. 23: 4) is from a flared rim bowl.

The tapered rim vessel from pit 917 (Fig. 23: 3), with lightly tooled decoration, may also belong with this group.

There are no obvious examples of furrowed bowls or scratch-marked bowls typical of the earlier part of the early Iron Age or other fineware forms. There are also no angular, situlate jars with finger-depressed decoration but there are sherds from more rounded forms with finger-tipped or other forms of decorated rims and sherds

32 with finger depressions around the girth. This might suggest the site dates from the very end of the early Iron

Age period, perhaps from around the 4th–3rd centuries BC.

Pit 620 with 101 sherds weighing 1813g is a key group for this period. The assemblage comprises sandy wares (SA1, SA2), sand and flint-tempered wares (SAFL); fine flint-tempered ware (FL2); sandy wares with organic matter (SAOR) or iron oxides (SAFE) and one vesicular shelly ware. Featured sherds include a jar with a slashed rim top; one with finger-nail decoration (Fig. 23: 5) and two with finger-tip decoration. One vessel has a grooved rim surface. There are eight sherds from vessels with finger depressions on the body (fabrics SA1FL2 and SA1) and one base sherd with a finger-smeared finish (fabric SAFE). Other vessels include a jar with an expanded rim (Fig. 23: 8) and two wide-mouthed jars with pronounced shoulders (Fig. 23: 6, 7).

The bulk of the later prehistoric assemblage appears to belong to the middle-later Iron Age characterised in particular by saucepan-style pots (Fig. 23: 10, 12, 14, Fig. 24: 22-3); beaded rim jars and bowls e.g. (Fig. 23. 11,

18, Fig 24: 21, 29-30, 34-5, 38, Fig. 25: 42, 44-5); simple or undifferentiated rim ovoid vessels (Fig. 24: 24, 28,

39) and necked jars and bowls with, and without, cordons (Fig. 23: 16, Fig. 24: 27, 31-3, 36-7, 40, Fig. 25: 43).

There are a number of decorated vessels including curvi-linear (Fig. 23: 9, 14, 16); impressed dot infilled panels

(Fig. 23: 15); incised and impressed tramline (Fig. 23:0. 19-20) and incised girth-grooves (Fig. 24: 26). A jar with simple curvilinear decoration in the Glastonbury style, but without the infilling, (Fig. 23: 16) comes with a radiocarbon date of 210–54 cal BC recovered from carbonized food residues adhering to it.

Less common are very large storage vessels as Fig. 24: 25 and small necked bowls (Fig. 23: 13, 17, Fig.

24: 26, 41). Most of the vessels in sandy fabrics have a burnished exterior and occasionally interior finish.

Nearly all the forms can be paralleled with mid-later Iron Age material documented from Hengistbury Head

(Cunliffe and Brown 1987). In particular, the necked cordoned jar (Fig. 24:.37) in a black sandy fabric of

Durotrigian type resembles some of the French imported jars (Cunliffe and Brown 1987, ill. 218). The assemblage lacks the more characteristic Durotrigian bowl and jar forms which start to appear at the very end of the Iron Age suggesting the site had fallen out of use by this time.

Overall sandy fabrics or sandy fabrics with other constituents such as iron oxides, flint and organic matter dominate the mid-later Iron Age assemblage. This compares well with the assemblage from Efford, near

Lymington (Jones 2007) which showed the same range and broad proportions of fabrics with sandy wares dominating followed by shelly, grog and flint-tempered fabrics.

Table 22: Prehistoric pottery summary

Group Fabric Description No No% Wt Wt% EVE EVE % Early Prehistoric EPFL coarse flint tempered 5 809 0 - EPGR Beaker 23 79 65

33 EPSH shell-tempered 14 137 36 Total 42 - 1025 - 101 - Later Prehistoric Flint FL1 coarse flint-tempered 76 2.3 1299 2.8 40 1.5 FL2 finer flint-tempered 71 2.1 1398 3.0 75 2.8 FLFE flint with frequent iron 10 0.3 77 0.2 0 - SAFL sandy with flint 283 8.5 3603 7.6 123 4.5 SA1FL2 fine sandy with flint 29 0.9 688 1.5 65 2.4 SA2FL coarser sandy with flint 17 0.5 400 0.8 13 0.5 SAFLFE sand with flint and iron 1 - 37 0.1 0 - SAFLH haematite-slipped 1 - 14 - 0 - Grog/flint GRFL grog and flint 5 0.1 79 0.2 7 0.3 Grog GR grog-tempered 403 12.1 8587 18.2 364 13.5 GRSA sandy with sparse grog 9 0.3 145 0.3 0 - GRSH/CA grog and calcareous/shell 3 0.1 75 0.2 13 0.5 Organic GROR grog and organic 34 1.0 136 0.3 0 - OR organic tempered 55 1.6 108 0.2 25 0.9 SAOR sandy with organic 4 0.1 39 0.1 0 - SAORH sandy with org haematite slipped 2 0.1 8 - 0 - Calcareous CA calcareous 13 0.4 236 0.5 15 0.6 SH1 coarse shelly 66 2.0 403 0.9 36 1.3 SH2 finer shelly 659 19.8 6852 14.5 307 11.3 SH2FL finer shelly with flint 15 0.4 185 0.4 0 - SA1SH finer sandy with shell 12 0.4 163 0.3 4 0.1 SAFESH iron-rich sandy with flint 1 - 4 - 0 - SAFECA iron rich sandy with calcareous 2 0.1 7 - 0 - Mixed SAFLOR sandy with flint and organic 1 - 33 0.1 0 - SAFLCA sand with flint and calcareous 3 0.1 19 - 0 - Sand DURO ?Durotrigian sandy 35 1.0 445 0.9 74 2.7 FESA ferruginous sandy ware 52 1.6 869 1.8 52 1.9 SA misc other sandy 30 0.9 280.5 0.6 0 - SA1 finer sandy 766 23.0 10392 22.1 618 22.8 SA1FE iron-rich finer sandy 32 1.0 566 1.2 35 1.3 SA2 coarser sandy 511 15.3 6683 14.2 582 21.5 SA2H haematite coated SA2 2 0.1 22 - 4 0.1 SA3 sub-angular quartz sandy 46 1.4 1016 2.2 42 1.6 SA4 fine sandy 2 0.1 182 0.4 10 0.4 TOTAL 3335 47100.5 2706

Roman (Table 23) The Roman assemblage comprises some 695 sherds of pottery weighing 9.5kg and with 9.5 EVE. The entire assemblage dates to the later Roman period. Recognizable traded wares are limited to just four sherds of Central

Gaulish samian (LEZ SA 2); sixteen sherds of Dorset black burnished ware (DOR BB1) and 115 sherds of

Hampshire grog-tempered ware (HAM GT). The samian comprises three sherds from two Dragendorff 37 bowls, one very abraded; and one sherd from a Dragendorff 31 dish. The DOR BB1, or BB1 related ware, is largely from plain or curved-rim dishes and at least one flat-rimmed bowl (Fig. 25: 47) whilst the HAM GT comprises almost exclusively necked and sharply everted rim jars (Fig. 25: 48).

The assemblage is dominated by grey medium to fine textured sandy wares which account for 33.9% of the late Roman assemblage. Some can be clearly recognised as products, for example, thumbed-pressed large storage jars (Fulford 1975, type 40.3); a jug (ibid. type 20) and a lid (ibid. type 23), unusual it that it has red-painted dot decoration. Most of the generic grey wares feature as jars but there is at least one rim from flanged-rim, conical bowl. Also likely to be New Forest products are the black-slipped grey wares which form

5.2% of the assemblage by count and include jars, some with burnished lattice decoration; flanged rim bowls; plain-rimmed dishes and a reeded-rim dish. Many of these grey wares are sooted from use. One grey ware

34 bodysherd from grave 218 is attached to a lead strap rivet. The grey sandy wares with sparse grog also appear to emulating New Forest types and may also be part of this industry.

Table 23 Roman pottery summary

Group Fabric Description No No% Wt Wt% EVE EVE % Imports LEZ SA Central Gaulish samian (Lezoux) 4 0.6 187 2.0 0 0.0 Regional DOR BB1 Dorset black burnished ware 16 2.3 282 3.0 51 5.4 Local/ BB1COPY imitation BB1 3 0.4 86 0.9 21 2.2 unknown BSGY black-surfaced grey ware 41 5.9 802 8.4 179 18.8 BSWW black-surfaced white ware 66 9.5 199 2.1 75 7.9 BWMIC micaceous black ware 1 0.1 17 0.2 9 0.9 BWSAFE black sandy wares with iron 1 0.1 9 0.1 0 0.0 BWSY black sandy 66 9.5 947 9.9 46 4.8 GY grey sandy ware 165 23.7 1567.5 16.5 173 18.2 GYFL1 grey sandy with flint 7 1.0 233 2.4 13 1.4 MISC misc sandy wares 9 1.3 48 0.5 5 0.5 GYF fine grey ware 1 0.1 1 0.0 0 0.0 NFO CC/RS New Forest cc/red-slipped ware 26 3.7 250 2.6 32 3.4 NFO WH New Forest white ware 1 0.1 7 0.1 0 0.0 NFO WH? white ware mortaria 5 0.7 99 1.0 0 0.0 NFO PA New Forest parchment ware 1 0.1 20 0.2 8 0.8 NFO RE New Forest grey ware 71 10.2 1011 10.6 107 11.3 OXID oxidised sandy 24 3.5 241 2.5 38 4.0 OXID2 well-fired oxidised 1 0.1 10 0.1 0 0.0 OXIDF fine oxidised 8 1.2 24 0.3 0 0.0 WSGY white-slipped grey ware 3 0.4 70 0.7 12 1.3 grog GRSJ grog-tempered storage jar 2 0.3 71 0.7 0 0.0 GYGR grey with grog 22 3.2 411 4.3 28 2.9 GYGRSA sandy grey with grog 36 5.2 1034 10.9 36 3.8 HAM GT Hampshire grog-tempered 115 16.5 1894 19.9 117 12.3 TOTAL 695 9520.5 950 Also present is a small group of New Forest specialist products including a single parchment ware (NFO

PA), and a flanged bowl (Fulford 1975, type 86) datable to the mid-4th century. There is a small group of 26 colour-coated wares largely from beakers (NFO RS 2; NFO CC) and six white wares, five of which are mortaria. Four of these appear to be standard New Forest examples with flint grits but one had rounded quartz trituration grits. Similar vessels have been found at the Plumley kilns, near Ringwood. As a group the assemblage is dominated by jars which account for 64.4% EVEs (Table 24) followed by bowls at 19.8%. The remaining 15.8% comprises lids, dishes, beakers and a flask.

Table 24 Roman pottery forms by Estimated Vessel Equivalent (x100)

Form EVE % EVE Jar 604 64.4 Bowl 186 19.8 Dish 56 6.0 Flask 35 3.7 Beaker 32 3.4 Lid 25 2.7 TOTAL 938

Late Roman pottery was associated with in the region of 78 cut features including pits, ditches, gullies, tree-boles, postholes and a single corn-dryer and grave. Only eight of these produced an assemblage of 25 or more sherds with the greatest concentration of material coming from pit 943 with 118 sherds. An added problem is the occurrence of residual later prehistoric material as seen in particular with grave 218 with a few later sherds

35 and the repaired pot but a much larger quantity of prehistoric material. There is also the difficulty in some cases, where single sherds were present, distinguishing later Roman grog-tempered wares from earlier grog-tempered ware.

Medieval (Table 25) Wares of medieval date account for 4.4% by sherd count, 5.7% by weight of the total assemblage. Most of the sherds comprise sandy wares with a single sherd of sandy ware with flint and at least one rim in a very vesicular shelly ware (MEDSH) (Fig. 25: 49). A distinction was made between a coarser sandy ware (MEDSA1) and a slightly finer variant (MEDSA2) but essentially these all seem to be variants within one industry.

Most of the vessels are handmade, unglazed cooking pots some with a scratch-marked finish (Fig. 25: 50.

52-7). At least one vessel has a thumb-pressed rim (Fig. 25: 58). A large number of the vessels have a sooted surface from use.

In addition there are a few sherds which come from jugs or pitchers with odd glaze splatter (e.g. Fig. 25:

51), in particular a strap handle, thumbed base and spout, probably from one vessel, in fabric MEDSA2 from pit

1730. A sherd with combed diagonal decoration came from pit 2038. Other less common vessels include a probably curfew from ditch 2295 (Fig. 25: 59) and a glazed dripping pan from ditch 2289 (Fig. 25: 60). The forms and fabrics fall within a similar range of gritty wares defined at Southampton broadly dating to the 11th-

12th centuries.

A large range of features contains pottery of this date so the quantity of sherds per feature is low. These include 17 pits, 15 ditches / gullies and the beam slots and postholes forming structures 2355 (B1), 2356 (B4),

2357 (B2) and 2359 (B3). The main structure, 2355 (B1), produced just 22 bodysherds, three are of which are re-deposited.

Table 25: Medieval pottery summary Fabric Description No No% Wt Wt% EVE EVE % MEDGL glazed sandy ware 1 0.5 7 0.2 0 0.0 MEDSA sandy ware 44 22.8 937 26.3 60 24.2 MEDSA1 coarse sandy 77 39.9 1253 35.2 80 32.3 MEDSA2 finer sandy 59 30.6 1178 33.1 103 41.5 MEDSAFL sandy with flint 1 0.5 7 0.2 0 0.0 MISC miscellaneous 2 1.0 9 0.3 0 0.0 SYHBW Surrey Hampshire Border ware 1 0.5 7 0.2 0 0.0 VER Verwood 8 4.1 164 4.6 5 2.0 TOTAL 193 3562 248

Post-medieval Eight sherds date to the post-medieval/ modern period. These include seven sherds of Verwood glazed ware from posthole 2133 and one sherd of Surrey-Hampshire Border ware from gully 2271.

36 Catalogue of illustrated sherds Fig. 23 1. Pale orange Beaker with a black, sooted interior. Cord-impressed decoration comprising horizontal lines and chevrons. Fabric: EPGR. Feature 39 (91). 2. Handmade, plain-sided vessel, possibly an urn. Brownish-grey in colour with a friable, very vesicular fabric. Fabric: EPSH. Pit 36 (88). 3. Handmade, thin-walled flared rim bowl. Dark brown in colour. Tooled line, slightly irregular decoration on rim and body. Fabric: SA2 finer. Pit 917 (1299). 4. Handmade, flared-rim, fine ware bowl with a waxy orange-red haematite slip on the exterior. Mid-brown burnished interior. Fabric: SA1OR. Ditch terminus/pit 339 (497). 5. Handmade, flat-tipped necked bowl. Finger-nail depressions on the upper rim surface and finger depressions around the girth. Red-brown in colour with black patches. Fabric: SA1 with some organic inclusions. Pit 620 (950). 6. Handmade, everted rim, jar. Patchy brown to orange-brown surfaces with a black core. Fabric: SA1FL (sparse very coarse flint stones up to 8-10 mm). Pit 620 (951). 7. Handmade, everted rim jar/bowl with a pronounced shoulder. Fabric: SA1FL. 620 (951). 8. Handmade, flat-topped, everted rim jar. Patchy dark brown to black in colour. Fabric: SA1FL. 620 (951). 9. Handmade, rolled-rim jar. Patchy orange-red to brown surfaces with burnished surfaces. Tooled curvi-linear line decoration. Fabric: SA1 (finer). Pit 1331 (1859). 10. Handmade, plain saucepan pot with intact profile (Hengistbury Head form PB 1.1). Burnished interior and exterior. Fabric: FL2. Pit 148 (262). 11. Handmade, round-bodied bowl with a simple undifferentiated rim. Highly burnished, black to orange-brown surfaces. Fabric: SA2. Pit 148 (262). 12. Handmade saucepan pot (Hengistbury Head form PB 1.1). Black burnished exterior with diagonal, burnished line decoration. Fabric: FL2. Pit 630 (975). 13. Handmade jar. Pale brown to grey with a burnished exterior. Fabric: GR. Ditch GN 2155 48 (156). 14. Saucepan style pot with tooled curvi-linear line decoration pot (Hengistbury Head form PB 1.1). Orange- brown exterior and black interior with smooth burnished interior and exterior surfaces. Fabric: SA2. Pit 1525 (3086). 15. Bodysherd from a decorated bowl. Rounded stabbed impressions probably set within a panel. Fabric: SA2. Pit 510 (693). 16. Handmade, necked, cordoned bowl. Black burnished surface with incised curvi-linear swag decoration. Sooted residue on the interior. Fabric: SA1. Pit 43 (97). 17. Handmade bowl with an incurving, slightly thickened, rim. Burnished exterior; worn abraded interior. Fabric: FL2. Pit 601 (861). 18. Handmade, slightly beaded rim jar (Hengistbury Head form BC 3.3). Burnished exterior. Fabric: OR. Pit 601 (861). 19. Handmade bowl with a beaded rim. Burnished black exterior and burnished orange/grey interior. Decorated with incised, infilled tram-line decoration. Slightly sooted in the interior. Fabric: GRSH. Pit 116 (176). 20. Handmade bowl, probably different but similar vessel to the one above. Oxidised exterior and grey interior. Decorated with tramline decoration. Fabric: GRSH. Pit 116 (176).

Fig. 24 21. Handmade, beaded rim, round bodied, bowl. Orange-brown surfaces. Slightly sooted exterior. Fabric: SH2. Pit 116 (176). 22. Handmade, saucepan-style pot (Hengistbury Head form PB 1.1). Red-brown with a grey interior. Smoothed but not burnished. Fabric: SA1. Pit 1635 (4154). 23. Saucepan pot, (Hengistbury Head form PB 1.1). Dark brown to black in colour with burnished interior and exterior surfaces, Fabric: SA2. Pit 207 (282). 24 . Handmade, large bowl with a simple rim. Burnished horizontally and vertically on the exterior and multi- directionally on the interior surface. Traces of burnt residue on the exterior. Fabric: SA2. Pit 207 (282). 25. Handmade, large jar with an expended rim. Pinkish-orange in colour. Fabric: FL1. Pit 207 (282). 26. Handmade, round-bodied jar with simple everted rim. Decorated with four parallel incised horizontal grooves. Burnt exterior. Dark grey in colour. Fabric: SH2. Pit 207 (282). 27. Necked bowl, handmade, wheel-finished. Black burnished finish. Fabric: SA2. Ditch 2187, 540 (787). 28. Handmade, beaded rim, globular bodied jar. Brown in colour. Burnished exterior; scraped finish to interior. Fabric: SH2. Ditch 2187, 540 (786). 29. Handmade, squat, beaded rim bowl. Black in colour with a burnished exterior. Sooted interior. Fabric: SA2/FL. Ditch 2187, 540 (786).

37 30. Handmade, globular bodied jar/bowl with an everted rim. Brown with a grey interior. Fabric: SH2. Ditch 2187, 540 (786). 31. Handmade, everted rim jar. Black burnished exterior shoulder and rim. Fabric: SA1. Ditch 2192, 416 (578). 32. Handmade, everted rim jar (Hengistbury Head form JD 4.11). Dark brown, burnished on the exterior and interior rim face. Fabric: GR. Ditch 2192, 416 (578). 33. Handmade everted rim jar (Hengistbury Head form JD 4.11). Dark brown to black with a burnished exterior. Sooted exterior. Fabric: GR (very lumpy texture). Ditch 2192 416 (578). 34. Very large handmade dolia-type jar (Hengistbury Head form JC 4.2). Orange-brown with a grey core and interior. Slightly sooted on the exterior. Fabric: SA3. Pit 624 (965). 35. Handmade bowl/jar with a slightly flattened top. Orange-brown to grey in colour. Fabric: SA1. Pit 544 (793). 36. Handmade necked everted rim jar. Patchy brown and grey exterior, oxidized red-orange interior. Fabric: SA2. Ditch 2155, 201 (376). 37. Handmade, wheel-finished necked, double-cordoned jar Hengistbury Head form JE 1.1).+ Black burnished surfaces. Fabric: SA2. Pit 603 (884). 38. Handmade, beaded rim jar. Black burnished exterior. Fabric: SA2. Ditch 313 (468). 39. Handmade jar/bowl with simple incurving undifferentiated rim. Orange brown in colour with a grey interior, originally with a burnished exterior finish. Fabric: SA2. Pit 1409 (1899).

Fig. 25 41. Handmade bowl with a slightly thickened rim. Roughly burnished exterior. Fabric: SA2. Pit 1445 (3050). 42. Handmade bowl with a slightly out-turned rim (Hengistbury Head form BD 4.4_. Black burnished exterior. Fabric: SA2. Pit 1445 (3050). 43. Handmade necked, cordoned bowl. Burnished exterior and interior rim face. Hole through the neck made after firing. Fabric; SA2. 1445 (3050). 44. Handmade, beaded rim bowl. Black burnished exterior. Fabric: SA2FL. Pit 1445 (3050). 45. Handmade, beaded rim bowl. Dark brown burnished surfaces. Fabric: GR. Gully 517 (753). 46. Handmade, beaded flange rim, curved-wall dish. (Hengistbury Head form BC 3.42). Black burnished surfaces. Fabric: as DOR BB1. Ditch 2289, 1644 (3158). Late Iron Age – early Roman.

Roman wares

47. Wheel-made, bead and flange rimmed deep dish. Black surfaces with a brown core. Fine sandy, slightly micaceous fabric. Fabric: Black sandy ware imitating BB1. Pit 643 (1052). 48. Handmade, sharply everted rim jar. Fabric: HAM GT. Pit 643 (1052).

Medieval wares

Fig. 25 49. Handmade expanded, flared rim, vessel. Grey in colour with a very vesicular fabric. Fabric: MEDSH2. 917 (1299). 50. Handmade, sharply everted, squared rim cooking pot. Fabric: MEDSA2. Pit 1730 (3194). 51. Spouted jug with a ridged neck and patchy glaze with a slight green tinge. Fabric: MEDSA2. Pit 1730 (3194). 52. Handmade everted rim cooking pot. Heavily sooted exterior. Fabric: MEDSA1. Pit 2025 (3499). 53. Handmade everted rim cooking pot. Slightly expanded, shaped, rim. Fabric: MEDSA1. Pit 2025 (3499). 54. Handmade, everted rim cooking pot. Oxidized. Fabric: MEDSA1. Pit 2025 (3499). 55. Handmade, everted rim cooking pot. Dark grey, smaller vessel. Fabric: MEDSA1. Pit 2025 (3499). 56. Handmade, everted rim cooking pot. Sooted exterior. Fabric: MEDSA1. Pit 2025 (3499). 57. Handmade, everted rim cooking pot. Pale grey in colour sooted in the exterior. Fabric: MEDSA1. Pit 2025 (3499). 58. Handmade, everted finger-tipped rim cooking pot. Pale oxidized sandy ware. Scratch-marked exterior. Fabric: MEDSA1. Pit 2117 (3590). 59. Handmade curfew with an expanded, flanged rim. Fabric: MEDSA2. Ditch 2295, 2018 (3489). 60. Handmade dripping pan. Decorated on the upper rim surface. Traces of a thin glaze on the interior surface. Sooted on the underside of the base. Fabric: MEDSA1. Ditch 2289, 1908 (3371).

38 Fired clay by Richard Tabor

A total of 740 fragments of fired clay weighing 35,869g derived from structural fragments and identifiable objects ranging from oven or furnace furniture to ‘loomweights’. The range of fabric mixtures exhibited examples of careful selection and others where large inclusions were clearly accidental. Material derived from

Middle Iron Age, Late Iron Age and Late Roman datable contexts although some of the latter is probably residual.

Fabrics The broad fabric groups comprise grog, quartz and sand mixtures and a distinctive fine silty matrix. Micaceous inclusions occurred across all groups and flint was absent only from the fine silty group (Table 27). All the fabrics comprised locally accessible materials.

F-mG1: Moderately hard, mid to dark grey laminate, sparsely micaceous fabric with buff pink outer surface including sparse poorly-sorted pink to grey grog (<2mm). F-feG1: Friable, pale grey fabric with buff pink outer surface including moderate to common poorly-sorted grog (<5mm) and reddish brown iron oxides (<2mm). F-fG1: Friable, laminate, pale grey fabric with buff pink to pale outer surfaces including sparse to moderate poorly-sorted rounded grog (<15mm) and rare angular flints (<2mm). F-femqG1: Moderately hard, mid to buff pink laminate, sparsely micaceous fabric including sparse poorly-sorted pink to grey grog (<2mm), red iron oxides (<2mm) and quartz (<0.25). F-fefmqG1: Moderately hard, mid to buff pink laminate, sparsely micaceous fabric including sparse poorly-sorted pink to grey grog (<2mm), red iron oxides (<2mm) and quartz (<0.25) and rare large flints (<20mm). F-mfeG1: Moderately hard, slightly soapy, buff pink, sparsely micaceous fabric including sparse grog (<2mm), rare red iron oxides (<1mm). F-feSG1: Friable, grey, sometimes laminate, sandy fabric including moderate poorly-sorted, often angular, grey grog (<10mm) and patches of reddish brown iron oxidisation. F-mQ1: Hard, fine grey fabric including sparse to moderate well-sorted rounded quartz (<0.25mm) and sparse mica. F-fQ1: Moderately hard, buff pink to mid grey fabric with pale pink to white surfaces including abundant well-sorted fine rounded quartz (<0.5mm) and rare fine or coarse angular or subrounded flint (1mm to 12mm). F-fQ2: Friable grey to brown fabric including abundant well-sorted fine rounded quartz (<0.5mm) and sparse poorly-sorted angular and subrounded flint (2mm to 40mm). F-fQ3: Friable yellow to grey fabric with yellow to buff pink surfaces including abundant well-sorted fine rounded quartz (<0.5mm) and rare to sparse poorly-sorted angular and subrounded flint (2mm to 40mm). F-S1: Hard, dark grey sandy fabric. F-mfeS1: Hard, grey sparsely micaceous sandy fabric with oxidised buff pink to grey outer surface including sparse brown iron oxides (<2mm). F-mfS1: Hard, grey sparsely micaceous sandy fabric with oxidised buff pink to grey outer surface including rare coarse flint (<15mm). F-mvS1: Moderately soft, sparsely micaceous sandy fabric with moderate subangular and subrounded voids (<4mm) probably formed following the dissolving of calcareous material such as calcite or more probably shell. F-mSi1: Moderately hard, sparsely to moderately micaceous silty, laminate, grey fabric with pale pink surfaces

Table 26. Quantification of fired clay fabrics No % sherds Wt (g) % wt Mean wt (g) F-feG1 10 1.4 85 0.2 8.5 F-mG1 91 12.3 854 2.4 9.4 F-mfeG1 40 5.4 1184 3.3 29.6 F-femqG1 66 8.9 898.5 2.5 13.6 F-fefmqG1 40 5.4 5177 14.4 129.4 F-fG1 20 2.7 2382 6.6 119.1 F-feSG1 8 1.1 100 0.3 12.5 F-mQ1 14 1.9 675 1.9 48.2 F-fQ1 47 6.4 2554 7.1 54.3 F-fQ2 90 12.2 5053 14.1 56.1 F-fQ3 194 26.2 10262 28.6 52.9 F-S1 6 0.8 320 0.9 53.3 F-mS1 10 1.4 879 2.5 87.9

39 F-mfeS1 17 2.3 523 1.5 30.8 F-mfS1 28 3.8 1644 4.6 58.7 F-mvS1 6 0.8 10 0.0 1.7 F-mSi1 53 7.2 3268 9.1 61.7

Some 650 of the 740 fragments (87.8% by count, 88.6% by weight) were from features dated by pottery,

association or morphology. 62.9% of the fragments (61.7% by weight) were from Middle Iron Age contexts,

17.2% (21.0% by weight) from Late Iron Age contexts and 19.8% (17.4% by weight) from Late Roman contexts

(Table 26). The lack of clear definition between the Middle and Late Iron Age phases may account for the

strong representation of certain fabrics in both phases, although the decline in mean sherd weights in certain

fabrics suggests that some are residual. All but those in F-fQ2 and F-mSf1 of the Late Roman fragments may be

so. The single large fragment in F-feSG1 was from a pit close to where examples in the same fabric were found

in Iron Age contexts. It should be noted that the Roman material F-mSf1 was from an oven, 1117, dated solely

by its form.

Table 27. Fired clay: Fabrics and mean fragment weight by datable contexts MIA LIA Late Roman Wt (g) Mean Wt (g) Mean Wt (g) Mean F-feG1 83 9.2 F-mG1 769 9.0 51 12.8 F-mfeG1 518 518.0 643 17.4 15 15.0 F-femqG1 331 11.4 103 14.7 11 5.5 F-fefmqG1 1910 91.0 3201 266.8 60 12.0 F-fG1 2382 119.1 F-feSG1 59 8.4 41 41.0 F-mQ1 138 27.6 252 84.0 F-fQ1 374 124.7 2180 49.5 F-fQ2 5053 56.1 F-fQ3 10262 52.9 F-S1 132 44.0 F-mS1 871 87.1 F-mfeS1 523 30.8 F-mfS1 1329 166.1 178 9.4 F-mvS1 10 1.7 F-mSi1 235 47.0 78 39.0

Morphology The assemblage is divisible into two overarching groups: structural material and tools. The former includes

fabric from hearths, ovens or furnaces and in this instance the latter is restricted to loomweights (Table 27).

Some fabrics may have been mixed with a view to the specific requirements of the use to which they were put.

All the bricks were in fabrics with finer inclusions, with the exception of a large fragment in F-mfS1. The bricks

were probably all oven furniture and needed properties suited to withstanding the highest temperatures. In

contrast, F-fQ2, probably exclusively from the floor of a corn-dryer, was a particularly coarse, friable, fabric.

The eclectic range of fabrics used for loomweights might suggest that some or all are indeed weights rather than

oven furniture since no special attention was given to their resistance to repeated thermal shock. Their possible

uses are discussed below.

Table 28. Fired clay: Quantification of morphological types by fabric Hearth/oven Corn dryer Brick Loomweights

40 No Wt (g) No Wt (g) No Wt (g) No Wt (g) F-mG1 8 52.0 1 249.0 F-mfeG1 2 570.0 F-fefmqG1 6 1506.0 7 3248.0 F-fG1 20 2382.0 F-mQ1 1 54.0 5 294.0 2 177 F-fQ1 39 2096.0 2 319 F-fQ2 90 5053.0 F-fQ3 194 10262.0 F-S1 2 165.0 F-mS1 10 879.0 F-mfeS1 1 256.0 F-mfS1 21 579.0 1 840.0 2 137.0 F-mSi1 1 859.0 8 750.0 19 746.0

Hearth, oven or furnace structural material

The only structural fragments demonstrably from features with which their uses were associated are those from

the Roman corn-dryers and a putative Roman oven. The remainder were generally dumped into pits, albeit

probably in close proximity to the places of use, given the volumes of material in several instances. Some of the

material is likely to have remained on the surface before becoming incidentally incorporated into the fills of

ditches and other pits.

The largest assemblage of fired clay associated with an oven or furnace structure was from two contexts in

Middle Iron Age pit 1331. The bulk of the material, 38 pieces (2374g) from (1857) and 156 pieces (7888g) from

(1859) was in the very coarse, friable fabric F-fQ3. Of these 37 had between one and six linear impressions left

by the framework over which the fabric had been daubed (Fig. 26: 1; Pl.11). The impressions criss-crossed at

variable depths in relationship to each other, implying that they formed a woven framework. Where sufficient

depth of impression remained to be measured diameters ranged from 8mm to 20mm. In general the impressions

had smooth surfaces but in a few instances both broad and narrow impressions had linear striations parallel with

their lengths (Pl. 12). The extensive surviving surfaces were smoothed, lightly oxidized and probably mainly

from the exterior (Pl. 13). Four fragments had a thin carbonized deposit on a rougher surface which may have

been the interior. No exterior fragments were matched with those of the interior but guided by the impressions

the wall of the structure is likely to have been at least 95mm thick. One of two fragments in the contrastingly

hard, fine, fabric, F-mfS1, had broad, probably near vertical linear impressions on the inner side which would

have been in contact with much narrower, very deeply set probably horizontal linear impressions on the outer

side (Fig. 26: 2). Their outer surfaces were distinguished from fragments in other fabrics by deep thumb-

moulding impressions (Pls 13 and 14). It seems likely that these fragments formed a specialized part of the oven

structure designed to withstand greater temperatures.

Middle Iron Age pit 1820 and Late Iron Age pit 711 included respectively five and one fragments with

linear impressions, all in a fabric with coarse flint inclusions, F-fefmqG1.

41 Fragments of a probable hearth base in fabric F-fG1from Iron Age pit 1417 had one smooth surface and surviving depths of up to 82mm. Two joining fragments of a superstructure formed an incurved open ring with an internal diameter of at least 160mm with a 58mm wide flat top. The wall thickness expanded to 65mm at

75mm below the rim. The outer surface appeared to have been smoothed and pressed with a broad spatula (Fig.

26: 3).

Fragments of very distinctive brown fabric F-fQ2, mainly from the lower fills of late Roman corn-dryer 1, typically had a single roughly smoothed surface and lacked any impressions and probably formed a floor. Some fragments survived to a thickness of 48mm. In general the thickest fragments lacking impressions may be from hearth or oven bases or possibly, as in the case of the possibly Roman oven in cut 1117, from small super structures for which no framework was required.

Bricks Several objects are brick-like. A block from MIA pit 229 has traces of at least four faces and although the remaining two sides are rough they may also be faces, giving a near-cube with dimensions of 60mm by 50mm by 48mm. Other blocks had no more than two surviving faces and were not measured. As noted above, most if not all of the blocks were probably used as oven furniture and as a group there is more evidence for care in the selection of the fabrics than for any other class, possibly indicating that they needed to withstand especially high temperatures.

Loomweights or oven furniture

There has been speculation concerning the function of perforated objects conventionally described as loomweights (Poole 1995). Those described below include examples which are broadly trapezoidal with single perforations and three which are, or probably are, triangular and have three perforations. Only the trapezoidal object from 245 was not associated with large amounts of hearth or oven fabric and it should be noted also that no spindlewhorls were recovered from the site. This is consistent with Poole’s findings at Danebury where, on balance, she thought it more probable that the triangular form was oven furniture.

(135) (199) Loomweight or oven furniture, F-mG1 (Fig. 26: 4). 94mm long, 0.47mm tapering to 37mm thick, block with one rounded and one flattened end. 10mm diameter pre-firing perforation set 24mm from rounded end and slanting towards it. Although only about 35% of the object is present it is likely that it had only one perforation. The flattened end is probably a base and the perforation was parallel to it. (245) (393) Loomweight or oven furniture, F-fefmqG1 (Fig. 26: 5). 0.76m high trapezoid, 70mm by 70mm flat base, 0.48 by 0.42 flattened top with slightly concave sides. An oval perforation parallel to the base was formed prior to firing by waggling a long object causing slight furrowing within the perforation and a raised rim of clay at one end formed on withdrawal. The 43mm long perforation was executed 15mm below the top with opening dimensions of 19mm and 14mm. The weight is largely complete. (711) (1074) Loomweight or oven furniture, F-fefmqG1 (Fig. 26: 6). Roughly equilateral, rounded corner triangular-shaped. With 170mm long, 94mm deep, sides block with one rounded and one flattened end. 10mm diameter pre-firing perforation set 24mm from rounded end and slanting towards it. There are three pre-firing perforations. Two diverge outwardly from each other from roughly the centre of one side which has been treated as the base for measuring angles

42 of roughly 60o and 50 o from horizontal. The two perforations exit the adjoining sides, on the one surviving side at 60mm above the base. A third perforation starting at 87mm above the base on the surviving surface slants at an angle of 55o from horizontal and would have exited the other surface. All the perforations have diameters of 17mm. (1331) (1857) Loomweight or oven furniture, F-fefmqG1. Broadly similar triangular form to the example from 711 but insufficiently complete to determine its orientation and the orientation of the three pre-firing slanting perforations, two of which were circular in section with diameters of 10mm-12mm and one of which was oval with 11mm by 8mm. The weight had a distinctively rough finish on one side (Fig. 26: 7).

Part from one side of a second loomweight of probably triangular form in pit 1331 in fabric F-mSi1 also had two slanting perforations of 10mm and 12mm. The perforation diameters contrast with one of 22mm penetrating a fragment from 1409.

The triple perforated triangular form occurs widely on Middle and Late Iron Age sites in Britain and

Europe and are of the Danebury clay weight type 1 (Poole 1995; Poole 1984, 401-3; Mepham 1989, 68, fig. 14;

Fasham et al. 1989, 112, fig. 102; Poole 1991, 372-80). The trapezoidal form is rarer. At Danebury Poole’s type

4 cylindrical form had single horizontal perforations close to upper end (Poole 1991, 379-80; fig. 7.49). Two triangular weights from Danebury had single perforations in one corner but a fragment from a third appears to have a more pyramidal shape. All were classified as type 2 (Poole 1984, 403; fig. 7.48). The single fragment of

‘loomweight’ from a Roman context was in a ditch and is most probably a residual Iron Age artefact.

Summary The group of ‘weights’ and their association with oven fabrics lends further support to the view that they were used in firing or cooking rather than in textile production. The other material is likely to derive from hearth, corn dryer or oven bases, or from the superstructures of the latter. Whilst it is tempting to suggest that the larger structures implied by the frameworks may have had an industrial rather than a domestic purpose there is no corroborative evidence such as crucibles or vitrified fragments.

Struck Flint by Steve Ford

A collection comprising 79 struck flints was recovered during the course of this phase of overburden removal and excavation. The struck flints were recovered from both topsoil contexts, the stripped surface and some cut features. Spatial control of where the material was recovered from was achieved. Topsoil finds and finds from the stripped surfaces were located individually. The typology of the collection is summarized in Table 29.

Most of the flintwork was recovered as residual finds in Roman features but some comments are in order about the general composition with regards to chronology. Some 23% of the flake component of the collection is

‘narrow flake’ (that is flakes exceeding a length:breadth ratio of 2:1). The categories were assigned by eye and both broken and intact flakes were used to prepare the statistics (Ford 1987). This proportion well exceeds the figures representing fortuitous production of narrow flakes in later Neolithic or Bronze Age assemblages clearly

43 indicating an early, that is Mesolithic or earlier Neolithic component is present in the collection. The collection includes a number of unambiguous examples of well made ‘blades’ and one blade core. Whilst many individual pieces of the flintwork are not closely datable, the collection almost certainly includes material of later Neolithic or Bronze Age date. One of the retouched pieces, a fabricator, is typical of (but not exclusive to) the Mesolithic.

It is considered that the flintwork here reflects areas of activity, possibly reflecting small scale occupation sites in both the Mesolithic and Neolithic/Bronze Age periods. This falls into the broader pattern of earlier prehistoric landuse on the Somerley plateau.

In addition (and excluded from the above statistics), two large patinated blades, both broken, were 65mm and 86mm long and are possibly long blades of Upper Palaeolithic date and related to the excavated occupation site found 500m to the west (Barton et al. 2009). These came from undated tree bole 610 and Iron Age pit 624, located just 2.5m apart in the north-east corner of the site (Fig. 4).

44 Table 29. Catalogue of struck flint

Possible Broken Cut Fill Group Type Area Intact Flake Intact Blade Broken flake Broken Blade Blade Spall Core Blade core Other 50 Topsoil strip24 1 strip 5 1 2 Fabricator strip 3 1(util) 2(1 util 4(1 burnt) 1 1 1 strip 8 2 1 strip 4 1 1 1 1 Scraper; Retouched flake strip 1 3 2 7 3 6 1 Scraper strip 2 2 3 9 61 2151 Gully Terminus 1 Tested nodule 32 84 2155 Ditch 1 42 94 2157 Ditch 2 127 179 Pit Scraper 135 199 Pit 1 216 292 Pit 1 435 671 2177 Gully Terminus 1 438 674 2177 Gully 1 444 680 2188 Ditch 1 514 751 Pit 1 1 522 760 Ditch 1 540 786 2187 Ditch Terminus 1 543 792 Pit 1 548 851 Pit 1 610 875 Treebole 1 palaeolithic (pat) 620 951 Pit 1 624 965 Pit 1 palaeolithic (pat) 1165 Corn Dryer 1 2 2 1331 1859 Pit 1 1323 1873 Pit 1 1347 1879 Pit 1 Pat (patinated)

1

Stone by David Williams

The range of sources for the querns, and possible querns, reaching the site is quite varied. It is possible that some of the fragments of the local “heathstone” listed below are broken pieces of quern, though none of them actually display any evidence for this. However, there is plenty of evidence that the site attracted quernstones from much further away, no doubt due to the limited choice of available stone in the region that was suitable for grinding purposes. This includes querns from the north of the site, the Chalk Downs, including a saddle quern, and as far away as the Pennine region, and then from the east in West Sussex. There is also a group of greensand querns whose composition and texture is unlike the greensand quarried from the Hythe Beds of West Sussex utilized by the Lodsworth quarry (Peacock, 1987). It is difficult to tie down the source, or sources, without known comparanda, but as Lodsworth may have monopolized the greensand deposits to the east of the site, a source amongst the Upper Greensand deposits of the Somerset/Wiltshire border to the west might be worth exploring.

1). (120) (1086) Large section of the lower stone of a rotary quern, with worn grinding surface and showing part of the spindle hole (D: c. 35cm; Th: 7.5cm at centre). The stone is a hard, compact, greenish-grey greensand with characteristic cherty swirls – most probably from the Lodsworth quarry in West Sussex (Peacock, 1987, fig. 13, no. 4; Shaffrey and Roe, 2011). 2). (127) (179) (A) Roughly rectangular slab of a hard, compact, greenish-grey greensand with characteristic cherty swirls – most probably from the Lodsworth quarry in West Sussex (Peacock, 1987; Shaffrey and Roe, 2011). Both surfaces have been worn smooth all over, while the two long edges display the characteristic wear pattern of a sharpening stone (L: 11.5cm; W; 7cm; Th: 2cm). The Lodsworth Quarry produced various forms of quernstones over a long period of time but there is no evidence of honestones being deliberately fashioned there (ibid.). Instead, it is more likely that this piece was originally part of a quern that broke and that it had a secondary use as a honestone (B) Part of a smooth, well-rounded, pebble of quartzite. 3). (643) (1052) Small fragment of quernstone with a worn grinding surface. The stone is a fine-grained, compact, greensand, though on this occasion not from the Lodsworth Quarry. 4). (941) (1376) (A) Two small separate segments from the edge of a disc-shaped rotary quern, with worn grinding surfaces. Both pieces are possibly from the same quern but it is difficult to be sure. The stone is a fine-grained, compact, greensand, though not from the Lodsworth Quarry. (B) There is also a small piece of dark reddish-grey ferruginous sandstone, with no evidence of working, most likely the local iron-rich heathstone”. 5). (1165) Two large segments from the edge of an ?upper stone of a disc-shaped quern, both showing a well-worn grinding surface (D: c. 50cm; Th: up to 4cm). The stone is a fine-grained, compact greensand, though not from the Lodsworth Quarry. The larger piece has part of a bivalve mollusc shell embedded in the fabric. 6). (1117) (1587) Three small fragments of quernstone, all with worn grinding surfaces, one of which displays evidence of a food-shoot, indicating an upper stone. The stone is a fine-grained, compact greensand, though not from the Lodsworth Quarry. 7). (1126) (1587) Small segment from the edge of a rotary quern, with a worn grinding surface. The stone is a fine-grained, compact greensand, though not from the Lodsworth Quarry. 8). (1128) (1584) A large part of a broken saddle quern, with characteristic worn dipped grinding surface (grinding surface: 12.5 x 11cm; Th: 6.5cm). The stone is a quartz sandstone with tightly-packed grains. Quite likely a sarsen from the Chalk Downs to the north of the site. The unbroken outer surfaces are all quite smooth, suggesting that this may originally have been a large water-worn “pebble/boulder”. 9). (711) (1072) Small segment from the upper stone of a rotary quern, with a well-worn grinding surface and showing part of a ?food-shoot hole (D: c. 24cm). The stone is a sandstone with tightly-packed quartz grains. Quite likely a sarcen from the Chalk Downs to the north of the site. 10). (1446) (3054) Small fragment of a rotary quernstone with a worn grinding surface. The stone is Millstone Grit, with a probable origin in the Pennine region of northern England. 11). (817) (1190) Five small fragments of quernstone, most of which display a worn grinding surface. The stone is Millstone Grit, with a probable origin in the Pennine region of northern England. 12). (708) (1056) Roughly oblong-shaped, worked fragment of a dark reddish-grey ferruginous sandstone, most likely the local iron-rich “heathstone” (L: 17cm; W: 15cm; Th: 2.5cm). The side of one of the top surfaces is broken but shows part of a “tongue” cut into the stone which, when complete, might have provided one half of a ? mortise and tenon joint. It is difficult to suggest what this stone was used for; certainly it is difficult to think of anything to do with querns. 13). (772) Small, roughly oblong-shaped, fragment of a dark reddish-grey ferruginous sandstone, most likely the local iron-rich “heathstone”. No evidence of working at all.

14). (916) (1297) Small irregular-shaped fragment of a dark reddish-grey ferruginous sandstone, most likely the local iron-rich “heathstone”. No evidence of working at all. 15). (1306) (1780) Two small irregular-shaped fragments of a dark reddish-grey ferruginous sandstone, most likely the local iron-rich “heathstone”. No evidence of working at all. 16). 817 The majority of a large disc-style upper stone (diameter 40 cm) with central food-shoot or spindle hole. From a rotary quern or, perhaps more likely given the size and the fact that the edge is broken and so the diameter could be wider, a millstone. Made from a hard, compact, greensand – possibly from the Lodsworth quarry in west Sussex (Peacock, 1987; Shaffrey and Roe, 2011).

Ceramic Building Materials by Danielle Milbank

A total of 2.874kg of ceramic building material (53 fragments) was recovered during the excavation. Of this, the majority of identifiable fragments were tile, with a moderate quantity of small fragments which could not be identified. The ceramic building material was examined under x10 magnifications and is summarized in Table 30.

Ceramic building material was recorded most frequently in small quantities, typically 50 to 200g, in a total of 18 contexts. The largest quantities were encountered in 2040 and 2402 (c. 400g).

The tile fabric ranged from medium (slightly weak or friable) to very hard and well-fired, and a fairly narrow range of fabric types and forms were present. The majority of the hard, evenly-fired types have frequent small well-sorted quartz sand inclusions. Occasional examples contained small and medium-sized flint inclusions. The colour was typically a pale red, with several examples of a grey (reduced) core.

A piece of tile was recovered from context 847. This is 15mm thick and comprises a hard, evenly-fired slightly laminated clay fabric with sparse groggy and manganese. The form is fairly even, the colour a mid red, and it is of late medieval or post medieval date.

Context 1730 contained two fragments of a slightly soft, evenly-fired fine clay fabric with sparse, small groggy inclusions. The colour is a pale orange red and the form is uneven and c. 11mm thick, with a striated upper surface and one peg hole present, along with a small amount of splashed glaze. The thinness, form and fabric suggests the pieces are of broadly medieval date, possibly 13th or 14th century.

Two further fragments of the same fabric and a very uneven form, with a small amount of pale green glaze, were also recovered from context 2040. These are 11mm thick, with edge-thickening and are of the same medieval

(13th or 14th century?) date.

Fragments from 2131 are of similar fabric, form and likely date, with a piece from 1908 with similar form and small flecks of a colourless glaze probably also dating to the medieval period.

Fragments from deposit 3254 are of a hard, evenly-fired sandy fabric with moderate small and medium groggy inclusions. The form is fairly neat but very thin (11mm) and the fragments are a pale buff colour, and are of likely medieval date.

Fragments recovered from 1742 are of a hard but slightly friable fabric, with frequent small and medium darker red groggy inclusions, sparse small manganese, larger pale groggy inclusions, and a pale orange red body.

The form is uneven, with one peg hole present and some edge-thickening, striations on the upper surfaces and strawmarks on the undersides. This indicates a medieval date for the fragments, possibly 13th or 14th century.

Four further fragments of this fabric type and form were recovered from deposit 3477 (feature 1947) which have a rough sandy underside and deep strawmarks on the upper surface. The pegholes are edge-thickened and the pieces are of likely medieval date, possibly 13th or 14th century.

Three fragments were recovered from deposit 3467 which are of a slightly coarse sandy fabric with a light red colour and reduced core. Two peg holes are present and the form and fabric suggest a broadly medieval date.

Context 2041 contained 6 fragments of a hard, evenly-fired sandy fabric with moderate small and medium groggy inclusions. The form is fairly neat but very thin (11mm) with some edge-thickening and the fragments are a pale red buff colour, and are of likely medieval date.

Context 2440 contained several pieces of the sandy, pale red/buff fabric, with a single example of a laminated fine red clay fabric with sparse groggy inclusions and some pale buff laminating.

Summary The majority are likely to represent pieces of plain peg tile, and two fragments with a peg hole were recovered.

Typically, they are 10 to 15mm thick and could not be closely dated. Tiles of this type were produced from the

12th century onwards, however they were generally limited to high-status buildings before becoming widespread by the 15th century.

Table30. Catalogue of brick and tile

Group Cut Deposit Type No Wt (g) COMMENT 847 1278 Pit 1 24 1016 1463 Pit 2 139 2259 1308 1789 Ditch Terminus 5 10 1730 3194 Pit 3 183 1736 3254 Pit 2 179 2286 1742 3261 Ditch 3 160 peg hole 2284 1818 3287 Gully 2 22 2289 1908 3371 Ditch 1 53 2296 1943 3467 Gully 3 105 peg hole 2351 1947 3477 Ditch 4 329 peg hole 2293 2040 3568 Gully Terminus 3 402 glazed 2293 2041 3569 Gully Terminus 6 342 2355 2108 3596 Beamslot 4 90 2131 3669 Pit 2 104 2358 2402 3991 Ditch Slot 5 410 2430 4171 Posthole 2 67 2440 4174 Posthole 4 240 2440 4175 Posthole 1 15

Burnt Bone by Ceri Falys

Burnt bone was recovered from in and around urn 39 (deposits 95 and 96, respectively), and pit 711 (deposit

1075). The bone from urn 39 was whole-earth recovered in a series of three 0.02m spits, while the bone from 1075 was hand collected on site. The urned remains were human, and those from pit 1075 were animal (Table 30).

Urned deposits 95 and 96 All bone was subjected to osteological analysis following the procedures suggested by Brickley and McKinley

(2004) and Buikstra and Ubelaker (1994). During the post-excavation processing, the samples from the urn excavated from context 39 were floated and wet-sieved to a 1mm mesh size. The burnt bone and other associated residues were separated for further analysis. The bone is uniformly buff in colour, which indicates an efficient cremation process (i.e. the skeleton was subjected to adequate time, temperature and oxygen supply for the organic components of the bone to be oxidized).

The preservation of the bone was generally poor, with an exceptionally high degree of fragmentation present.

The bone from each spit was sorted using a sieve stack of 10mm, 5mm, and 2mm mesh sizes, and weighed. A total of 14g of buff coloured bone was present for analysis from urn fills 95 and 96. The relative weights from each of the sieves for the urned deposit are as follows: 10mm = 6g, 5mm = 1g, 2mm = 6g. All bone from the soil surrounding the urn (i.e. deposit 96) is of a 2mm mesh size, and weighs just 1g. The high degree of fragmentation present is illustrated by comparing the weight of bone in each mesh size category with a maximum fragment size of

41.6mm (Table 31).

The identified skeletal fragments are: several pieces of very thin cranial vault, a developing proximal epiphysis (“head”) of a humerus or femur, a midshaft portion of a radius, and a distal manal phalanx. A lack of element duplication suggests the presence of only one individual. The overall size and stage of development of the identified elements indicate the individual was an infant (<1 year old) at the time of death. It is not possible to determine the sex of non-adult skeletal remains, and no pathological alterations or non-metric traits are observed.

Pit 711 A total of 8 pieces of burnt animal bone were recovered from pit 711 (deposit 1075), uniformly white and, although several of the pieces were small in size, they demonstrated generally good surface preservation. The largest element is a long bone shaft of a medium sized animal, likely a pig. The epiphyses were unfused and absent, indicating the animal was immature at the time of death. No further information could be retrieved.

Table 31. Inventory of burnt bone. Cut Deposit Colour Max Frag Size (mm) Tot wt (g) Age Sex Comments 39 95 buff 41.6 x 25.3 13 Infant n/a 3 spits, human 39 96 buff 14.4 x 6.9 1 I I not identifiable 711 1075 white 121.1 x 23.2 26 immature n/a ?pig Key: I = indeterminate

Animal Bone by Danielle Milbank

The bone recovered comprises a small quantity of highly fragmented bone collected from just 8 contexts and presumably represents a tiny sample of bone that has not survived due to ground conditions. The condition of the bone was poor, with high fragmentation and frequent surface erosion. Where pieces of sufficient size were present, only cattle or large mammal (horse, cattle) bones could be classified (Table 32).

Table 32. Catalogue of animal bone

Cut Deposit Group Type No Frags Wt (g) Comment 201 376 2155 Ditch 9 4 540 786 2187 Ditch Terminus 6 8 540 788 2187 Fill of Pot 787 3 12 540 856 Ditch Terminus 5 3 Cattle tooth 711 1075 Pit 8 24 1331 1859 Pit 4 2 Large mammal 1437 2085 Pit 10 12 Cattle tooth 1445 3050 Pit 4 52 Cattle tooth and large mammal

Glass by Danielle Milbank

A single piece of glass was recovered from pit 1016. The piece weighs 14g and is a pale blue colour, with frequent small bubbles, and is 6mm thick. It is typical of Roman glass but could not be more closely dated. The form suggests it represents part of a small flat-sided vessel.

Metalwork by Susan Porter

A total of 221 metal objects were recovered from the site with a combined weight of 1,126g. Of this assemblage there was single item of copper alloy, a single ferrous blade, and two lumps of iron slag, with the remainder of the assemblage comprising hobnails and nails of which a number may represent coffin nails.

Copper Alloy A single fragment of copper alloy (CAT 59) was recovered from posthole 732. The artefact comprises three narrow copper alloy wires twisted together, surviving for a length of 40mm. The length is slightly curved and possibly suggestive of an item of jewellery such as a bangle or necklace, although a more functional purpose should not be ruled out.

Iron Nails formed the greater part of this assemblage, but pit 214 yielded a small wedge-shaped blade with tang (CAT

40). It survived to a length of 40mm (including tang) and was 15mm wide at its widest part. The point is missing, however due to the tapering of the blade it cannot have been much larger than what survives perhaps another

10mm at most. The blade seems to be simply made using a folded piece of metal, there is no indication of the type

of handle into which the tang would have slotted. The blade itself is not diagnostic and cannot be closely dated but eh context is Roman.

Nails

Nails in both complete and fragmentary form were recovered from several features across the site, three large assemblages of nails were from pit 100 and graves 417 and 218 are discussed below. The remainder of the nails were recovered in fragmentary form from several features (Table 33).

Table 33: Catalogue of nails other than those described in text (all ferrous).

Deposit Type Cat No object no Wt (g) Comment 172 Pit 39 Nail 1 11.5 near complete 670 Ditch 58 Nail 1 2.5 base of stem fragment 3185 Ditch 61 Nail 1 10.5 43mm length, bulbous head 3185 Ditch 62 Nail 1 7 35mm stem fragment 3185 Ditch 63 Nail 1 6.5 38mm stem fragment 3254 Pit 64 Slag 1 50 fe slag 3577 Posthole 65 Nail 1 1 stem fragment 16mm 3879 Posthole 66 Nail 1 19.5 68mm length A total of 38 metal artefacts were recovered from pit 100, of which 17 were fragments of large nails, with squared heads and cross-sections. A single complete example, Cat. no. 7 was 65mm in length with a squared head

18mm wide. The remaining stem fragments measured between 20-36mm in length, where nail heads survived they were between 15 and 23mm in width. The remaining 21 items recovered from this pit were hobnails varying in size between 13-17mm in length with heads between 6-13mm wide, all heavily corroded. Cat. no. 38 appears to be complete with the curved base fixing still in place.

Grave 417 yielded a total of 16 thick set heavy duty nails weighing between 5.5 and 19 grams and it seems likely given their relatively uniform size, (16-22mm heads and incomplete lengths between 14 and 60mm), and location with a grave cut, that these were coffin nails.

Grave 218 produced 155 nails, 95 of which were hobnails and are likely all that remains of a pair of shoes buried with the individual. The remaining nails were recovered from three identified deposits within the grave cut and are likely to represent stray hobnails and probable coffin nails. Context 357 was assigned to the 95 hobnails thought to represent shoes (Pl. 8), and contained Cat. nos 67-161, with Cat. nos 67-93 comprising hobnails the corrosion of which has preserved elements of the leather shoe. For the most part the location within the sole cannot be confirmed however several fragments can be confirmed as edging pieces. No. 78 is the largest piece likely to be from the central part of the sole it was 45mm wide and 30mm long with the suggestion of two possibly three layers of leather between the top and bottom of the sole of the shoe. No. 93 was 65mm in length and formed an edge strip of the shoe. A further 68 loose hobnails were recovered from this context with heads on average c.10mm wide and varying in length between 5–20mm. The quantity of hobnails present indicate a heavily nailed sole, perhaps with

nails along all of the sole rather than arranged in a simple pattern. the leatherwork does not survive to any extent although several nails with attached corrosion give the suggestion of at least 3 layers of leather.

Three further deposits encountered within the grave cut (377, 380 and 381) also contained nails, which were more substantial than the hobnails and seem likely to represent coffin nails. Deposit 377 contained 3 nail stem fragments between 13–50mm in length. Deposit 380 contained a single stray hobnail, probably part of 357, and 13 larger nails, between 25–75mm in length with substantial square heads 12–25mm wide. Nails of similar size were recovered from deposit 381. These nails were noted as having been arranged in a line 1.1m from the eastern end of the coffin, and may the upper part of the coffin, having collapsed outwards as it decayed. Seven further loose hobnails were also recovered from this deposit.

Summary For the most part assemblages comprising solely of nails can tell us very little, however in this instance it is possible to suggest that two individuals (likely to be of Roman date) interred on this site were buried in wooden coffins, with the individual from grave (218) being buried wearing his/her hobnailed shoes (Pl. 8). The nails and hobnails recovered from pit 100 are likely to represent discarded materials, the 21 hobnails may either be discarded shoes or hobnails that were thrown away having been replaced in the shoe. The remaining metalwork recovered from the site is likely to have been deposited as discarded rubbish or perhaps in the case of the copper alloy bangle, broken and perhaps lost. No further work is anticipated on this assemblage.

Slag by Susan Porter

Two lumps of iron slag were recovered from pits 1040 (77.5g) and 1736 (50g). Neither piece was diagnostic of any particular process.

Clay Pipe by Danielle Milbank

A single clay pipe stem was recovered from gully 340 (498), which is plain and undecorated. The bore diameter is suggestive of a mid 17th to early 18th century date (though it must be noted that this dating can be only tentative when applied to a single example).

Charred Plant remains by Rosalind McKenna

A programme of soil sampling was implemented during the excavation, which included the collection of soil samples from 123 sealed contexts (Appendix 4) generally of 30–40 litres. The samples were wet sieved using a

0.25mm mesh and the flots examined under a low-power binocular microscope at magnifications between x12 and

x40. Taxonomy and nomenclature follow Stace (1997) for plant remains and Schweingruber (1978) and Hather

(2000) for charcoal. Details of methodology and identification guides used are in the archive.

Charred plant macrofossils were present in 50 samples. The preservation of the charred remains varied from sample to sample and even within samples: where abundant remains were present within a sample the preservation also varied from poor to very good.

Indeterminate cereal grains were recorded in 47 of the samples. Identified cereal grains were recovered in the form of wheat (Triticum sp.) – six samples, barley (Hordeum sp.) – five samples, and oat (Avena sp.) grains – three samples. Oat awn fragments were also present in three samples. Given that floret bases from oats were not present it is not possible to ascertain whether the remains were of the cultivated variety. As the samples produced only a few grains of this species, it is probable that the oats are of the wild variety and became incorporated into the assemblages as weeds of other crop species.

Wheat was the most abundant grain present within the samples. It is probable, based on the general size and remaining identifying morphological characteristics that the species of wheat utilised was that of a hulled wheat, such as emmer, einkorn or spelt. In hulled wheat, the chaff is fixed firmly to the grain and is therefore more difficult to remove. Although de-husking would have been a time-consuming activity in the past, wheat chaff does give the grains protection in the field and in storage, providing a useful barrier against water and insect damage.

The dominance of wheat is probably due to the likelihood of the glume wheat coming into contact with fire when it is parched to remove the grain from the glume.

Barley was also present, in smaller numbers in the samples. Barley was often grown as a dredge crop along with oats as a buffer against adverse weather; it was also mixed with oats to make coarse bread.).

Another, more indirect, indicator of cereals being used on site is the number of remains of arable weeds that were found in 47 of the samples. These weeds are generally only found in arable fields, and are doubtless incorporated into domestic occupation samples with crop remains. Along with grasses (POACEAE), remains of goosefoot/orache (Chenopodium/Atriplex), docks (Rumex), and cleavers (Galium aparine) also fall in this group.

All these species would almost certainly have been brought to the site together with harvested cereals.

Just two of the samples produced medium sized suites of remains, both in terms of quantity and diversity, one late Iron Age and one middle to later Iron Age. These medium sized suites of remains were dominated by indeterminate cereal grains, with smaller amounts of identifiable cereal gains, weed seeds and some chaff fragments.

Only one of the samples produced a large suite of remains in terms of quantity and diversity. Sample 18 came from pit 147 (269) dated to the late Iron Age. Over 1900 indeterminate cereal grains were recorded. Of the

identifiable remains, 830 wheat grains, 254 oat grains and 122 barley grains were present, together with a suite of weed seeds such as grasses, goosefoot/orache, knotgrass, dock and the cabbage family which are species typically associated with cultivation. Other evidence for cereal grains and processing was present in the form of numerous different types of chaff. Glume bases and spikelet forks were recorded alongside culm nodes and other chaff fragments.

If cereal processing were occurring at the site, it would be expected that some remains (most probably in high numbers) of cereal chaff would be found. There was chaff present in fifteen of the samples, but only in small amounts in comparison to the number of grains recorded.

The results show a consistent pattern for deposition of crop processing debris. Two explanations for this pattern in the data are possible:

1. Semi clean grain (i.e. grain which was largely separated from its surrounding cereal chaff, and has been fine sieved but still includes larger sized weeds from the crops) was parched first before removing any remaining crop contaminants. These deposits could represent that proportion of parched crops which was charred during parching and then discarded. 2. Both crop processing waste (chaff and weed contaminants) which was burned as fuel, and charred cereal grain, possible resulting from parching before milling, were discarded into the same features, resulting in the mixtures of cereal grain, chaff and weeds of crop encountered.

Charcoal fragments were present in nearly all of the samples but preservation was generally poor. Identifiable remains were present mostly in small numbers in 100 of the samples. The total range of taxa comprises oak

(Quercus), willow/poplar (Salix/Populus), and hazel (Corylus avellana) with oak overwhelmingly predominant in all phases, with willow/poplar being dominant in two of samples, and hazel being dominant in three.

Summary The remains of plant macrofossils recovered from the samples showed the utilization of wheat and barley as well as indeterminate cereal grains, and chaff fragments, together with a range of weed seeds typically associated with cultivation. The fact that the samples have produced broadly similar results suggests that these secondary deposits do not result from deposition of debris from accidental charring events, but instead represent a consistent pattern of charring cereal grain, chaff and crop weeds over the period of occupation. The one exception to this may be sample

18, which could represent a single charring event.

Previous work carried out at the site (McKenna 2009) also produced very small numbers of indeterminate cereal grains and chaff, along with wheat and weeds associated with cultivation. Other sites in Hampshire of roughly similar dates, such as Nursling, Southampton (Adam et al. 1997), have also produced suites of remains where emmer/spelt wheat is present alongside smaller amounts of barley and oats, and weed seeds and chaff fragments.

Radiocarbon dating

One sample was submitted to the Chrono radiocarbon dating laboratory at the Queen’s University of Belfast. The result was calibrated using Calib rev 7.0 with data from INTCAL 13 (Reimer et al. 2013) and are detailed in Table

34. The result is quoted at 2-sigma (95.4% probability).

Table 34. Radiocarbon date

Lab ID Feature Material Radiocarbon Age (BP) Calibrated Area under curve date BC at 2-sigma UBA22372 Pot from pit 43 (97) Pottery residue 2135 + 31 351-299 16.3% 277-223 0.7% 210-54 83.0%

Conclusion

In interpreting the significance of the results of the archaeological work it is necessary to take into account the very particular physical setting which has influenced the resources available to the local population during the periods of occupation. The low plateau itself is covered with poor soil which does not easily retain moisture and hence would be prone to parching if exposed or under only thin vegetation during the summer months. Much of the surrounding area is wooded, recent managed conifers now often replacing mixed forest. The soils on the scarp to the east are similar and those west of the River Avon in the valley bottom fluctuate between having high ground water and, conversely, a tendency to drought which would limit crop yields. This might imply that in areas of the plateau free of woodland, grazing of livestock would have been more productive than the growing of crops but that cereals may have been available from east of the river where the soils were more suitable (CSAI 2017). Where woodland persisted it would have been a resource in its own right for fuel and construction material, as well as offering a landscape suitable for pannage. At times the river itself would have allowed the forming of local and more wide- ranging trading networks. In the 18th century it was possible for vessels to enter it at Christchurch and to continue to within two miles of Salisbury. In 1685 two vessels carrying 25 tons were unloaded at Crane Bridge in the city

(Defoe 1927, 188; Morley Hewitt 1965, 42). It is unfortunate that for all periods on the site the survival rate of bone was extremely poor and that in general crop and associated plant remains were distributed only thinly. The character of the material recovered from both Iron Age and Roman contexts would allow some processing of cereals to have taken place before they were brought to site.

The earliest known activity on the Somerley plateau is represented by an important Upper Palaeolithic flint knapping site, broadly contemporary with a much larger site at Hengistbury Head (Barton et al. 2009). Flint scatters of varying density from the Mesolithic or Early Neolithic periods have been found nearby and radiocarbon dating placed a pit less than 500m south of the present site in the latter period (Ford 2002, 5, fig. 3; Taylor 2008,

11). No features of these periods have been identified during the present phase of archaeological work, but two

possibly Palaeolithic flint blades were recovered from two (later) pits close to one another, a coincidence which possibly suggests the disturbance (in prehistoric times) of another in situ Palaeolithic site.

In this phase of work, a Beaker cremation pit was the earliest feature discovered. During investigations of areas to the south a complete Beaker with an arrow head was found in what was first identified as a treethrow and sherds from another Beaker were recovered from a pit (Raymond 2009, fig. 9; Raymond 2008). At least two other pits from the present phase of archaeological work are likely to be of Bronze Age date, and struck flints were collected (again, from later features) from a relatively narrow band of the north part of the site, more or less corresponding with the later trackway ditches. A long standing route across the plateau leading to the Avon valley may either have come into existence because of the presence of Beaker burials or have influenced their siting assuming that there had been more in the area, but there is no direct evidence for its date of inception. The track may have been linked to a route crossing the river, possibly a predecessor of Ellingham Drove which is bridged

900m east-south-east of the site. Two Middle Bronze Age pits have been found north of the drove on the east side of the river and Late Bronze Age and Early Iron Age pottery suggests that there may have been continuity of settlement there (Butterworth 1995).

Iron Age

Much of the Iron Age occupation evidence had been erased prior to the archaeological work but enough remains to demonstrate that the community grew and re-fashioned the landscape until late in the period. The west to east route across the plateau to the north of the site probably determined the particular location of earlier Middle Iron Age settlement but the division of land was modified on the east side to take into account the line of the scarp. There is no evidence that the area of habitation was fully enclosed during the period, although arcs surviving from ring gullies suggest that some structures stood close to ditches. The limited number of post holes and lack of perceptible patterning of them suggests that stake-supported wattle and daub was the principle mode of construction although the impacts of the dearth of dating evidence and of truncation should not be underestimated given that the ring gullies themselves are far from complete. It is likely that clusters of pits formed in association with particular structures early in the occupation, although this cannot be demonstrated as dating evidence is only quite broad within the period. There is stronger evidence for clustering in the later phase. Most of the datable pits show no cogent signs of deliberately structured deposition although rich deposits of ceramic material, derived from exposure to high temperatures, imply industrial as well as domestic activity on site. It may be significant that one of only two pieces of slag from the site was from an Iron Age pit (1040), but the pit is dated only by three sherds of pottery, and a single small piece of slag could easily have arrived on site from elsewhere. During the later Iron Age

phase deliberate deposits of whole or near-complete vessels, some inserted into recuts of filled-in pits and ditches, are suggestive of the enactment of rites. Aside from these larger, deliberate, deposits the pottery is in general scattered thinly.

There is limited knowledge of Iron Age activity at the southern end of the Avon valley in Hampshire, with the notable exception of Hengistbury Head (Cunliffe 1987). There was an apparent hiatus of occupation during the period on the valley bottom east of the river at although nearby rectilinear enclosures identified by the NMP

1km south of Gorley are provisionally dated as Iron Age and/or Roman (Coles and Ford 2016, 79; Young 2008,

39, fig. 23). They intersect with arcing linears of a different, possibly earlier date, and are overlooked by Gorley

Hill and were the first of the period to have been discovered on this part of the floor of the Avon valley

(Light et al. 1994, fig. 22). On the other hand, at least one phase of multi-period enclosure systems is related directly to a slightly sinuous long linear feature at Breamore, on the west side of the river, 10km north of the site

(Young 2008, 38-9, fig. 24). It was suggested that nearby linear features defining a possible broad south to north drove-like feature on the valley floor might be of Bronze Age date based on an orthogonal relationship with other linear features and their association with ring features of a scale commensurate either with small barrow or roundhouse ring ditches (Young 2008, 36-7, fig. 20).

Fieldwalking within a 5km by 12km area extending northwards along the Avon Valley from 2km north-east of the site recovered pottery which was taken to indicate possible settlement in several loci. The study area took in three previously known discrete hilltop enclosures set on promontories at the head of the scarp to the east of the river, each within a potential territory divided naturally from the others by east to west flowing tributaries. No equivalent enclosures are recorded on the west side of this section of the valley but most of the small amount of pottery recovered by the survey judged to be from the earlier part of the period was from that side. This in part reflects the greater number of modern fields surveyed there, although the absence of Iron Age activity from east of the river at Ibsley is consistent with a preference for the west side of the valley (Light et al. 1994, 75-6, fig. 22;

Coles and Ford 2016). Fieldwalking is poorly suited to the identification of later Bronze Age and Iron Age activity areas due to the lack of durable artefacts associated with those periods and, more generally, the role of artefact movement, as well as destruction, in the ploughsoil should be taken into account (Tabor 2008, 31-3). In contrast, detailed analysis of the air photographic evidence further north from the area between the rivers Avon and Test and the subsequent polyfocal research of the Danebury Environs Programme has revealed enclosed and unenclosed settlement associated with extensive field systems (Palmer 1984; Cunliffe 2000). In addition, excavations along the route of the M3 have made important contributions to understanding of settlement development in central

Hampshire during the period (Fasham 1985; Fasham et al. 1989).

The influence of Bronze Age linear ditches over the subsequent development of landscapes is well attested in

Wessex (ie. Combe Down and Dunch Hill, Bradley et al. 1994, figs. 10 and 15). The extensive data collected in the

Danebury environs project shows far more examples of field systems with differing orientations which pre- or post-date the larger ditch but it is suggested that most the large linear boundaries of the Late Bronze Age survived

‘into, and possibly throughout the Iron Age’ (Palmer 1984; Cunliffe 2000, 155-60). At Gussage Hill and

Thickthorn Down on Cranborne Chase the orientation of an Iron Age field system was determined by a Neolithic monument, the Dorset Cursus (Bowen 1979; Barrett et al. 1991, 232-6). In all these examples the Iron Age field systems can be considered as much in their relationships to a range of discrete enclosures, ranging from to

‘banjos’, as they can to linear ditches. However, these areas of west central Hampshire and east Dorset are set on chalk upland of sharply different character to the gravel terrace on which the Somerley site is set and to the Avon

Valley which it overlooks.

Whilst isolated pits may well be situated at a distance from settlement, large groups are only likely to occur close to one. Many of the pits in the northern part of the site for which there was no dating evidence are likely to be contemporary with the ring gullies which in turn are datable only by their morphology. Their overall pattern suggests parallels with the early Middle Iron Age development of an unenclosed settlement alongside a long linear ditch at Easton Lane, Winchester which extended across the infilled Bronze Age to Early Iron Age Winnall Down enclosure 100m to the south-east (Fasham et al. 1989, 58-68; Fasham 1985, 18).

The chalk natural at the Winchester sites is well suited to the identification of small features. It was found that the ring gullies were often penetrated by concentric rows of stakeholes and in several instances they surrounded circles of post settings. The ring gullies were interpreted as settings for wattle and daub walls covered with roofs supported by internal posts (Fasham 1985, 18-22). The enclosed Early Iron Age phase at Winnall Down included similarly-built but fewer round houses and several small rectangular structures of a form usually considered to be for above-ground storage, possibly of cereals (Fasham 1985, 13). The group of similarly-sized small rectangular structures dominating the central area of the enclosure at Gussage All Saints were surrounded by pits and scoops into the natural which were interpreted as working hollows (Wainwright 1979, 16-20, fig. 16). At both of these sites there were great reductions in the numbers of rectangular structures during the Middle Iron Age although they differ in other respects with a single structure within a recut enclosure at Gussage contrasting with multiple unenclosed roundhouses at Winnall Down (Wainwright 1979, 21, fig. 17; Fasham 1985, fig. 15). By analogy, the apparent dearth of rectangular structures at Somerley supports the placing of the initial Iron Age occupation in the middle of rather than early in the period. In general the reduction in the number of small rectangular structures on a site and an increase in the number of pits are thought to reflect a change in the prevalent

method of storage, particularly of cereals, from above to below ground. It remains likely that storage would have been the primary use for many of the Iron Age pits at Somerley. It has been suggested that the shapes and sizes of pits may relate to their period of use and at Gussage pits which narrowed towards their apertures and increased in size formed a higher proportion in each successive phase (Jefferies 1979, 15). At Somerley there was a higher proportion of smaller pits which are earlier but three of the largest pits are also deemed earlier. There is a lack of the classic ‘beehive’ or ‘bell’ form pits, both of which have narrow apertures considered to favour storage but this may be due to truncation given the shallowness of many of the Somerley pits relative to those from other Wessex region sites. Nonetheless, the open cone forms seem particularly unsuited to that purpose.

As the means for storage of a staple food pits have been used to estimate population capacity at sites.

Jefferies suggested that one pit might supply three to six people for a year with their needs for cereal and notes

Bersu’s suggestion that a pit might be re-used over a period of ten years if kept free of infestation (Jefferies 1979,

15). Sharples has noted that hillforts appeared to have had a storage capacity greatly exceeding the needs of their own populations. Offering a view beyond the individual site, he has proposed that smaller settlements were obliged to contribute labour for the building and maintaining of the locally dominant hillfort, and that the labour force was fed with resources levied from the smaller settlements and stored within the hillfort. The smaller settlements would then have been locked into a cycle of dependency on the dominant settlement (Sharples 1991, 260). Even without full knowledge of the extent and density of settlement and storage capacity on the Somerley plateau the number of pits with the potential to be used for storage suggests that the local population was not in a dependent relationship of that kind. On the other hand, if the lack of direct evidence for crop production is a true indicator that the community produced few, if any, cereal crops of its own, there would have been a need for regular exchange with other communities. One Middle to Late and two Late Iron Age pits stand out for the amounts of cereal grain in them, wheat in particular (McKenna above, and Appendix 4). Each of the pits included crop waste which allows the possibility of occasional episodes of fuller on-site processing. Querns for grinding cereals were all from distant sources. Those in sarcen, presumed erratics from the chalklands to the north, hint at southward traffic along the river, but a Pennine millstone grit quern from an Iron Age context is a surprising discovery on so modest a site, possibly indicative of secondary exchange.

This would imply that the Somerley community had commodities suitable for exchange such as livestock on or off the hoof, or manufactured goods. In the absence of demonstrable stock enclosures during the earlier Middle

Iron Age phase it may be suggested that animals may have been grazed extensively. The only diagnostic bone fragments are from cattle or large mammals but the assemblage is far too small to draw any conclusions about a location for which sheep, pigs and ponies may have been equally well adapted. There is evidence of enclosure

which would have suited both livestock-holding and arable use in the later Iron Age. The substantial remains of oven furniture and superstructure testify to possible manufacturing but a single piece of slag is not sufficient to argue that metalworking was part of an economy which included the trading of metalwork. However, the site’s location may have afforded control or influence over routes of exchange which included the river.

Roman

The pottery suggests that the Iron Age occupation continued into the early 1st century AD. Exclusively Iron Age sherds in the lower fills of pits including later Roman sherds in the upper fills imply that there was a phase of no more than low intensity land use (if any) between the two periods, should it be accepted that most of the Roman pottery is late and allowing for the impact of manuring for the distribution of the few demonstrably earlier sherds.

The surface of pit fills had subsided then stabilized prior to renewed intensive activity around two centuries later when their residual hollows trapped newly introduced material. However, manuring implies cultivation or at least the improvement of grazing and there is little to suggest that the land was divided into suitable plots. Such continuity as there may have been from the Iron Age is represented by a few lengths of ditch which are of broadly similar orientation to the old plateau trackway. On the other hand, a new later Roman droveway differed in orientation and was established slightly to the south of the old track. There is no evidence that it continued as far as the scarp; indeed at least one ditch and a corn-dryer may have been barriers to a projected course. The implication may be that this section of the River Avon had ceased to be a means for transport during the earlier Roman period, depriving the settlement of its economic base. This contrasts with the interpretation of finds from fieldwalking on the Avon valley floor only a few kilometres further north where there was little disruption to the perceived continuity of settlement from the Late Iron Age to Roman periods (Light at al. 78). Certainly there was renewed activity east of the modern crossing at Ellingham in the 3rd century where two ditches appeared to have formed part of an enclosure (Butterworth et al. 1995, 74–5).

A recent summary of knowledge concerning the division of the Roman landscape concluded that whilst much is known about enclosure on the chalklands in the region, the off-chalk areas remain poorly understood, and it is possible that with poor soils much of the Somerley plateau was left to forest regeneration (Fulford 2014, 166).

Wood may have been the resource which led to resettlement of the area during the later Roman period, along with the best suited associated livestock, pigs and ponies. The presence of graves shows that people not only lived but probably died at or near the site. The clustering of the graves in the north of the excavated area implies that the full extent of a designated burial space may have been exposed in which case the population may have been small and the settlement short-lived, even allowing for the disturbance of the cemetery by the modern track. The two corn-

dryers, as well as querns, testify to the processing of cereals on the site but there is sparse evidence for arable or, indeed, livestock enclosures. There must also have been occupation as facilities which required significant time and skill in their construction as well as their operation would not have stood in isolation. It has been suggested that buildings at nearby Ibsley were of non-earthfast construction, although it was noted that linear features appeared to be converging on an area of unexcavated woodland under which the settlement core might lie (Coles and Ford

2016, 27, fig. 38). Such a mode of construction or the disturbance caused by earlier gravel extraction between the north and south areas of the site would allow similar explanations for the lack of identifiable Roman structures at

Somerley but truncation by modern land use may have erased some evidence.

Given that much of the cereal grain was indeterminate it is not clear whether the small amount of oats and barley and lack of identifiable wheat from the corn-dryers, or for that matter, any confirmed Roman context, is significant (Appendix 4), although the assemblages contrast markedly with richer returns from three later Iron Age pits and with those from some corn-dryers from further afield, for example at Catsgore, Somerset and Dorchester,

Dorset (Hillman 1982, 139; Jones and Straker 2002, tab. 33b). Parching of cereals improves their digestibility for animals and oat is the grain best suited to horses so that the dryers may have been used for parching winter fodder.

The querns indicate the reduction of cereal to flour for human use whilst demonstrating renewed contacts along the southern coastal strip, West Sussex providing at least one source during the later Roman period. A saddle quern which was the only example of sarcen in a later Roman context is likely to be residual or to have been re-used opportunistically and it is possible that the river was not a conduit for material from further inland. Fields at Ibsley dated to the 4th century, with a possible 3rd-century origin, and occupation sites with associated possible manure scatters below Gorley Hill suggest that crop production had resumed on the more suitable soils on the valley floor east of the Avon, offering a likely source of cereals for settlements to the west (Coles and Ford 2016, 79; Light et al. 1994, 76-8, fig. 23; Young 2008, 39, fig. 23).

Somerley might be regarded as a marginal site for meeting subsistence needs and the population living there would have depended on exchange with communities from further afield to attain a standard of living commensurate with better resourced settlements. The disruption in the late 4th and early 5th centuries AD of economic routines developed during the Roman occupation may have led to unsustainable pressure culminating in abandonment. Once again the perceived hiatus on the plateau contrasts with the evidence from the valley floor where chaff-tempered pottery found on several Roman settlement sites implies continuity of occupation into the

Early Saxon period (Light et al. 1994, 79). In a comparable setting at Ibsley charcoal from a sunken-floored building associated with organic-tempered pottery yielded a radiocarbon date from the decades either side of

AD600 (Coles and Ford 2016, 30).

Medieval

The earliest recognizable Medieval occupation appears to follow another hiatus in intensive land-use and dates to the late 11th or earlier 12th century, hence broadly contemporary with a group of paddocks previously identified to the south of the present site and possibly earlier than the burgeoning of settlement on the valley floor to the north

(Timby 2004, 7; Oram 2004, 8; Taylor 2009; Light et al. 1994, 85). The new orientation of fields paid heed neither to that established during the Iron Age linked to the ancient track, nor to sparse ditches of the later Roman period.

It gives every sign of being thoroughly planned. The organization of movement close to the buildings of the two main phases appears to relate to stock- rather than crop management. The buildings themselves evolved from a fairly simple rectangular plan to a developed later Medieval domestic plan and as such represent a significant addition to the literature for rural architecture of the period. They are small compared to halls of the time and the pottery is sparse so that throughout its occupation the settlement is unlikely to have been more than a farmstead.

The regular layout and extent of the field system suggests that it was probably a satellite of a manor (Smith 1999,

35).

The pottery implies that the earlier Medieval phase predated the first known reference to Somerley in 1272

‘when Thomas Baldwin granted Nicholas son of Philip le But an annuity from a messuage of land there’ (VCH

1911). At that time it formed part of the holding of the manor of . In Domesday Book. Harbridge was in the Ringwood Hundred held by Bernard, the king’s chamberlain. Prior to the Norman conquest it had comprised 5 hides valued at £4 10s but this had fallen to 3 hides and 1 virgate valued at 70s. There were ten households with four ploughs, 80 acres of meadow and woodland for 2 pigs (Williams and Martin 2003, 127). The site lies slightly beyond the western fringe of the forested area designated for royal hunting by William I in or before 1079, an act which precluded much enclosure from that time until the present. Some enclosure during that time, was probably illegal (Smith 1999, 1-2, 49). Legal enclosure in the forest during the 13th century included ten vaccaries (Smith

1999, 21-2). Typically they were extensive oval tracts which contrast sharply with the linear system in a broadly similar landscape at Somerley. Aside from providing grazing for dairy herds for the crown and commoners the forest was used for pannage, the earthworks of late Medieval and early Post-medieval pig pounds providing physical evidence of an older tradition (Smith 1999, 31-3).

The settlement seems to have gone into decline well before the end of the Medieval period. This may have coincided with the fall in population associated with the plague during second half of the 14th century, a period during which arable land was often abandoned but when herdsmen would also have been in short supply (Smith

1999, 50).

Acknowledgements

The excavation was funded by Tarmac Southern Limited. The authors wish to thank all who participated on site

and behind the scenes. The excavation was supervised by Andy Taylor and the team consisted of, Kyle

Beaverstock, Dan Bray, Aiji Castle, Aidan Colyer, Steve Crabb, James Earley, Matt Gittins, John Kaines and

James McNicoll Norbury. Illustrations are by the authors, Andy Mundin and Amanda Tabor.

References Adam, N J, Seager Smith, R and Smith, R J C, 1997, ‘An early Romano-British Settlement and Prehistoric field boundaries at Dairy Lane, Nursling, Southampton’, Hampshire Stud 52, 1–58 .Addyman, A and Leigh, D, 1973, ‘The Anglo-Saxon Village at Chalton, Hampshire: Second Interim Report’, Medieval Archaeol 17, 1–25 Addyman, A, Leigh, D and Hughes, M, 1972, ‘Anglo-Saxon Houses at Chalton, Hampshire’, Medieval Archaeol 16, 13–31 Anthony, S, 2002, Nea Farm Quarry Phase 3, Somerley, near Ringwood, Hampshire, An archaeological evaluation, Thames Valley Archaeological Services (TVAS), unpubl rep 02/03, Reading Anthony, S and Ford, S, 2003, Nea Farm Somerley, near Ringwood, Hampshire phase 2, A post-excavation assessment, TVAS unpubl rep 01/41c, Reading Barrett, J, Bradley, R and Green, M, 1991, Landscape, Monuments and Society: The prehistory of Cranborne Chase, Cambridge Barton, R N E, Ford, S, Collcutt, S N, Crowther, J, Macphail, R I, Rhodes, E and Van Gijn, A, 2009, ‘A Final Upper Palaeolithic site at Nea Farm, Somerley, Hampshire and some reflections on the occupation of Britain in the Late Glacial Interstadial’, Quartär 56, 7–35 BGS, 1990, British Geological Survey, 1:50000, Sheet 314, Solid and Drift Edition, Keyworth Bowen, H C, 1979, ‘Gussage in its setting’, in G Wainwright, Gussage All Saints: An Iron Age Settlement in Dorset, London, 179-83 Bradley, R, Entwhistle, R and Raymond, F, 1994, Prehistoric Land Division on Salisbury Plain: The work of the Wessex Linear Ditches Project, Engl Heritage Archaeol Rep 2, Swindon Brickley, M and McKinley, J (eds), 2004, Guidelines to the Standards for Recording Human Remains, IFA Pap 7 Brown, L, 2000, ‘The regional ceramic sequence’, in B Cunliffe, B, 2000, The Danebury Environs Programme: The Prehistory of a Wessex Landscape, volume 1, Introduction, Oxford, 79–124 Brown, L, Corney, M and Woodward, P, 1996, ‘An Iron Age and Romano-British Settlement on Oakley Down, Wimborne St Giles, Dorset’, Proc Dorset Natur Hist Archaeol Soc 117, 67–79 Buikstra, J E and Ubelaker, D H, 1994, Standards for data collection from human skeletal remains Arkansas Archaeological Survey Research Series, 44, Fayetteville, Ark Butterworth, C, 1995, ‘Archaeological Excavation and Watching Brief at Ellingham farm, Near Ringwood, Hampshire, 1988-1991, Proc Hampshire Fld Club Archaeol Soc 51, 59–76 Butterworth, C, Cleal, R and Morris, E, 1995, ‘Discussion’, in C Butterworth, ‘Archaeological Excavation and Watching Brief at Ellingham farm, Near Ringwood, Hampshire, 1988-1991, Proc Hampshire Fld Club Archaeol Soc 51, 74–5 Cadman, G, 1983, ‘Raunds 1977-1983: An Excavation Summary’, Medieval Archaeol 27, 107–22 Cass, S, 2008, Nea Farm, Phase GP 4, Somerley, Ringwood, Hampshire: an archaeological excavation, TVAS unpub rep 07/53,Reading Coles, S and Ford, S, 2016, Neolithic, Bronze Age, Roman and Anglo-Saxon Occupation, and Bronze Age Burial at Ibsley Quarry, Ibsley, Ringwood, Hampshire, TVAS Monogr 25, Reading Cunliffe, B, 1987, Hengistbury Head, Dorset, volume 1: The Prehistoric and Roman Settlement, 3500 BC–AD 500, Oxford Univ Comm Archaeol Mongr 13, Oxford Cunliffe, B W, 1995, Danebury: an Iron Age Hillfort in Hampshire, Vol 6: A Hillfort community in perspective CBA Res Rep 102, York Cunliffe, B (ed), 2000, The Danebury Environs programme; the prehistory of a Wessex landscape (vol 1), Engl Heritage/Oxford Univ Comm Archaeol Monogr 48, Oxford Cunliffe, B, 1984, Danebury: An Iron Age hillfort in Hampshire. Volume 2. The excavations, 1969-78: the finds, London Cunliffe, B and Poole, C, 1991, Danebury: An Iron Age hillfort in Hampshire. Volume 5. The excavations, 1979- 88: the finds, CBA Res Rep 73, London

Cunliffe, B and Poole, C, 2000, Windy Dido, Cholderton, Hants, 1995: The Danebury Environs Programme: The Prehistory of a Wessex Landscape, vol. 2 pt. 7, Oxford Univ Comm Archaeol Monogr 49, Oxford Currie, C and Fletcher, J, 1972, ‘Two Early Cruck Houses in North Berkshire identified by Radiocarbon’, Medieval Archaeol 16, 136-42 Davies, S, Bellamy, P, Heaton, M and Woodward, P, 2002, Excavations at Alington Avenue, Fordington, Dorchester, Dorset, Dorset Natur Hist Archaeol Soc Monogr 15, Dorchester Defoe, W, 1927, A Tour through the Whole Island of Great Britain, volume 1, London Dyer, C, 1986, ‘English peasant buildings in the later Middle Ages (1200-1500)’, Medieval Archaeol 30, 19–45 Fasham, P J, 1985, The Prehistoric Settlement at Winnall Down, Winchester: excavations of MARC3 site R17 in 1976 and 1977, Hampshire Fld Club Archaeol Soc Monogr 2, Gloucester Fasham, P J, Farwell, D E and Whinney, R J B, 1989, The Archaeological Site at Easton Lane, Winchester, Hampshire Fld Club Archaeol Soc Monogr 6, Gloucester Fitzpatrick, A, 2008, ‘Later Bronze Age and Iron Age’ in C Webster (ed), The Archaeology of South West England: South West Archaeological Research Framework resource Assessment and Research Agenda, Taunton, 117–44 Ford, S, 1987, ‘Chronological and functional aspects of flint assemblages’, in A G Brown and M R Edmonds (eds), Lithic Analysis and Later British Prehistory, BAR 162, Oxford, 67–85 Ford, S, 1992, Nea Farm Quarry, Somerley, near Ringwood, Hampshire, an assessment of the proposed sand and gravel extraction (input to Environmental Statement), TVAS unpubl rep 92/21, Reading Ford, S, 2001a, Nea Farm Quarry, Somerley, near Ringwood, Hampshire Phase 2, an archaeological evaluation, TVAs unpubl rep 01/41, Reading Ford, S, 2001b, Nea Farm Quarry, Somerley, near Ringwood, Hampshire Phase 2, An archaeological evaluation - stage 2 (test-pitting), TVAS unpubl rep 01/41b, Reading Ford, S, 2002, ‘Struck flint’ in S Anthony, Nea Farm Quarry Phase 3, Somerley, near Ringwood, Hampshire, An archaeological evaluation, TVAS, unpubl rep 02/03, Reading, 5 Ford, S and Hall, M, 1993, Somerley Estate, Ringwood, Hampshire, Archaeological Evaluation, TVAS unpubl rep 92/21b, Reading Fulford, M, 2014a, ‘The Roman Period: Resource Assessment’, inG Hey and J Hind, Solent-Thames Research Framework for the Historic Environment: Resource Assessments and Research Agendas, Oxford Wessex Monogr 6, 155–78 Fulford, M, 2014b, ‘The Roman Period: Research agenda’, in G Hey and J Hind, Solent-Thames Research Framework for the Historic Environment: Resource Assessments and Research Agendas, Oxford Wessex Monogr 6, 179–84 Gardiner, M, 2000, ‘Vernacular buildings and the development of the later Medieval domestic plan in England’, Medieval Archaeol 44, 159–79 Hewett, C, 1973, ‘Songers, Cage Lane, Boxted, Essex’, Medieval Archaeol 17, 131–2 Hey, G and Hind, J, 2014, Solent-Thames Research Framework for the Historic Environment: Resource Assessments and Research Agendas, Oxford Wessex Monogr 6, Oxford Holbrook, N, 2008, ‘Roman’, in C Webster (ed), The Archaeology of South West England: South West Archaeological Research Framework resource Assessment and Research Agenda, Taunton, 151–61 Jefferies, J, 1979, ‘The Pits’, in G Wainwright, Gussage All Saints: An Iron Age Settlement in Dorset, Dept of Environment Archaeol Rep 10, London, 9–15 Jones, J and Straker, V, 2002, ‘Macroscopic plant remains’, in S M Davies, P S Bellamy, M J Heaton, M and P J Woodward, Excavations at Alington Avenue, Fordington, Dorchester, Dorset, 1984-87, Dorchester, 118–21 Leech, R, 1982, Excavations at Catsgore, 1970-1973: A Romano-British Village, Western Archaeological Trust Excavation Monogr 2, Bristol Light, A, Schofield, J and Shennan, S J, 1995, ‘The Middle Avon Valley Survey: a study in settlement history’, Proc Hampshire Fld Club Archaeol Soc 50, 43–101 Lock, G, 1995, ‘Pits and Propitiation’, in B Cunliffe, Danebury: An Iron Age hillfort in Hampshire. Volume 6. A hillfort community in perspective, York, 80–8 McKenna, R, 2009, ‘Carbonized plant remains’, in A Taylor, ‘Nea Farm, Phase GP5, Somerley, Ringwood, Hampshire, 2007, An Archaeological Excavation (Draft publication report)’, TVAS rep 07/114, Reading Mepham, L, 1989, ‘The Iron Age Occupation at Dibble’s Farm, Christon’, Proc Somerset Archaeol Natur Hist Soc 132, 23–81 Morley Hewitt, A, 1965, The Story of Fordingbridge, Fordingbridge CSAI, 2017, Soilscapes, http://www.landis.org.uk/soilscapes/ (accessed: 13th February 2017), Cranfield Soil and Agrifood Institute Oram, R, 2004, ‘Nea Farm, Somerley, Ringwood, Hampshire: an archaeological excavation’, TVAS unpubl rep 04/53, Reading Peacock, D P S, 1987, ‘Iron Age and Roman quern production at Lodsworth, West Sussex’, Antiq J, 67, 61–85

Pearce, J E, Vince, A G and Jenner, M A, 1985, A Dated Type-Series of London Medieval Pottery Part 2: London- type Ware, LAMAS Spec Pap 6, London Pine, J, 2003, ‘Nea Farm Quarry, Somerley, Ringwood, Hampshire Phase 4, An archaeological evaluation’, TVAS unpubl rep 03/68, Reading Pine, J and Bennett, N, 2010, ‘Nea Farm, Phase GP6, Somerley, Ringwood, Hampshire, An Archaeological Excavation Draft Publication Report’, TVAS unpubl rep 09/68, Reading Platt, D, 2010, ‘Nea Farm, Phase GP7, Somerley, Ringwood, Hampshire, An Archaeological Excavation (Draft Publication Report)’, TVAS unpubl rep 10/17, Reading Poole, C, 1984, ‘Objects of baked clay’, in B Cunliffe and C Poole, Danebury: An Iron Age hillfort in Hampshire. Volume 2. The excavations, 1969-78: the finds, London, 398–407 Poole, C, 1991, ‘Objects of baked clay’, in B Cunliffe and C Poole, 1991, Danebury: An Iron Age hillfort in Hampshire. Volume 5. The excavations, 1979-88: the finds, London, 370–82 Poole, C, 1995, ‘Loomweights versus oven bricks’, in B Cunliffe, Danebury: An Iron Age hillfort in Hampshire. Volume 6. A hillfort community in perspective, York, 285–6 PPG16, 1990, Archaeology and Planning, Dept Environment Planning Policy Guidance 16, HMSO Raymond, F, 2008, ‘Prehistoric Pottery’, in A Taylor, ‘Nea Farm, Phase GP3, Somerley, Ringwood, Hampshire, 2005, An Archaeological Excavation, (Draft publication report)’, TVAS rep 05/62, Reading, 6–9 Raymond, F, 2009, ‘The Beaker’, in A Taylor, ‘Nea Farm, Phase GP5, Somerley, Ringwood, Hampshire, 2007, An Archaeological Excavation, (Draft publication report)’, TVAS rep 07/114, Reading, 7–9 Reimer P J, Bard E, Bayliss A, Beck J W, Blackwell P G, Bronk Ramsey C, Buck C E, Cheng H, Edwards R L, Friedrich M, Grootes P M, Guilderson T P, Haflidason H, Hajdas I, Hatté, C, Heaton T J, Hogg A G, Hughen K A, Kaiser K F, Kromer B, Manning S W, Niu M, Reimer R W, Richards D A, Scott E M, Southon J R, Turney C S M, van der Plicht J, 2013, ‘IntCal13 and MARINE13 radiocarbon age calibration curves 0-50000 years cal BP’, Radiocarbon 55(4), 1869–1887 Schweingruber, F H, 1978 Microscopic Wood Anatomy, Birmensdorf Shaffrey, R and Roe, F, 2011, ‘The Widening use of Lodsworth Stone: Neolithic to Romano-British Quern Distribution’, in D F Williams and D P S Peacock (eds), Bread for the People: the archaeology of mills and milling, BAR Int Ser 2274, Oxford, 309–24 Sharples, N, 1991, Maiden Castle: Excavations and field survey 1985-6, London Smith, A, 1996, ‘Somerley Estate, Ringwood, Hampshire, Archaeological Excavation’, TVAS rep 95/64b, Reading Smith, N, 1999, ‘The Earthwork Remains of Enclosure in the New Forest’, Hampshire Stud 54, 1–56 Stace, C, 1997, New flora of the British Isles, Cambridge Sunter, N and Woodward, P, 1987, Romano-British Industries in Purbeck: Excavations at Norden; Excavations at Ower and Rope Lake Hole, Dorset Natur Hist Archaeol Soc Monogr 6, Dorchester Tabor, R, 2008, Cadbury Castle: The hillfort and landscapes, Stroud Taylor, A, 2008, ‘Nea Farm, Phase GP3, Somerley, Ringwood, Hampshire, 2005, An Archaeological Excavation, (Draft publication report)’, TVAS rep 05/62, Reading Taylor, A, 2009, ‘Nea Farm, Phase GP5, Somerley, Ringwood, Hampshire, 2007, An Archaeological Excavation, (Draft publication report)’, TVAS rep 07/114, Reading Timby, J, 2004, ‘Pottery’, in R Oram, ‘Nea Farm, Somerley, Ringwood, Hampshire: an archaeological excavation’, TVAS unpubl rep 04/53, Reading, 6-7 VCH, 1911, 'Parishes: Harbridge', in A History of the County of Hampshire: Volume 4, ed. William Page (London, 1911), pp. 604-606. British History Online http://www.british-history.ac.uk/vch/hants/vol4/pp604-606 (accessed: 22nd February 2017) Wainwright, G, 1979, Gussage All Saints: An Iron Age Settlement in Dorset, Dept of Environment Archaeol Rep 10, London Wainwright, G and Davies, S, 1995, Balksbury Camp, Hampshire: Excavations 1973 and 1981, Engl Heritage Archaeol Rep 4, London Weaver, S, 1995, ‘Somerley Estate, Ringwood, Hampshire, Archaeological Evaluation’, TVAS unpubl rep 95/64, Reading Webster, C (ed), 2008, The Archaeology of South West England: South West Archaeological Research Framework resource Assessment and Research Agenda, Taunton Williams, A and Martin, G, 2003, Domesday Book: A Complete Translation, London Whimster, R, 1979, Burial Practices in Iron Age Britain, vol 2, unpubl PhD thesis, Univ Durham Young, A, 2008, The Aggregate Landscape of Hampshire: Results of NMP Mapping, Engl Heritage/Cornwall