Two Thousand Years of Salt Making at Lymington, Hampshire

Proc. Hampshire Field Club Archaeol. Soc. 64, 2009, 9-40 (Hampshire Studies 2009)

TWO THOUSAND YEARS OF SALT MAKING AT LYMINGTON, HAMPSHIRE

By ANDREW B. POWELL with contributions firm KAYT BROWN, GRACE PERPETUA JONES, LORRAINE MEPHAM, DAVID NORCOTT, RUTH PELLING and CHRIS J. STEVENS with illustrations by ROB GOLLER and ELIZABETH JAMES

ABSTRACT has always been an essential element of coastal salt production in Britain. In the 18th century, A programme of archaeological fieldwork on land Lymington's prosperity derived largely from at Manor Farm, Pennington, south of Lymington, the thriving salt making industry based in the revealed evidence of the coastal salt making industry Keyhaven and Pennington Marshes which flank which was a major feature in the medieval and post- the north shore of the western Solent between medieval landscapes, and'upon which the town's Avon Water and the Lymington River estuary prosperity in the 17th and 18th centuries was largely (Fig. 1). built. It also revealed an Iron Age settlement, compris- Evidence for the long-term nature of salt ing at least ten roundhouses and numerous pits, some making in the area is provided by documents apparently clay-lined; finds ofbriquetagefrom the set-dating from the 11th to the mid-19th century, tlement indicate that salt making was a significant as well as by maps, drawings and historical component of its possibly seasonal economic base. accounts. Today, traces of the salterns can be Although medieval salt making at Lymington is seen in aerial photographs and, in places, are attested to in documentary sources these have little still visible on the ground as low banks, mounds to say about the methods used until after well doc- and ditches (Momber et al. 1994). Yet there has umented changes in the industry in the early 17th been only limited archaeological investigation century, when saltern owners were required to replaceof this important industry, which at its height their 'mounds' with 'floor pans'. Following these was exporting its product across the Atlantic to major developments, there are a number of precise Canada and the United States. descriptions and graphical depictions of salt pro- A programme of archaeological works duction methods at Lymington. Despite the fact that revealed evidence not just of medieval and 12th—13th century pottery was recovered from the site,post-medieval salt making, but also for its late the archaeological evidence for historic salt making Iron Age precursor. The fieldwork, including corresponds more closely to the descriptions of the post-earthwork surveys, archaeological strip-and- medieval industry, although in the absence of firm record excavations and watching briefs, was dating evidence questions remain about the precise undertaken on c. 8 hectares of land at Manor date and function of the features uncovered. Farm, Pennington, south of Lymington, centred on NGR 431650 092500, in advance of gravel extraction and subsequent landfill (Fig. 2). INTRODUCTION A key factor for salt making along the Lymington coasdine was the storm protec- 'Salt', said Pythagoras, 'is born of the purest of tion offered by Hurst Spit to the west. The site parents: the sun and the sea'. While the British occupies generally level ground at c. 0.7-1 m climate does not allow for the production of above Ordnance Datum (aOD) and, even crystalline salt by just the solar evaporation of accounting for sea level rise since the medieval sea water, the use of the sun and die wind to period, would, if open to the sea, have been produce a concentrated brine from sea water liable to flooding on spring tides when the

9 10 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Southampton

\eV

Fig. 1 Site location E*OWELL: TWO THOUSAND YEARS OF SALT-MAKING AT I.YMINflTON, HAMPSHIRE 11 mean high water reaches 0.87 m at Hurst Point at the north end of the site (Fig. 3). This area, and 1.02 m aOD at Lymington. The site is now which comprised the highest ground within bounded to the south by a bund that formerly the site, probably lay just above the tidal range, separated Pennington Marshes from the on the edge of the salt marsh. One sherd of farmland to the north, while further south is flint-tempered pottery, recovered along with the sea-wall built in the 18lh century. burnt flint and fired clay from a probable tree- The site lies within a belt of land mapped as throw (7281) cut by the gully of roundhouse 7 alluvium (Geological Survey of Great Britain (below), could be late Bronze Age in date, but 1075). The alluvium marks the line of a small no other material of this dale was found on the watercourse that runs off the 'Angular Flint and site. Gravel of the Downs' (AFGD) which lies inland of the coastal marshes. The AFGD deposits line Roundhouses all the extant watercourses in the area (e.g. the Ten possible roundhouses (numbered 1-10 Lyminglon River and Avon Water) and appear from north-west to south-cast) were recorded, to mark the course of several former streams defined by curved gullies c. 8-12 m in internal that flowed into the marshes between these diameter (summarised in Table 1). They rivers. All the watercourses cut through Pleis- include ac.8 m long arc of gully (roundhouse tocene Plateau Gravel. During excavation of 4) with a projected possible diameter of 13-14 the late Iron Age settlement at the northern m, making it significantly larger than the other end of the site the natural geology was recorded roundhouses, and possibly indicating some as sandy clay loam. The local soils are clas- other function. sified as EfTord 1 (marine and river terrace In spite of their often dose proximity, none gravel: well-drained fine loamy soils, often over of the structures overlapped. In several cases, gravel; Soil Survey of England and Wales 1983) however, they appear to have been rebuilt - the although, on the ground, elements of the sur- gully of roundhouse 9 had been recut at least rounding Shabbington (river terrace drift: once (possibly twice), while roundhouses 2 and slowly permeable seasonally waterlogged fine 6 both had concentric gullies. The three gullies loam over clayey soils) and Wallasea 1 (marine in roundhouse 6 partly overlapped, but the two alluvium: stoneless non-calcareous and calcare- in roundhouse 2 were separated by up to 0.5 ous clayey soils) soils were observed. in, representing either a slight variation in size Fifteen episodes of ficldwork were under- (8.7 m and 10 m diameter) or a slight shift in taken as different areas were cleared for gravel position. extraction, the excavation in most areas being Most of the roundhouses had entrances preceded by a topographic survey (Wesscx facing between north-east and south-east, Archaeology 1999; 2000). The site was divided although roundhouse 5 was notable in having a by a bentonite dam into two parcels of land, the west-facing entrance (positioned less than 2 in strip between them not being examined. The from the south-east facing entrance of round- various areas are here combined into Areas A, house 3). In addition, roundhouses 2 and 10 B and C (Fig. 2). appear to have had entrances facing both east and west. The gullies were generally of similar dimen- RESULTS sions, averaging c. 0.5 m wide and 0.2 m deep. Roundhouse 9, which had a slightly square IronAgp shape, had a more substantial gully, its first recut being up to 1.7 m wide and 0.5 m deep; in The evidence for late Iron Age occupation, both of its main phases its entrance terminals comprising mainly roundhouses, ditches had postholes cut into the gully base. The gully and pits, was found exclusively in Area G, a of roundhouse 1 appeared in places almost as a triangle of land measuring 100 m by 240 in, scries of stakeholes/postholes. 12 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Key: • Limits of excavation Mound 092700 ™ Cut feature

092500

Area A

350 ffl 100m

Fig. 2 Fieldwork areas, showing Iron Age settlement (Area C) and medieval/post-medieval salterns (Areas A and B)

A range of finds was recovered from the quantities of briquetage and burnt flint. The roundhouse gullies, including late Iron only animal bone from the site came from the Age pottery, fired clay including briquetage gully of roundhouse 9. (fragments of vessels, rods and blocks) and Few features were noted within the round- burnt flint. This material was often concen- houses - two possible postholes in roundhouse trated in the gully terminals, the southern 6, two pits and a c. 3 m long slot (7216) lying terminal of roundhouse 7, for example, con- 3 m inside the entrance in roundhouse 7, taining a handled jar (Fig. 7.10) and pieces and a number of pits in roundhouse 9. None, of a fired clay loomweight, while the southern however, could be clearly associated with the terminal of roundhouse 2 contained large structures. POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 13

Table 1 Summary of Iron Age roundhouses

Roundhouse Internal diameter (m) Orientations of entrances Number of gullies/cuts

1 9.2 SE 1 inner 8.7 2 outer 10.0 ESE-WNW 2 3 9.5 ESE 1 4 <14.0 - 1 5 9.7 WNW 1 inner 9.6 middle 10.7 6 outer 11.7 E >3 7 9.6 SSE 1 8 11.2 NE? 1 9 10.5 ESE 3 10 8.0 E - W 2

Pits where a layer of clay was recorded only on the Numerous small circular features were recorded base, as for example in pit 2798 (group 3). It in Area C, many of them occurring in small was observed during the excavation that those groups (PG 1-6) (Fig. 3 and Table 2). Many of pits holding rain water rapidly accumulated a those that were excavated proved to be similar in well-sorted puddled clay 'lining' in their bases, form and contents, averaging 0.7 m in diameter the clay having eroded out of the surrounding and 0.2 m deep, with moderately steep sides and sandy clay loam geology. flat or slightly concave bases. They contained a similar range of finds to that found in the round- Other features house gullies, comprising late Iron Age pottery, A pear-shaped feature (2997) east of round- fired clay (including briquetage fragments) and house 10, was 1.1m long, with the rounded end burnt flint. A number also contained varying 0.6 m wide and 0.2 m deep, narrowing to c. 0.3 quantities of charcoal, but in only one (pit 2999, m and the base sloping up to the east. Its single group 3), which had a patch of burnt clay above fill of dark brown silty clay contained over 2.5 a basal clay layer, was there any evidence for in kg of burnt flint as well as briquetage, including situ burning (Fig. 4). vessel fragments, but while there were occa- The majority of the pits were recorded as 'clay- sional flecks of charcoal, there was no evidence lined'. Some did have a layer of clay covering of in situ burning. the base and extending up the sides of the pit Between roundhouses 9 and 10 there was a (e.g. pits 2999, group 3; 7124 group 4; 7226, shallow oval cut (2882), 1.2 m by 1.5 m and up group 1) (Fig. 4), and a reasonable interpre- to 0.16 m deep, with a single fill with a scatter tation is that these pits were designed to hold of burnt flint on its surface, while pit 2732 (cut water, possibly brine, so playing some role in by gully of roundhouse 8), measured 0.8 m in the salt making process. There is some doubt, diameter and 0.2 m deep, with a basal clay layer however, as to the identification of a number extending up the sides, and a single fill contain- of the pits as clay-lined, particularly in those ing fired clay, burnt flint, and a sherd of pottery. 11 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Fig. 3 Iron Age settlement in Area C

NW SE F w sw NE

"» \--W 7125 -\ 7228 7001/ ^~~ j J 7126 ^ 7227 - ^ C r -^ 7002 I 2999 7'124 7226

0 I m —i ID

Fig. 4 Clay-lined Iron Age pits: 2999 (group 3), 7124 (group 4) and 7226 (group 1) POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 15

Table 2 Summary of Iron Age pit groups

Group Pit Width Depth Fills Contents Evidence for clay lining

PG1 7226 0.7 0.3 2 burnt flint Clay layer extending halfway up vertical sides (Fig. 4) 7249 0.7 0.2 1 burnt flint fired clay (incl. briquetage vessel frag.) + 3 possible postholes PG2 7301 0.7 0.1 2 Late Iron Age pottery Clay layer on base burnt flint fired clay (incl. briquetage vessel frags.) Emmer wheat glume base 7152 0.6 0.1 1 fired clay burnt flint burnt soil? PG3 2798 0.6 0.25. 2 Late Iron Age pottery Clay layer on base burnt flint fired clay charcoal 2999 0.7 0.2 3 Late Iron Age pottery Clay layer on base and burnt flint sides, below patch of fired clay clay burnt in situ (Fig. possible iron-working slag 4) + 1 unexcavated pit PG4 7124 0.8 0.25 2 burnt flint Clay layer on base and fired clay sides (Fig. 4) 7043 0.6 0.1 1 burnt flint - fired clay + 4 unexcavated pits PG5 7300 0.8 0.1 3 charcoal Clay layer on base and burnt soil/clay sides 7305 0.6 0.25 3 charcoal Clay layer on base burnt soil/clay 7247 0.6 0.15 fired and burnt clay (incl. possible briquetage Clay layer on base frags.) PG6 7130 0.7 0.1 1 -- 7209 0.9 0.15 burnt flint — fired clay charcoal 7170 0.7 0.12 burnt flint - fired clay (incl. briquetage block frags.) 7204 0.7 0.1 burnt flint - fired clay charcoal + 1 unexcavated pit and 2 postholes 16 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Linear features ditches, although not excavated at that point, Area C was crossed by a grid of regularly spaced and while the quantity of material suggests that east-west ditches of probable post-medieval/ the ditch was contemporary with the Iron Age modern date, some of which cut across the occupation of the site, the phasing of these roundhouses (Fig. 3). However, a number of linear features remains uncertain. ditches had other orientations, some with less regular lines, and a few appear to pre-date Medieval, post-medieval and modern some Iron Age features. These include a poorly To the south of the area of Iron Age activity, denned and slightly meandering linear feature the archaeological remains across the rest of (7299) orientated north-east and traced for over the site (Areas A and B, Fig. 2) were different 15 m from a possible terminal in the entrance in character, comprising both cut and topo- of roundhouse 6. It also passed through the graphic features, and consisting mainly of low entrance of roundhouse 7 but was cut by the earthwork mounds and banks, areas of associ- building's gully at the north, beyond which it ated shallow ditches arranged in regular grid appeared to peter out. It was up to 1.2 m wide patterns and a range of pits, all these features and 0.2 m deep, with a single fill containing late likely to be direcdy related to salt production. Iron Age pottery, fired clay and burnt flint. Although it was possible, during most episodes A similar feature (2812) on approximately of strip and record and watching brief, only to the same orientation, but offset some 3 m to digitally survey many of these features, sample the south-west, ran south-west from a possible excavation has helped to clarify their nature, if terminal within roundhouse 6, c. 6 m south- not precisely their function. west of the terminal of feature 7299. After 30 Dating these features is also problematic m it turned to the west (parallel to the modern since, although there is documentary evidence ditch bounding the area of higher ground), for salt working on the Lymington coasdine before petering out after a further 10 m. through the medieval and post-medieval Although this feature cut across the gully of periods and beyond, few artefacts were found roundhouse 6, its shared orientation to feature in this area of the site. The only datable finds 7299 and similar dimensions and contents from Area A and B were 26 sherds (470 g) of - pottery, fired clay including briquetage and 12th-13th century pottery recovered from two burnt flint (one section containing 4.6 kg of relatively limited areas within the site. burnt flint) - suggest that they may be contemporary; a fragment of modern brick from the Early features feature is probably intrusive. Although no features clearly datable as medieval Some 35 m to the east, and broadly parallel to were identified on the site, a large number of features 7299 and 2812, there was a more clearly undated pits, many in irregular clusters, were denned and substantial ditch (2727), which at recorded within the ditch systems in Area B. the north passed between roundhouses 8 and 9 Many were broadly similar in character, c. 1-2 and at the south continued into an undefined m wide and 0.3-0.7 m deep, often with irregular spread of soil. It was 1.2 m wide and 0.3 m profiles (due in part to erosion of the pit sides) deep with a concave profile, and in its single and uneven bases. Their fills were quite mixed, excavated section, a charcoal-rich lowest fill was often including lumps of redeposited natural, overlain by material slumped from either side, lenses of very dark peaty material and laminated then a layer of possibly natural silting above silts. While some contained fragments of burnt which was an upper fill comprising a large or fired clay, none showed any traces of clay quantity of dumped burnt flint (some 76 kg was linings and their function is unclear. Although recovered) and charcoal, along with late Iron no datable material, either Iron Age, medieval Age pottery and fired clay including brique- or post-medieval was recovered from these pits, tage block fragments. This ditch was surveyed they appear to pre-date the ditches among which as cutting one of the post-medieval/modern they were recorded, and the recovery from one POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 17 pit (7313, Fig. 3) of four seeds of Anthemis cotula of the blocks showed signs of modification, (stinking mayweed) suggests that these features with ditches being recut, in only one, at the are more likely to be associated with a medieval north-east corner of Area B, had this appar- phase of salt making than with the late Iron endy involved the complete realignment of the Age phase, to which pit 7313 had initially been ditches - at right-angles to their previous orien- tentatively assigned (see Pelling, below). tation. Particularly in Area A, it was possible to Five lengths of ditch in Area A also pre-dated establish some relative phasing between blocks those features clearly identifiable as relating to where adjacent blocks overlapped. the historic saltworks, none of them having the There was little difference between the dominant north-south alignment of the saltern profiles of the blocks' feeder ditches and the ditches in this part of the site. A group of three cross ditches running off them, although in (4001) - two parallel ditches 30 m apart aligned some blocks the feeder ditches were wider and WNW.-ESE. and one perpendicular between deeper. The ditches averaged c. 2 m wide and them - were located at the north (Fig. 5), while where excavated proved to be 0.3-0.4 m deep at the south-east and south-west, respectively, with shallow, generally V-shaped profiles. Most ditches 4002 and 350 were aligned approxi- were filled with very dark, almost black, sandy mately NE.-SW. (Fig. 2). In contrast to the later silt, some with orange and grey motding, the ditches they contained no burnt inclusions and lower fills often being finely laminated, as would their function is unclear. be expected of water-lain deposits. However, The group 4001 ditches were sealed by a despite the black, apparently charcoal-rich layer of clay, up to 0.1 m thick. A similar layer nature of these fills only small quantities of was observed more widely across Area A (e.g. charcoal were recovered from die samples layer 213, Fig. 6), and also in Area B and was taken. provisionally interpreted as a pre-saltern flood In some blocks the feeder ditches bounded deposit, as it was cut by the saltern ditches and three sides, with the cross ditches extending overlain by mound 3010 (but see Discussion from opposite sides towards a central mound below). It was covered by a thin smear of black (see below). The feeder ditches must have been silt identical to the material filling the later connected to channels running either direcdy ditches. from the sea or, more likely from seawater holding ponds fed by the high tides, but no such Ditch blocks features were identified on the site. The spacing The most noticeable feature revealed by the of the cross ditches within each block was rela- fieldwork in Areas A and B was an array of sub- tively uniform across the site - generally c. 7-9 square blocks of connected ditches, each block m between the inner edges of adjacent ditches, comprising a number of parallel cross ditches although this was often narrowed where ditches connected at one end to a 'feeder' ditch that had been recut on slightly different lines. The bounded two or more sides of the block. relatively few features recorded within the rec- Some of these ditches were faindy visible as tangular strips between the ditches, mosdy pits, earthworks, and recorded in the topographic appear not to have been located with reference surveys. Their form was most clearly discernible to the ditches and therefore seem not to be in Area A. directly associated with them. The blocks varied in size from c. 50-80 m square, and while those in Area A were all Banks aligned approximately north-south, many of A number of the ditch blocks were clearly those in Area B had different orientations, bounded by low earth banks, and although giving the impression that the latter in par- these had suffered from levelling, their dispo- ticular had developed piecemeal over a period sition, particularly in Area A, was highlighted of time, possibly filling in the intervening by the topographic survey. The block at the spaces between existing blocks. While many south-west corner of the site, for example, was IS HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Fig. 5 Detail of Areas A and B including Mound 3010 and adjacent ditch blocks POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 19 bounded by banks to the east, north and west these ditch terminals suggests that the eastern (the bund to the south concealed the block's edge of the original mound had lain further southern extent). While the feeder ditch would west. When the later phase mound was con- have passed through a break in the bank, with structed or extended eastwards, these ditches water flow possibly controlled by a sluice gate, had silted up with black sandy silt to a depth of the breaks identified in these banks during the no more than 0.1 m; they had then been back- topographic survey all post-date the abandon- filled with grey brown sandy clay containing ment of the salt works. some gravel, occasional charcoal and flecks of Such banks may have had a range of burnt/fired soil/clay. A small pit (278), sealed functions- defending each ditch block by a spread of burnt material (342), was also against river flooding or inundation from the overlain by the later phase mound (Fig. 6). sea, providing dry access routes across the The later phase mound, as revealed in salt working area and to the central mounds section, appears initially to have been bounded within the blocks, and also possibly defining by a bank or 'kerb' of redeposited sandy clay the boundaries of individual saltern properties, (337 and 343), possibly derived from the exca- identifying the block as the basic unit of salt vation of the group 4011 ditches. This enclosed production. an area 17 m wide. A further redeposited layer Bank 2392, which ran north from a large (336) had been dumped against the inner edge earthwork mound in the centre of Area B, of the kerb on the south side of the mound, was flanked on either side by probable feeder but across the rest of the base there was a layer, ditches supplying water to the blocks to the up to 0.3 m thick, of red burnt soil (272), in east and west, the ditch to the east apparently some places apparently burnt in situ, contain- recut on two occasions. A section excavated ing frequent burnt pebbles and fragments of close to the mound showed that the bank was charcoal. A concave hollow, 0.5 m in diameter built up over a palaeosol cut by the ditches, and and 0.3 m deep, in the top of this layer was comprised varied layers of dumped material filled with a dump of grey-green clay (341). surviving to a height of c. 0.4 m, including Overlying the burnt soil was up to 0.6 m of redeposited sandy clay (possibly from the ditch black sandy silt (271), containing charcoal, construction) and dark brown silty clay, with ash, burnt clay and vitrified soil. This extended small patches of orange heat-affected soil/clay. over the clay bank and 2-3 m beyond it, giving the mound an overall width of c. 25 m; while Mounds its edge covered some of the terminals of the The mounds were irregular in shape, located block 4011 's cross ditches, this may be due to either between ditch blocks or in their centres the subsequent levelling and spreading of the and reached by banks from one side. While mound material. some survived as low earthworks, others appear The three large mounds in the central and to have been completely levelled during the northern parts of Area B had been cut into by 19th century conversion of the site to grazing numerous pits, of varying form and content, land, although their positions may be deduced some laid out in lines. The investigation of from the empty spaces between blocks of one approximately east-west line of pits, at ditches, as for instance towards the south-west the western end of the northernmost mound, of Area B (Fig. 2). showed that all were subcircular, between The structure of one mound (3010) in 0.9 m and 1.2 m wide, most having similar the centre of ditch block 4011 was revealed profiles with steep sides and a slightly concave in section (Figs 5-6). This mound may have base. All had linings of yellow/orange clay, on comprised two phases of construction, since average 50 mm thick, in most cases extending its eastern edge overlay the terminals of cross across the base and up the sides, and they had ditches possibly belonging to a ditch block clearly been used to contain water. Although underlying the bentonite dam. The position of they ranged from 0.2 m to 0.5 m deep, given 20 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Key: I Kerb of redeposited natural sand-clay | Black soil (rich in burnt material) j Red burnt soil

Ditch 266 Complete section

Detail ^^^^^

272 Nj**1 / 272 ^^^^343 |^J|

213 342 V^ 279 r 344 344 - S S —' 7 Pit 278 213 1m a

Fig. 6 Section across mound 3010, Area A the later levelling of the mounds, they may Rectangular clay-walled structure originally have been significantly deeper, and A rectangular structure (2640) with an internal therefore capable of holding a larger volume floor area of over 20 m2 was recorded in the than their recorded depths would suggest. centre of the central mound in Area B, and Some had a thin layer of compact silt on the was the only feature of its kind recorded on the base, but the main fills in each pit comprised site. It was 9.4 m long and 4.9 m wide exter- finely laminated dark grey-black silty clay, with nally, with clay walls up to lm thick surviving occasional inclusions of burnt clay. to a height of 0.3 m. It was filled with layers of These clay-lined pits, possibly used for material probably dumped after the feature holding brine, clearly contrast with the groups had gone out of use. of undated and unlined pits sited off the The underlying mound material had been mounds (as described above), whose groupings capped by a 60 mm thick layer of grey clay that had less linear arrangements. extended out beyond the walls of the structure. POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 21

In the section at the west of the feature this was Table 3 Quantification of pottery by period overlain by a thicker layer of clay that appeared (number and weight of sherds) to form a foundation layer for the walls, suggesting that the clay capping may have formed Period Number Weight (g) a watertight floor. In the opposite section, however, there were several intervening layers ?Late Bronze Age 1 15 of apparently dumped material, interpreted as Late Iron Age 298 3323 part of the mound, between the clay capping layer and the walls, suggesting that there was Roman 2 104 no functional relationship between them. The Medieval 26 470 nature of this feature is unclear, therefore, although almost certainly associated in some Totals 327 3912 way with the salt making process. less commonly comprise grog-tempered wares FINDS (11 % by number, 9% by weight), flint-tempered fabrics (9%/5%) and shell-tempered wares Pottery by Grace Perpetua Jones (medieval pottery (<1%). With the exception of the Q3 and Q4 identified by Lorraine Mepham) fabrics, most vessels were probably manufac- tured locally. The flint-tempered sherds include A total of 327 sherds (3912 gr) of prehistoric, one of probable late Bronze Age date, but as a Roman and medieval pottery was recovered. group were not chronologically distinctive. Analysis of fabric and form has followed the The Roman fabrics comprise single sherds standard Wessex Archaeology recording system of Oxfordshire mortarium and New Forest for pottery (Morris 1994). The assemblage parchment ware mortarium. The medieval derives from 52 contexts, although only two fabrics are predominantly coarsewares, (context 7143 in ditch 2812 and context 7286 including south-east Wiltshire/East Dorset in the gully of roundhouse 7) contained more Laverstock-type coarseware (some scratched- than 25 sherds, the minimum number consid- marked), four sherds of sand/flint-tempered ered reliable for an estimation of phase (PCRG and two sandy ware sherds. All fabrics are quan- 1997, 21). The condition of the material is tified by fabric code in Table 4, and detailed in poor to fair, with post-depositional concretions Appendix 1. present on some sherds, particularly those from context 7286. The mean sherd weight is 12 gr. Forms The assemblage has been quantified by period Sixteen rim forms were identified, most repre- in Table 3. sented by single examples, with the exception of the R5, R7, R8 and Rll forms, each with two Fabrics vessels. The late Iron Age fabrics are dominated by Out-turned or everted rim jars sandy wares, which account for 60% of the Iron R2: Jar with high, rounded shoulder and out-turned Age pottery by count and 72% by weight. Sandy rim (Fig. 7.1). ware Q4 is probably a Durotrigian product from R3: Everted rim jar (Fig. 7.2). the Dorset industries of the Wareham-Poole Harbour region (Cunliffe and Brown 1987, Bead-rimmed j ars 319), as may Q3. Fabrics dominated by vesicles, Rl: Bead-rimmed jar (Fig. 7.3). presumably from leached out shell, account for R4: Necked jar with bead-rim (slightly triangular), a further 19% of the number and 14% of the broken at neck. weight. Some of these additionally contained R6: Bead-rimmed jar with high, carinated shoulder flint and quartz inclusions. Fabrics occurring (Fig. 7.4). 22 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Table 4 Quantification of pottery fabrics, by Jars with flattened beaded rim number and weight R12:Jar with internally-bevelled rim and small bead (Fig. 7.8). Fabric Number Weight (g) R16: Bead-rimmed jar, flat-topped and channelled rim, internally expanded (Fig. 7.9).

Prehistoric fabrics Handled jar Fl 26 159 R13: Barrel-shaped jar with slightly out-turned rim and two opposing applied handles (Fig. 7.10). Gl 20 168 G2 13 138 Miscellaneous jars R8: Upright-necked jar with everted rim, broken at Ql 54 219 neck. Q2 34 220 R9: Upright-rimmed jar (Fig. 7.11). R10: Shouldered jar with flat-topped rim. Hint of Q3 10 320 fingertip impression on rim top (Fig. 7.12). Q4 10 137 Two jars with out-turned or everted rims were 69 1499 Q5 recorded, one in a sandy fabric (R2, Q4) and S99 2 2 one grog-tempered example (R3, Gl). Both VI 37 209 are of late Iron Age date, although the R3 may extend into the post-conquest period of the 1st VF1 12 49 century AD. The closest parallels from Heng- VQ1 8 171 istbury Head are the JD4.11 jars. Six vessels are characterised by undifferentiated rims, two are VQ2 447 barrel-shaped (R5) and two are more round- Romano-British wares bodied (R7), however the orientation of two E211 19(Rll) could not be ascertained due to the small size of the sherds. Three are in vesicular fabric E212 1 95 VI, two are made from a sandy ware and one Medieval wares is in a grog-tempered fabric. The form is par- E422 7 168 alleled in the Hengistbury Head series, JC2.0, which were particularly common in vesicular Q400 13 129 and grog-tempered fabrics (Cunliffe and Brown Q401 6 173 1987,208). Four bead-rimmed jars are present, but all rep- Total 327 3912 resenting quite different forms. The Rl is a typical late Iron Age largejar, possibly a storage jar, with a rim diameter of approximately 400 mm. It is par- R14: Small bead-rimmed jar or beaker (Fig. 7.5). alleled by Hengistbury Head formJC 3.1 and had

1 been made from a fabric now characterised, by Undifferentiated rim, from barrel-shaped to round vesicles and quartz. The form of the R4 could not bodiedjars be ascertained as too litde is present, but it has a R5: Undifferentiated rim, probably from barrel- slighdy triangular terminal and is of late Iron Age shaped/ 'baggy' jar (Fig. 7.6). R7: Undifferentiated rim on round-bodied/ovoid date, potentially from either side of the conquest. jar (Fig. 7.7). The R6 has a high, carinated shoulder, with well- Rll: Undifferentiated rim, orientation uncertain. defined bead and point of carination. Both the R15: Possible jar with very slight bead rim, orientation R4 and R6 are in sandy fabrics. The R14 is a small uncertain as too little is present. Alternatively jar or beaker form. may be a lid. The R12 and R16 are both round-bodied POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 23 jars with flattened beaded rims, in Durotrigian and 2732, located in and around roundhouses type fabrics (Q3 and Q4). These are paralleled 6, 7 and 8. at Hengistbury Head by form JC4.1, described Only two sherds of Roman pottery were as a Dorset product, part of the Durotrigian recovered, both from late Roman mortaria. assemblage, dated 50 BC-AD 50 (Cunliffe and The medieval material is of 12th to 13th Brown 1987, 209). Also typical of the Durotri- century date and derives from most of the areas gian ceramics are jars of rounded profile and investigated. Most of the sherds are in sandy/ elongated, lipped rim, sometimes handled flint-tempered coarsewares, although slighdy (Cunliffe and Brown 1987, JD4.41, 50 BC- finer, sandy variants include two jug rims. The AD 50), one example of which was present at coarsewares include Laverstock-type ware from Efford (R13, Fig. 7.10), although the fabric is south-east Wiltshire/East Dorset. not a Durotrigian one. The miscellaneous jars (R8, R9 and R10) List of illustrated late Iron Age sherds (Fig. 7) were too incomplete or undiagnostic to identify parallels, however a hint of a fingertip 1 Round-shoulderedjar with out-turned rim (R2), impression on the top of the flint-tempered Q4 (fabric), PRN (Pottery Record Number) 3, R10 suggests a possible late Bronze Age date. context 2812, ditch 2812 This vessel was recovered from tree-throw 7281 2 Everted rim jar (R3), Gl, PRN 20, context 2961, ditch 2757 which was cut by the gully of roundhouse 7 and 3 Large bead-rimmed jar (Rl), VQ1, PRN 2, therefore this date is entirely possible. context 2812, ditch 2812 4 Bead-rimmed jar with high, carinated shoulder Discussion (R6), Ql, PRN 37, context 7143, ditch 2812 The earliest pottery from the site is a single 5 Small bead-rimmed jar or beaker (R14), G2, flint-tempered rim, probably of late Bronze PRN 81, context 800, gully of roundhouse 3 Age date, from tree-throw 7281, however most 6 Small barrel-shaped jar with undifferentiated of the prehistoric assemblage is of late Iron Age rim (R5), VI, PRN 31, context 7042, gully 2777, date. It derives predominandy, but not exclu- roundhouse 6 sively, from excavation Area C where the Iron 7 Round-bodied/ovoid jar with undifferentiated Age features and buildings were recorded. Diag- rim (R7), Q2, PRN 38, context 7159, gully 7252, nostic forms and fabrics suggest either local roundhouse 9 manufacture or Durotrigian products brought 8 Jar with internally-bevelled rim and small bead (R12), Q3, PRN 62, context 7324, gully of in from the Wareham to' Poole Harbour region roundhouse 9 of Dorset. Typical forms include the jars with 9 Bead-rimmed jar with flat-topped and chan- flattened beaded rims (Efford R12, R16) nelled rim, internally expanded (R16), Q4, which probably derive from middle Iron Age PRN 84, context 808, gully of roundhouse 4 types (Cunliffe and Brown 1987, 319) and the 10 Barrel-shaped jar with slighdy out-turned rim handled jar (Efford R13), although the fabric and two opposing applied handles (R13), Q5, of the Efford example appears to be slighdy PRN 69, context 7286, gully of roundhouse 7 earlier in date, possibly dating from the 2nd to 11 Upright-rimmedjar (R9), VQ1, PRN 42, context 1st century BC. The jar was recovered from the 7196, gully 7253, roundhouse 9 gully terminal of roundhouse 7. 12 Shouldered jar with flat-topped rim and possible fingertip impression (R10), Fl, PRN With the exception of the deposition of 52, context 7282, tree-throw 7281 much of this handled jar in the terminal of the gully of roundhouse 7, the remaining Iron Age potterj' was spread across the site in small Fired clay by Grace Perpetua Jones and groups. Assemblages of 29 sherds or less were Kayt Brown recovered from the gullies of roundhouses 2, 3, 7 and 9. Assemblages of a similar size were A total of 997 fragments of fired clay, weighing present in clay-lined pits 2102, 2999,7301,2798 11761 gr, was recovered from 105 contexts 2 4 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Fig. 7 Iron Age pottery across the site, all but 65 fragments (370 gr) tal samples represents evidence for heating, deriving from features in Area C. Afurther 1562 presumably during salt production, creating gr of burnt loam identified from environmen- light-weight, vesicular nodules. POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 25

Approximately one third of the fired clay represent a small container, perhaps for the assemblage, by weight, could not be identi- transport of the finished product (Fig. 8.3). fied, consisting of amorphous lumps, usually The use of domestic vessels as drying moulds devoid of surfaces or any diagnostic traits. has been proposed at the middle Iron Age Another third consists of fragments from fired site at Mill Hill, Sussex (Bradley 1992, 43) clay slabs. These were usually constructed so perhaps it is not implausible that a similar from poorly wedged fabric FC VI (Appendix practice could have occurred at Efford. Two 1). The upper surface of these objects is other briquetage rims (contexts 7229 and usually buff, orange or yellowish brown, 2923, Fig. 8.5 and 8.4 respectively) are again often with a relatively smooth surface, almost comparable to fragments from Wytch Farm a slurry finish. The underside is more rough (Cleal 1991, fig. 64.4) and Danebury (Poole and less oxidised, usually grey with areas of 1991 fig. 7.68.31). buff or brown. Most are around 35 mm in Briquetage container body fragments were thickness, however no complete slabs were generally flat, approximately 6-8 mm thick, present and the length and width of these with buff or orange/yellow brown surfaces, objects is unknown. They are usually uneven, sometimes with unoxidised interiors, and often appearing slightly warped. One slab, some with pink/grey/white colours associ- from context 800 (gully of roundhouse 3), is ated with salt contact. A number of fragments unique in that 13 circular impressions were from context 7250 could be re-joined (Fig. present on the upper surface, penetrating 8.10), comprising the base of a thin walled it by 5 mm (Fig. 8.11). They had been made vessel with flaring base. Two wedge-shaped prior to firing, probably with a stick. At 115 fragments (context 7321, 7285) may also be mm x 100 mm x 43 mm it was the largest slab associated with salt production (Riehm 1961, present in the assemblage, although three 188). fragments from ditch 2727 (context 7326) The use of triangular loomweights was joined to form a piece 155 mm x 165 mm x indicated by the presence of three fragments, 35 mm. It is unclear whether or not these clay each with a perforation across the corner slabs were used in salt production or utilised measuring approximately 15 mm in diameter in a domestic context. (Fig. 8.7-9). Two other clay objects, of unknown A small assemblage of briquetage, char- function, comprised rounded fragments with acterised by its organic tempering and an incomplete central perforation (contexts discoloured surfaces (resulting from contact 2917 and 7261). with salt), included rim fragments which are often irregular in shape, the size and orienta- List of illustrated fired clay (Fig. 8) tion of which could not easily be ascertained (Fig. 8.1-5). The rims measured between 8 1 Briquetage rim, context 2991, gully 7252, mm and 11 mm in thickness and all occurred roundhouse 9 in organic-tempered fabric FC V2. Two rim 2 Briquetage rim, context 7286, gully of round- sherds, shouldered vessels with upright rims, house 7 both from context 7286, are quite distinct 3 Briquetage rim, context 7286, gully of round- forms; although the late Iron Age assemblage house 7 from Wytch Farm includes illustrated vessels 4 Briquetage rim, context 2923, gully 7252, which hint at possible similar rounded forms roundhouse 9 (Cleal 1991, fig. 64, nos 13 and 18), none 5 Briquetage rim, context 7229, gully of round- are directly comparable. The larger of these house 9 Efford rims (Fig. 8.2) may be a more exagger- 6 Fired clay object, context 2861, ditch 2727 7 Loomweight fragment, context 7286, gully of ated version of a trough form with moulded roundhouse 7 rim as identified at Danebury (Poole 1991 8 Loomweight fragment, context 2969, gully fig. 7.68.31), whilst the smaller version may 7253, roundhouse 9 26 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Fig. 8 Fired clay and briquetage

9 Loomweight fragment, context 7285, gully of 11 Fired clay slab, context 800, gully of round- roundhouse 10 house 3 10 Salt container body, context 7250, pit 7249 POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE

Table 5 Plant remains and charcoal from pit 7313

Context Sample Sample volume (I)

Charred plant remains Alriplex sp. Orache Chenopodiaceae Stetlaria media (L.) Vill. Chickweed Rumex sp. Docks FaUopia convolvulus (L.) A. Love Black bindweed Corylus avellana Hazel nut shell fragment Calluna vulgaris (L.) Hull Heather seed capsule Galium aparine L. Goosegrass/cleavers Rubus/ruticosus L. agg. Bramble/blackberry Eleocharis palustris (L.) Roem. & Schult. Common spikerush Canst sp. 3 sided Sedges Carex sp. 2 sided Sedges Unidentified seed Cyperaceae rhizome Monocotyledon rhizomes Poaceae culm node Culm lengths Roots/stem lengths Charcoal Volume of charcoal >2mm (ml) Cf. Ulmussp. Elm Quercus sp. Oak Quercus sp. Heartwood Oak cf Quercus sp. cf. oak Pomoideae Apple/Pear/whitebeams Prunus spinosa L. Sloe/blackthorn Prunus spinosa L. heartwood Sloe/blackthorn Indet Total no. identified fragments % of assemblage identified by volume TWaterlogged Ranunculus acris/repens/bulbosus L. Buttercup Atriplex sp. Orache Chenopodium album L. Fat hen Rumex acetosella L. agg Sheep's sorrel cf. Rumex sp. Docks Rubusfruticosus L. agg. Bramble/blackberry etc Anlhemis cotula L. Stinking mayweed Polamogeton sp. Pondweed Carex sp. Sedges 28 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

Other finds by Lorraine Mepham and flots were therefore tested for the presence of iron working residues and hammerscale with a Almost all the burnt, unworked flint (over 99 magnet. The bulk samples were supplemented by kg) came from Area C, where the associated two monoliths, two soil samples and one kubiena pottery suggests a late Iron Age date. The bulk of taken for the analysis of sediments. it came from a single feature - 76 kg from ditch 2727, and only two other contexts produced Methods more than 1 kg (7143 in ditch 2812 and 2998 in pear-shaped feature 2997). Its association The bulk samples were processed by standard with briquetage in ditch 2727 and feature 2997 flotation methods; the flot retained on a 0.5/ suggests it was used in salt production. 0.25 mm mesh and the residues fractionated Other material types were represented in into 5.6 mm, 2 mm and 1 mm and 0.5 mm much smaller quantities. They include seven fractions and dried. The coarse fractions (>5.6 pieces of undiagnostic worked flint, a Roman mm) were sorted, weighed and discarded. The silver coin (republican denarius, unstratified); flots were scanned under a xlO - x30 stereo- 13 fragments (3 gr) of unidentifiable animal binocular microscope and presence of charred bone, a post-medieval iron horseshoe, three remains quantified (assessment table, in fragments of lead (unstratified) and modern archive), to record the preservation and nature brick. of the charred plant and charcoal remains. In the majority of samples charred plant remains other than charcoal were found to be limited ENVIRONMENTAL EVIDENCE and sufficient detail was recorded during the By Ruth Pelling with Chris J. Stevens assessment to characterise the flots, forming the basis of the following discussion. The changing pattern of settlement and indus- Two samples from pit 7313 (initially inter- trial activity at Efford must be closely related to preted as of possible late Iron Age date) were the local patterns of coastal change, with the selected for detailed analysis of charcoal and expansion of salt marshes as the coastline moved any other macrofossils present (Table 5). southwards. Likewise the long tradition of salt This feature was within one of the groups of making, which became increasingly industr- otherwise undated pits within the later ditch ialised over time, may have had a significant system in Area B, which were problematic impact on the local vegetation, particularly in in terms of their interpretation. All charcoal relation to fuel usage. The aim of the sampling fragments retained in the 2 mm and larger was to attempt to retrieve evidence for the arable sieves were extracted from the flots. Up to 100 economy at the site as well as other evidence fragments of charcoal were then selected from for plant use and in particular changing fuel each sample for identification. The extracted patterns over time. Changes in fuel use may be charcoal fragments were prepared for identifi- related to changing industrial or semi-indus- cation according to the standard methodology trial activities or may be related to changing of Leney and Casteel (1975, see also Gale and vegetation and fuel availability. This evidence Cutler 2000). Identification was undertaken was supplemented by sediment descriptions. according to the anatomical characteristics Environmental samples were taken from a described by Schweingruber (1990). Identi- range of feature types and processed for the fication was to the highest taxonomic level recovery and assessment of charred plant remains possible, usually that of genus, and nomencla- and charcoal. These included 40 bulk samples ture is according to Stace (1997). of generally 10 litres; one sample of 140 litres from ditch 2727 was artefact sieved (4 mm) for Results charcoal. Siliceous material was noted in all the samples suggesting some sort of industrial residue, The flots were generally small to moderate in POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 29 size (up to 60 ml from a 10 litre sample). Many from context 2861 produced some 800 mis of consisted almost entirely of rooty material (80- material sieved through a 4 mm mesh from 140 90%) and low numbers of weed seeds which litres. were not charred and may therefore be indicative of stratigraphic movement. Silicaceous Pit 7313 material and small lumps of burnt fine grained Two samples from pit 7313 (contexts 7311 and mineral material were present in all samples. 7312), which was initially considered to be of Evidence for cereals was recovered from possible Iron Age date, were examined in more two Iron Age samples: pit 7301 (pit group 2) detail (Table 5). Each flot was generated from produced a single glume base of probable 10 litres of deposit, producing 500 ml and 350 emmer wheat (Triticum dicoccum), and the gully ml of material respectively. The flots consisted of roundhouse 2 (section 7283, context 7286) of charcoal, stem or root material and large produced a single rachis of barley (Hordeum quantities of both silicaceous material and vulgare). A small number of weed seeds was also lumps of apparently burnt, fine sediment recovered from the pit, including sedges (Carex with occasional larger sand inclusions. The sp.), common spikerush (Eleocharis palustris charcoal in both samples was dominated by type) and knotgrass (Polygonum aviculare). Quercus sp. (oak) including fragments contain- Charred cereal remains are most frequently ing tyloses indicative of heartwood. Fragments a product of the regular processing of cereals of Pomoideae (apple/pear/whitebeam etc) as they are taken from storage and the waste and Prunus spinosa (blackthorn/sloe) were also thrown upon hearths or fires. The absence of present and a single fragment tentatively iden- such remains probably indicates that the inhab- tified as Ulmus sp. (elm). This range of mixed itants of the site did not process grain upon tree taxa is likely to reflect local mixed vege- the site or that any cereals brought to the site tation rather than any deliberate targeting of may have arrived already processed as flour, species. bread or beer. Alternatively, processing waste Of more interest was a number of convoluted may have been used for other purposes, or the roots or stems, and the rhizomes and tubers of occupation of the site may have been of such monocotyledons such as grasses or sedges. While short duration that few remains were produced the species of these items were not determined, or survived. they are indicative of uprooted grasses and Several of the samples did produce remains shrubs. Two seed capsules identified as Calluna of grass or sedge stems, as well as a few seeds, vulgaris (heather) in context 7312 provide some mainly of wetland species such as sedge (Carex indication of possible species for the woody root sp.), spikerush (Eleocharis palustris) and blinks and stem material. This context also produced (Montia fontana ssp. chondrosperma). Other a number of trigonus seeds of Carex sp. (sedges) species were also occasionally present such as including two preserved as silica skeletons, as knotgrass (Polygonum aviculare), black bindweed well as seeds of Eleocharis palustris (common (Fallopia convolvulus), cleavers (Galium aparine) spikerush), Rubus Jruticosus (bramble/black- and docks (Rumexsp.). These were commonest berry) , Rumex sp. (docks), Fallopia convolvulus from the samples in pit 7313 (see below). (black bindweed), Atriplex sp. (orache) and While generally scarce, charcoal occurred indeterminate Chenopodiaceae. in high quantities in occasional samples. This assemblage also includes seeds of wet Round wood and stems were present in ground or marshy conditions (Carex sp. and several samples, for example possible Iron Eleocharis) as well as species more characteris- Age ditch 2812, undated pit 7313 and two tic of disturbed ruderal habitats. A fragment of probable post-medieval pits (2522 and 2524) Corylus avellana (hazel) nut shell was present in the group south of rectangular structure in context 7311. A small number of possible 2640. The samples from ditch 2727 conversely waterlogged seeds present in context 7312 contained charcoal but little round wood; that included a range of possible wet or marshy 30 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY ground species such as Ranunculus acris/repens/ Murphy 1995) the association of the thicker bulbosus (buttercup) and Carex sp. (sedges), or roots or stems and amorphous burnt and sili- species of disturbed ground such as Atripkx so. caceaous material is more unusual. Similar (orache), Rumex acetosella (sheep's sorrel) and deposits were recovered from excavations on Rubus fruticosus (bramble). A seed of Potamoge- the Isle of Barra (Wessex Archaeology 2008) ton (pondweed) suggests open water. consisting of burnt soil, stems and seeds of Context 7312 also contained four seeds sedges which were interpreted as burnt turves. of Anthemis cotula (stinking mayweed) which The mix of fuel types (wood charcoal and peat/ are slightly anomalous within the assemblage heather) may be indicative of secondary refuse in that this species is closely associated with rather than primary fuel, or it may represent a arable fields and cereal cultivation of the low grade industrial process involving whatever Romano-British and medieval periods. While fuel was easily available. its occurrence in a late Iron Age context is The paucity of charred remains and charcoal possible this would be more likely to occur probably reflects the nature of activity on the on a site with better evidence for arable agri- site rather than poor preservation. On a salt culture. Given the absence of Romano-British working site of this kind, the dark colours in deposits on this site the presence of this species the samples may indicate the use of materials therefore hints at a medieval rather than Iron other than wood for heating seawater. For Age date for this feature. instance, peat would leave few large fragments, and coal (imported for the post-medieval Conclusions saltworks) would not be recorded in the flots. The 'slag'-like concretions recorded during the Despite the archaeological evidence for assessment (in archive), which are the same as Iron Age activity at the site, the bulk samples the burnt amorphous material present in pit produced little evidence for cereal remains 7131, are a by-product of heating and burning beyond rare individual items of glume bases, but do not represent coal or the burning of rachis or arable weed seeds. The fact that these peat, or industrial processing of metals (i.e. remains are present at all suggests some cereal- slag). Nor has it been determined whether related activity although clearly on a small scale, they represent the burning of elements with cereals possibly brought onto the site as occurring naturally in the local highly oxidised flour or even bread. This may reflect the pre- and gleyed soil, in the soils uprooted with the dominantly industrial, rather than domestic, heath, or from salt making. nature of the site. The charred remains from the pit 7313 are interpreted as being related to fuel use rather DISCUSSION than burnt domestic refuse, including charcoal but also possibly peat or heather burning Salt making suggested by seed capsules of heather. If mature heather was uprooted rather than cut it is likely British coastal seawater consists of c. 3.5% by that it would include sedges and/or grasses or weight of salt, and in order to extract that salt other heathland species growing around it. The the water has to be evaporated off. The history roots and thick stem fragments have not been of coastal salt making in Britain, from prehis- positively identified but are thought to include tory through to the modern period, reveals more than one species, which may include the different means by which salt was extracted heather (Calluna vulgaris). The silicaeous from sea water with ever greater efficiency, by material and burnt amorphous material may harnessing a combination of natural and indus- represent silty peat deposits uprooted with the trial processes. heather. While heather is relatively frequendy Two main methods have been used to con- identified from archaeological sites (e.g. centrate brine (Gilman 1998). One is to 'wash' POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 31 the salt from sand (and/or silt) collected from ing stands of grasses and reeds and bisected the shore; the salt becomes impregnated in the by deeper flowing channels - an environ- sand during the high tides, and dried and con- ment similar to that in the present Keyhaven centrated by the sun and wind. This method Harbour. However, its low-lying and exposed of sand washing, sometimes referred to in the position is likely to have made it unsuitable for medieval period as 'sleeching' or 'muldefang', year-round settlement and it is possible that it involved skimming off a thin layer of sand, and was occupied only on a seasonal basis, as part trickling water through it in a trench filled with of a cycle of agricultural and other economic turfs, straw or reeds to act as a filter (Sturman activities. Evidence for seasonal coastal settle- 1984). The filter retains the sand, while the ment in the Iron Age has been found on the brine passes through and is collected. A charac- Avon Levels in Somerset, based on the tran- teristic feature of this process are large mounds shumant exploitation of coastal grazing, and consisting mainly of the discarded, washed characterised in part by the repeated renova- sand. tion of roundhouses (Gardiner et al. 2002). The other method of producing brine is Certainly the summer months would have to use solar evaporation in 'salt pans', some offered the best time for both making salt and times called 'sun works', a method described for grazing livestock on the coastal marshes by Georgius Agricola in the 16th century (Bradley 1975), and there is considerable and represented in rather idealised illustra- evidence for Iron Age salt making along the tions (Agricola 1556). Seawater, collected in Hampshire coast, such as in Langstone and ponds at the spring tides, is fed into a series of Chichester Harbours (Fox 1933; Bradley 1992; wide shallow basins in which the salt content Allen and Gardiner 2000). gradually increases as the water evaporates. Although the roundhouses on the site were Archaeological evidence for salterns using close together, none of them overlapped and it this method may include the remains of the is possible that they represent a single substan- holding ponds, the ditches or channels that tial contemporary settlement, with evidence of fed the seawater to the pans, and the pans the rebuilding indicating their reoccupation themselves, although their shallow construc- over a number of years. Whether they were tions makes the latter vulnerable to subsequent all domestic structures is unclear, however; ploughing. While such methods were common while some had entrances facing between east in warmer climates from antiquity, there is no and south, two (roundhouses 2 and 10) had clear evidence that they were used in Britain opposed entrances, and roundhouse 5 faced until the post-medieval period. north-west - unusual for a domestic structure. Once the brine had been produced, by Despite the density of roundhouses, however, whatever method, the concentrate needed to the site produced very low levels of domestic be further condensed and dried in vessels over refuse, the finds consisting mainly of pottery, hearths, fuelled by wood or peat, and latterly either locally produced or imported from the coal. This was a process common to both pre- Wareham to Poole Harbour region of Dorset, historic and historic salt production, indicated fired clay and burnt flint. While the presence by the evidence of burning in the form of of a number of loomweight fragments within charcoal and burnt soil, although the nature of the fired clay assemblage points to a domestic the vessels and the form of the hearths changed craft activity of the type normally associated over time. with Iron Age settlements, the almost total absence of animal bones (one context) and Iron Age crop remains (two contexts) makes it difficult to determine the wider economic basis of the The Iron Age settlement occupied relatively setdement. dry land on the edge of the salt marsh which As discussed above, there is doubt as to the consisted of fine silts and mud support- identification of clay-lined pits in the Iron Age 32 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY settlement. Even if some were genuinely clay- logical Services 2001), while to the north-east lined, their possible role within the salt making at Exbury there was a promontory fort on the process is unclear, although they may have Beaulieu River (Sumner 1917, 119). The uni- been used for holding water, possibly brine. vallate fort at Tournerbury on Hayling Island There are no indications, however, as to how is in a similar location (Bradley and Fulford brine might have been concentrated. Much 1975), and although such fortified sites are of the fired clay from the site was fragmentary relatively rare in lowland areas they may have in nature, although the identification of some been established to control the exploitation of pieces from flattish briquetage blocks/slabs the rich coastal and estuarine resources in the and rods indicates that water was boiled off area, including salt. the brine in fired clay containers. However, no hearths, either domestic or salt making, were Medieval and post-inedieval identified, and while the burnt flint may have been a product of the brine boiling process, There are references in the Domesday Book only four contexts produced more than 1 kg; of 1086 to salterns at 12 manors in Hampshire, the 76 kg recovered from the upper fill of ditch including six at Hordle, just west of Lymington. 2727 was clearly redeposited. The production of The earliest reference to salt making at salt, therefore, may have been undertaken away Lymington, however, concerned the grant of a from the settlement focus with distinct areas of tithe of Lymington salt to Quarr Abbey on the industrial and setdement activity shifting over Isle of Wight in 1132, which fits well with the time. Although the pits were found close to the 12th-13th century date range for the pottery roundhouses they need not have been direcdy from the site. It is suggested that the Lymington associated with them, the different pit groups industry may already have been well established being found both inside, outside and cut by the by that date (Lloyd 1967). roundhouse gullies. Such early references, however, give few Late Iron Age salt making at Lymington clues as the method of salt production in the is likely to have been combined with other medieval period, although there are indi- economic and agricultural activities, the cations from both Sussex and Dorset (and number of roundhouses alone suggesting further afield) that sand washing (as described that the seasonal occupation of this coastal above) was the standard method (Holden and landscape for possibly varied activities was one Hudson 1981, 123; Keen 1987). Moreover, that made sound economic sense. Salt making early maps of Hampshire, such as Norden's was a potentially lucrative industry, and as bri- map of 1595, show the Lymington coastline quetage is found on sites far inland from the dotted with substantial mounds, which may be coast, the identification of briquetage vessels a representation of the characteristic feature of used for transporting salt may indicate this this method (Fig. 9). Although Rudkin (1975, site's involvement in long distant exchange 39) refers to sun works having been used at and trade. The recovery, albeit unstratified, of Lymington since the 13th century, references a Roman republican silver denarius is a hint at to medieval 'salt-pans' (the normal transla- the potential wealth that this trade may have tion of medieval Latin salina) may denote salt generated. works in general, rather than specifically the A likely social context for such activity is the use of floor pans. (Some confusion can be multivallate fort at Buckland Rings 4km north caused by the fact that the term 'pan' is used, to the north, which dates from the 4th century variously, to refer not only to salt works in BC to the 1st century AD (Hawkes 1936). general, but also, more specifically, to salterns There was another, smaller earthwork, also using floor pans to concentrate the brine; it possibly Iron Age, at Ampress Hole, just east of is also used to denote the individual shallow the fort beside the junction of Passford Water ponds within such salterns, as well as the lead, and Lymington River (Thames Valley Archaeo- and later iron, vessels used for boiling the POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 33 P^^M^i '

Fig. 9 Detail of John Norden's 1595 map of Hampshire (reproduced with permission of Hampshire County Council Museum Service)

brine). Although a tentative medieval date has salterns, that of 1698 by Dummer and Wiltshaw been suggested for pit 7313 on the basis of the (Lloyd 1967, plate 4), for example, showing an environmental evidence, the function of such array of large rectangular 'salt panns' stretching features is unclear. along the shore west from the Lymington River, Following a decline in the 14th century, salt interspersed with wind pumps and buildings - production at Lymington was boosted at the probably boiling houses. beginning of the 17th century by the granting The central questions to be answered of a patent by James I 'for the new invention about the features identified in Areas A and of making white and bay salt', and in 1625 all B, therefore, are: what was their function and the mudlands on the coast at Lymington were where do they fit within the development of the granted to the salt trade. However, these new Lymington salt industry? The medieval pottery methods required major structural alterations (the only datable finds from these features) (VCHH, 470), and there were protests by 14 and the presence of low earth mounds might Lymington saltern owners against an order suggest an early date, but the grid-like arrange- requiring them to replace their mounds with ment of ditches within larger blocks appear floor pans. This, too, implies that sand washing, closer in form to the descriptions of the 17th a process resulting in the creation of mounds, century and later salt works. However, there had been the main method employed up to have been no excavations of this type of site that date. Maps dating to the end of the 17th within Hampshire, and none of the few saltern century, however, indicate the new form of the sites excavated elsewhere in the country display 34 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY this layout (e.g. Hall and Coles 1994; Healey by weakening ye salt. When they think its fit 1975; Holden and Hudson 1981; Barford 1988, to boyle they draw off water from ye ponds by McAvoy 1994; Gilman 1998; Ridgeway 2000). pipes ... '. A 1784 drawing by Thomas Row- In 1879, Edward King reported that the landson shows a regular patchwork of square marshes, by degrees, had been more or less floor pans, each bounded by low banks, the levelled and the salt making ponds filled up, wider area dotted with the windmills (Wark so as to render land useful for grazing (King 1963). 1879), and the 1846 Sheringham chart shows Unfortunately, the cartographic evidence the whole site, and the land between it and for the extent of the post-medieval salterns the shore as rough grazing land drained by a at Lymington, which has a bearing on the wide ditch running from the north of the site date of excavated features, is ambiguous. A to the sea. However, as any early sand washing 1781 chart by Lt. Murdoch MacKenzie places mounds may already have been largely levelled most of the site outside (northwest of) the as a requirement of the early 17th century devel- salterns, although the precise location of the opments, the low mounds recorded during the salterns' landward boundary would not have excavation may be the remains not of medieval been important in a naval chart focussing on mounds but of raised areas (possibly reusing the navigation of the Solent. In contrast, the material from the earlier mounds) constructed 1st Series Ordnance Survey map of 1811 (Fig. for the siting either of the hearths used for 10), which shows the salterns as a regular grid boiling the brine, or of the wind pumps used pattern that appears to cover almost all of the for pumping the brine to the boiling houses. site, contains clear inaccuracies and surveying The windmill mounds are reported as being 30 errors. feet across (c. 9 m) across and 5 feet high (c. There are a number of contemporary descrip- 1.5 m); the windmills were demolished in 1873 tions of the final phase of the Lymington salt (Lloyd 1967, 97). works. In 1813, when there were 68 'pans' in The bank or kerb underlying the mound in the Lymington and Pennington salterns (Lloyd Area A, which was built (or at least extended) 1965), Charles St. Barbe, the principal owner after a period of use of one of the ditch blocks, of the Lymington salterns, wrote 'The sea may have been designed to keep water out of water is first admitted into feeding ponds, from a central dry area, and the thick layer of red whence it flows into levels, in which there are burnt soil inside the bank, in places burnt in partitions, forming pans, as they are called .... situ, indicates that this space was probably the these receive the sea water from the feeding site of hearths. The overlying mound material, ponds to a depth of about 3 inches (c. 8 cm), comprising a thick layer of black sandy silt and from which it passes from the higher to the containing charcoal, ash and burnt clay, may lower pans, exposed to the action of the sun represent a mix of hearth debris and material and the wind, until the brine becomes of a suf- intermittently cleared out of the surround- ficient strength to be pumped up by small wind ing ditches, with whose fills it was similar in engines into a cistern, whence it is conveyed by appearance. troughs into respective iron pans for boiling' A detailed account of the Lymington salterns (Vancouver 1813,420). around 1700 was provided by Celia Fiennes in Further details were provided by Loudon Through England on a Side Saddle in the Time (1825, 562-3): 'At Lymington, in Hampshire, of William and Mary (Morris 1949), in which sea water is evaporated to one sixth of the she described how 'Ye seawater they draw into whole by the action of sun and air. The works the trenches and so into severall ponds, yt are in which the sea water is heightened into brine secured in ye bottom to retain it and it standeth are called sun-works or out-works. These are for ye sun to exhale ye watry fresh part of it constructed on a flat down or oozy beach, and if it prove a dry summer they make the within a mole which is raised if necessary to best and most salt for ye raine spoyles ye ponds keep out the sea; there is a large reservoir, or POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 35

mm Fig. 10 Detail of the 1811 Ordnance Survey Old Series map of Hampshire (reproduced with permission of Hampshire County Council Museum Service) feeding pond, communicating with the sea by medieval date appears to be at odds with a sluice, and adjoining to this reservoir a long the (admittedly small) finds assemblage, as trench parallel to which there are several square it requires the 26 sherds of medieval pottery ponds, called brine pots, in which the water is all to be residual, deriving from an earlier evaporated to a strong brine, and afterwards it phase of salt working of which no clear undergoes and artificial evaporation and puri- structural evidence was found. In turn, the fication in boilers'. interpretation offered would be stronger had From these varied sources of information post-medieval finds been recovered from the the most likely interpretation of the features site, but their absence could be explained if recorded in Areas A and B is that they represent the main foci of activity - the boiling houses 17th century or later floor pans and ditches, (of which, again, no structural evidence was and associated low mounds for the siting of found) - were located outside the bounda- hearths and/or windmills. However, none of ries of the site. the above descriptions, or other studies of the Each ditch block appears to represent a single salterns at Lymington (e.g. Ravenscroft 1914; saltern unit, possibly bounded by an earthen Cross 1965), exactly match the arrangements bank. St. Barbe recorded that 'the extent of of ditches uncovered on this site. Apart from ground required for evaporation, exclusive of the obvious function of feeding seawater into the feeding ponds and cistern, is about 3 roods, the salterns, their precise function remains or 120 perches to each pan'. This is equivalent unclear and the following interpretation is to an area c. 55 m square, consistent with the necessarily tentative. Furthermore, a post- sizes of the ditch blocks (as described above) of 36 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

50-80 m square. At spring tides, seawater would ately laid as part of the saltern construction. have passed through sluices to fill the reservoirs The excavation records describe the clay or feeding ponds behind the sea wall. Such layer (213) below the mound in Area A as ponds are clearly shown on the 1781 Mackenzie possibly 'deliberately spread', while the clay chart flanking the shoreline, the nearest to the layer recorded below the central mound in site being some 150 m to the south. From there Area B (2518) was also described as 'laid' and the sea water was fed by trenches into the feeder compared to the clay used to line the clay- ditches which in turn fed the cross ditches in lined pits; the same context record also refers each saltern. to probable repairs to the clay layer with a No internal earthworks associated with the slightly grey clay. cross ditches were noted within each saltern. However, the 1879 edition of 'Old Times Conclusions Revisited' says that salterns were 'divided into shallow ponds about 20 feet (c. 6 m) square by The programme of fieldwork has added low mud banks about 6 inches (15 cm) high, valuable new information about salt working just wide enough for a man to walk upon with on the Lymington coastline, but leaves many caution'. It is possible that such features did important questions unanswered. While the not survive the 19th century levelling of the Iron Age remains provided clear circumstan- salterns or were not archaeologically visible. tial evidence of salt working, they offered An 1828 account (quoted by Lloyd 1967, 97) little insight into the actual operation of the describes such ponds (referred to as 'pans') industrial process. The presence, however, of a as measuring 25-120 square yards (5-10 m substantial, if seasonal, settlement, capable of square), therefore potentially lying between exploiting the varied economic resources of the the cross ditches, and says that the liquor coastal marsh and saltmarsh environment, of was moved six times from pan to pan. If this which salt was probably an important element, was this case, then each saltern may have helps provide an economic context for the comprised in the region of 50 interconnected hillfort at Buckland Rings and other possibly ponds, with the increasingly concentrated late Iron Age enclosures along the coast and brine possibly being fed inwards towards the estuaries. mounds from where it was pumped to the The complex of features associated with the boiling hearths. A 1693 map of Lymington by later saltworks are also, in some respects, difficult Captain Greenville Collins shows what may to interpret. The interpretation offered above be a stylised representation of such a saltern, requires, for instance, that the few datable finds the 'Salt Pitts' comprising an approximately (medieval pottery) were residual, and that the square block of six parallel ditches, with per- later activity left no artefactual remains (or at pendicular 'cross-bars' between them creating least no archaeologically recovered finds) within smaller divisions. the features excavated. However, in contrast to According to Celia Fiennes (Morris 1949), the little that we know about medieval coastal the ponds were lined with clay in order to salt making in Britain, there are clear formal retain the brine: 'They are very careful to similarities between the features uncovered keep their ponds well secured and mended and the descriptions, illustrations and carto- by good clay and gravel in the bottom and graphic representations of the post-medieval sides'. The traces of a clay layer recorded and later sun works at Lymington. While there extensively across the site, sealing the pre- would seem to be little doubt that the arrays saltern features, were initially interpreted as a of ditches were used in some way to feed floor flood deposit, and indeed there is a record of pans, the precise operation of those floor pans a great storm in 1703 in which some salterns is not clear, as no internal earth ridges were were flooded (Lloyd 1967, 90). However, it noted that would separate individual ponds. is possible that this clay layer was deliber- Moreover, as the descriptions refer to the brine POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 37 being passed through a series of ponds, which of ditches spaced 5-7 m apart (Stone 2003, could in theory be fed from the outer, feeder fig. 9), significantly different to the salterns ditch, the purpose of the cross ditches is also recorded on this site. not clear. It is clear that many questions raised by The reported levelling of the salterns in the this programme of fieldwork- about the 19th century to create coastal pastures clearly layout, development and operation of the impacted on the upstanding earthworks, as well Lymington salterns through the medieval and as features cut into them. While the clay-lined post-medieval periods, as well as in the Iron pits may have been used to hold brine, there Age - remain to be answered. However, there is was no direct evidence for the latter part of a high potential for further targeted investiga- the salt making process - converting the brine tion of this important archaeological resource to salt. There was clear evidence of burning to help resolve these issues. associated with salt making, but no hearths were identified; two unstratified fragments of lead were the only possible evidence of the ACKNOWLEDGEMENTS boiling pans. Apart from the large, rectangular, clay-walled feature of uncertain function, no The work was funded joindy by New Milton buildings, wind pumps or other structures were Sand and Ballast, Onyx (Hampshire Waste recorded. Services) and Hampshire County Council. Many of the descriptions of the Lymington Wessex Archaeology is grateful to Dave salterns date from late in their history, and it is Carswell of New Milton Sand and Ballast, Dave possible that the precise operation of such sun Andrews of Onyx and Ian Wykes of Hampshire works may have changed significantly from the County Council. David Hopkins (Hampshire time of their initial introduction. The documen- County Council) monitored the excavations tary references to the replacement of mounds and is thanked for his support during the with floor pans suggests that they were first project and for helpful comments on this constructed in the early 17th century, although report. The excavations were supervised by the possibility of an earlier, even medieval date, Mark Dunkley, Paul Gajos, Steve Webster and cannot be entirely ruled out. Jamie Wright. The fieldwork was managed for As the saltmarsh and mudflats were Wessex Archaeology by Anthony Firth and reclaimed for salt production, the earliest Steve Webster. Philippa Bradley managed the floor pans may have ended up being sited post-excavation analysis and edited this report. furthest from the shore, and therefore the Grace Perpetua Jones undertook preliminary earliest to be turned back into grazing land, post-excavation analysis. The specialists are and this site, although not far from the 18th thanked for their contributions to this report. century sea wall, lies on the north-western Patrice de Rijk commented on the burnt soil (inland) edge of the salterns as they are and slag-like material. Steve Webster, Julie depicted both on the 1781 MacKenzie chart Gardiner, Karen Walker and Lorraine Mepham and the 1811 OS map (Fig. 10). The bund are thanked for commenting on the draft which defines the southern edge of the site report. Wessex Archaeology is grateful to Gill may mark a boundary between an area of Arnott of Hampshire County Council Museum early floor pans, and those closer to the coast Service for permission to reproduce parts of which may correspond closer to the late post- the John Norden 1595 map of Hampshire medieval descriptions. It is certainly evident and the 1811 Ordnance Survey Old Series from the remains of adjacent salterns, visible map of Hampshire. The archive is currently in aerial photographs, that there is consider- stored at Wessex Archaeology under project able variation in their layout, and a survey of a codes 47074-8, 50241-2, 51680, 51880, 52235, small area near Iley Dock in Keyhaven Marsh, 53257-8 but will be deposited with Hampshire for example, revealed a regular grid pattern Museum Service in due course. 38 HAMPSHIRE FIELD CLUB AND ARCHAEOLOGICAL SOCIETY

REFERENCES

Secondary Sources Geological Survey of Great Britain (England and Wales) 1975 Lymington, 1:50,000, Drift Agricola, G 1556 DeReMetallica, Basle (trans Hoover, Sheet 330. H& Hoover, LH 1912). Gilman.P (ed.) 1998 Monuments Protection Programme: Allen, M J & Gardiner, J 2000 Our Changing Coast: The Salt Industry, English Heritage Con- A Survey of the Intertidal Archaeology of sultation Report. Langstone Harbour, Hampshire (CBA Hall, D & Coles, J 1994 Fenland Survey: An Essay in Research Report 124), York. Landscape and Persistence, London. Barford, P M 1988 After the red hills: salt making in Hawkes, C F C 1936 The excavations at Buckland late Roman, Saxon and medieval Essex, Rings, Lymington 1935, Proc Hampshire Colchester Archaeological Group Bulletin 31 Fid Club 13 124-64. 3-S. Healey, H 1975 A Medieval Salt-Making Site in Bicker Bradley, R1975 Salt and settlement in the Hampshire Haven, Lincolnshire, in de Brisay & Evans Sussex borderland, in de Brisay & Evans (eds), 1975,36. (eds), 1975, 20-5. Holden, E W & Hudson, T P 1981 Salt-making in the Brisay, K W de & Evans, K A (eds), Salt: The Study of Adur Valley, Sussex Archaeol Collect 119 an Ancient Industry, Colchester Archaeo- 117-48. logical Group. Keen, L 1987 Medieval salt-working in Dorset, Bradley, R 1992 Roman salt production in Chichester Proc Dorset Natur Hist Archaeol Soc 109 harbour: rescue excavations at Chidham, 25-8. West Sussex, Britannia 23 27—44. King, E 1879 Old Times Revisited in the Borough and Bradley, R & Fulford, M 1975 Excavations at Tourn- Parish of Lymington Hants, Lymington. erbury, Hayling Island, 1959 and 1971, Leney, L & Casteel, R W 1975 Simplified procedure Proc Hampshire Fid Club Archaeol Soc 32 for examining charcoal specimens for 63-70. identification, f Archaeol Sci 2 153-9. Cleal, R 1991 Briquetage, in Cox, P W & Hearne, Lloyd, A T 1965 The Salterns of Lymington, Milford & C M Redeemed from the Heath, the Archae- Hordle, Christchurch. ology of the Wytch Farm Oilfield (1987-90) Lloyd, A T 1967 The salterns of the Lymington area, (Dorset Natural History and Archaeo- Proc Hampshire Fid Club Archaeol Soc 24 logical Society Monograph Series 9), 86-102. Dorchester, 144—149. Loudon, J C 1825 An Encyclopaedia of Agriculture, Cross, D A E 1965 The Salt industry of Lymington,/ London. Ind Archaeol 2 86-90. McAvoy, F1994 Marine salt extraction: the excavation Cunliffe, B & Brown, L 1987 The later prehistoric of salterns at Wainfleet St Mary, Lincoln- and Roman pottery, in Cunliffe, B shire, Medieval Archaeol 38 134-63. Hengistbury Head, Dorset, Volume 1: Momber, G, Rackley, A & Draper, S 1994 New The Prehistoric and Roman Settlement, Forest Coastal Arckaeological Resource, 3500BC-AD500 (Oxford University Hampshire & Wight Trust for Maritime Committee for Archaeology monogr Archaeology. 13), Oxford, 205-321. Morris, C (ed) 1949 The fourneys of Celia Fiennes, Fox, C F1933 Salt works at Hook, Warsash, Hants, Proc London. Hampshire Fid Club Archaeol Soc 13 105-9. Morris, E L 1994 The Analysis of Pottery, Wessex Gale, R & Cutler, D 2000 Plants in Archaeology, Archaeology Guideline 4, Salisbury. Westbury and Royal Botanic Gardens, Murphy, P 1995 Plant macrofossils, in Andrews, P Kew. Excavations at Redcastle, Furze, Thetford, Gardiner, J, Allen, M J, Hamilton-Dyer, S, Laidlaw, 1988-9 (East Anglian Archaeology M & Scaife, R G 2002 Making the most Report 72), 131-5. of it: late prehistoric pastoralism in the PCRG, 1997 The Study of Later Prehistoric Pottery: Avon Levels, Severn Estuary, Proc Prehist General Policies and Guidelines for Analysis Soc 68 1-39. and Publication (Oxford: Prehistoric POWELL: TWO THOUSAND YEARS OF SALT-MAKING AT LYMINGTON, HAMPSHIRE 39

Ceramics Research Group Occ Papers 1 logical Investigation, Reading, unpubl. & 2), 2nd ed., Oxford. report. Poole, C 1991 Briquetage containers, in Cunliffe B Vancouver, C 1813 General View of the Agriculture in & Poole, C Danebury An Iron Age Hillfort Hampshire, including the Isle of Wight, in Hampshire Vol 5 The Excavations London. 1979-1988: the finds (CBA Res Rep 73), VCHH, The Victoria History of the Counties of England: 404-407. Hampshire, vol V, London, 1912. Ravenscroft, W 1914 The old Lymington salterns, Wark, R R (ed.) 1963 Rowlandson's Drawing for a Proc Hampshire Fid Club Archaeol Soc 7 Tour in a Post Chaise, San Marino CA, 81-5. Huntingdon Library. Ridgeway, V 2000 A medieval saltern mound at Wessex Archaeology, 1999 Efford Landfill, Penning- Millfields Caravan Park, Bramber, ton: Earthwork Survey and Watching West Sussex, Sussex Archaeol Collect 138 Brief, Salisbury, unpubl. client report, 135-52. 47074.01 Riehm, K 1961 Prehistoric salt-boiling, Antiquity 139 Wessex Archaeology, 2000 Efford Landfill, Penning- 181-191. ton: Earthwork Survey Plans and Sections, Rudkin, E H 1975 Medieval salt making in Lincoln- Salisbury, unpubl. client report, shire, in de Brisay & Evans (eds), 1975, 47075.02 37-40. Wessex Archaeology, 2008 AUasdale Dunes, Barra, Schweingruber, F H 1990 Microscopic Wood Anatomy, Western Isles, Scotland: Archaeological Eval- 3rd ed., Birmensdorf, Swiss Federal uation and Assessment of Results, Salisbury, Institute for Forest, Snow and Landscape unpubl. client report, 65305.01. Research. Soil Survey of England and Wales, 1983 Soils of South Historic Maps East England, Sheet 6,1:250,000. Stace, C 1997 New Flora of the British Isles, 2nd ed., 1595, John Norden's map of Hampshire, Henry Cambridge. Overton edition, published 1670 Stone, G 2003 The Lymington-Keyhaven Medieval and (HMCMS:FA1999.58). Post-Medieval Salt Industry: A Neglected 1689, E Dummer and Captain TWiltshaw, Lymington Archaeological Resource, unpubl. B. A. dis- River East of Hurst Castle (HMCMS: sertation, Exeter University. FA1991.24.10). Sturman, C J 1984 Salt-making in the Lindsey 1693, Captain Greenville Collins, Chart of the marshland in the 16th and early 17th Solent and the Isle of Wight (HMCMS: centuries, in Field, N & White, A (eds), FA2001.97). A Prospect of Lincolnshire, Lincoln, 50-6. 1783, Murdoch Mackenzie, Admiralty Chart, Survey Sumner, H 1917 Ancient Earthworks in the New Forest, of Southampton River (HMCMS: London. FA2000.21.1). Thames Valley Archaeological Services, 2001 Ampress 1811, Ordnance Survey Old Series map of Hampshire Park, Lymington, Hampshire: An Archaeo- (HMCMS:FA1998.91).

Authors: Andrew B. Powell, Kayt Brown, Grace PerpetuaJones, Lorraine Mepham, David Norcott, Ruth Pelling, Chris J. Stevens, Rob Goller, Elizabeth James, Wessex Archaeology, Portway Hose, Old Sarum Park, Salisbury SP4 6EB, web: www.wessexarch.co.uk

APPENDIX 1. FABRIC DESCRIPTIONS

Pottery fabrics amount (20%) of voids, up to 11 mm on surface, probably from shell, and a moderate Iron Age fabrics to common amount (15-20%) of flint, some calcined some not, angular, up to 3 mm, mod- Fl: A soft fabric containing a common amount erately sorted. (20-25%) of flint, mostly calcined, angular, up VQ1: A soft, soapy fabric with a common to very to 4 mm, poorly sorted; rare (1%) iron oxides, common amount (25-30%) of voids, possibly 3mm, sub-rounded. from leached shell fragments, 1-2 mm, mod- Gl: A soft, soapy fabric containing a common erately sorted; and a common amount (20%) amount of grog (20%), sub-angular to angular, of quartz, coarse-grained, sub-rounded to sub- up to 3 mm, poorly sorted. In section the grog angular, well sorted. appears as voids. VQ2: A soft, sandy fabric containing a common G2: A soft, soapy and slightly silty fabric containing amount (20%) of voids, up to 5 mm where a moderate to common amount (15-20%) of visible, in a fine sandy clay matrix with rare grog, angular, up to 1mm, well sorted, and a (1%) red iron inclusions, up to 1mm and moderate amount (15%) of voids, up to 5 mm, detrital flint, up to 8 mm. possibly from grog, poorly sorted, in a silty micaceous fabric. Roman fabrics Ql= A soft, sandy fabric containing an abundance of sub-angular to angular quartz, medium- E211: Oxfordshire mortaria. grained, well sorted, with rare (2%) angular E212: New Forest parchment ware mortaria. coarse to very coarse (up to 1.5 mm) grains. Iron staining visible in break. Medieval fabrics Q2: A soft, harsh fabric containing a common amount (25%) of quartz, clear and opaque E422: Laverstock-type coarsewares. fragments including milky and rose-coloured, Q400: Sandy/flint-tempered medieval fabric. rounded to sub-angular, coarse to very coarse- Q401: Medieval sandy ware. grained, well sorted, and occasional detrital flint fragments, up to 2 mm, angular. Fired clay fabrics Q3: A coarse, sandy fabric containing abundant (40%) quartz, sub-angular to angular, coarse- FC VI: A soft, slighdy silty textured fabric with a grained, well sorted. 1st century AD, could be moderate to common amount (15-20%) of either side of conquest. Very similar to Q4 but linear voids, up to 6mm, and a sparse amount grey in colour. (5-7%) of rounded red iron oxides, up to 1 Q4: A coarse, sandy fabric containing abundant mm. No actual quartz grains are visible. The (40%) quartz, sub-angular to angular, coarse- fabric is quite marly. grained, well sorted. Very similar to Q3 but this FC V2: A soft, slighdy silty textured fabric with an is generally black in colour, similar to Black abundant amount (40%) of linear voids, up to Burnished ware 1. 6 mm, and sparse monocotyledon stems. Q5: A soft, sandy fabric containing abundant (40%) FC Ql: A soft, but harsh sandy fabric, containing quartz, sub-rounded, medium to coarse-grained, a common amount (20-25%) of coarse-sized well sorted, and common (20%) red iron oxides, quartz grains, rounded, well sorted, and sparse rounded, up to 1 mm, moderately sorted and (3%) iron oxides, rounded, 1 mm. On the sparse (5-7%) voids from organic inclusions. surface, larger (up to 2 mm), more angular S99: Small sherds of shell-tempered ware. milky quartz fragments can be seen. VI: A soft and very soapy fabric containing an FC II: A soft, soapy fabric containing a moderate abundant amount (40%) of voids, up to 4 mm, amount (15%) of rounded iron inclusions, up platy, poorly sorted, possibly from shell. to 3 mm, well sorted, in a silty clay matrix. VF1 ;A soft, harsh fabric containing a common