liomnev A4al.sl/ C'ou~talant1 I,andscnpe Changr t/i~.oz~ghthe Ages (ed A Long, S Hzpkrn and H C'larke), OCrLC1Xfonogrrryh 56, 2002, 121-126

8. The Rumenesea Wall, Romney and Walland Marshes: A Commentary

J.R.L. Allen

Recent sjjn/hetic accounts qf tl7e Walland and Romney Marshes accept to different degrees the definitive role advocated for the Rumencsea Wall in the critical early phases oj'the settlement oj the area, bzrt call jbr a .further exumination of several issues oj'juct and opinion, chiefly in the light of' recent archaeological inves/igations. On the balance of evidence at present c~vailable,the Rurnenesea Wall would .seem to be a bridging seahank (Wealden scarp to barrier) qf Saxon date which, with a contemporaneou.s coastal barrier andprobablji other engineered banks, allowed roughljl one-half of the area /o be trtrnsformed into a settled and,farmed landscape. The context and role of the Wall call,fbr.fi~rtherwork, however, and a number qf'new and e.speciallj~environmental research needy are szrggestcd.

Introduction The Rumenesea Wall Pon-erful natural forces and lengthy human interferencc The nuIncrous earthworks of especially have created. in the Walland Marsh and have long been rccognised as important landscape fcatures, proper on the English Channel. perhaps the most complex but it is only recently that the significance of the struchire Holocene geological sequence and associated historic variouslj- called the Rumenesea Wall or Yoke Sewer Wall landscape to be found on the British coast. Despite research (see below) has been recognised. Hence it is appropriate extending over many decades and recent monographic briefly to summarise the evidence for, and character of, treatments (Eddison and Green 1988; Eddison 1995; this stnicture. Eddison rt al. 1998), much relnains either unki~ow~lor The naille Ruinenesea Wall (Allen 1996) applies to a uncertain about the evolution of this considerable area (c. field monument which, although no longer a standing bank, 250 km2) of shifting gravel-sand barriers behind which can be traced continuously over the ground from New sheltered sandy-muddy tidal i~~lets,including river outlets, Romney to (Allen 1999) and which, betu-eel1 and muddy-peaty marshes. Two attempts at a general Snargate and the Wealdell scarp at Appledore, arguably synthesis (Eddison 2000; Rippor~2000) invite cominent lies either beneath or very close to the impressive: later in the light of recent archaeological field work bearing on earthwork known as the (Fig. 8.1). The early reclamation or land-claim in the area and the role of Rumenesea Wall does not appear on Elliot's (in Lewin the earthwork \vhich has coine to be known as the 1862) early maps of enclosing banks. In Green3s (1968) Rumenesea Wall. The analysis brings to light a number of important soil map. the Wall is shown as a definite bank future research needs. only for a distance of about two kilometres northwest of 122 Jolzn Allen

...... averaged about 0.5 m, and in one place a rise of 1.61 m was noted: at only one of the selected sites did the ground prove to fall southwestward (by 0.15 m). This method failed only over an elevated, roughly one kilometre tract north-east of . but where a bank had survived to be recorded in RAF (1916) air photographs. At Owen's Bridge, northwest of Old Romney, a modest rise of 0.18 m in level in the same direction was recorded over the line of a wall standing at right-angles to the Rulnenesea Wall and apparently cut by the 13th-century Rhee Wall (Green 1968: Allen 1996). The course of the Wall betu-een Snargate and the Wealden scarp at Appledore is uncertain. It appears to emerge from beneath the Rhee Wall, a younger structure (Fig. 8.1), but may be represented in part by a lengthy bank Green (1968, soil map) recognised very close to the Rhee Wall on its northern side. As Green's map makes especially clear, the Rhee Wall between Appledore and Snargate pursues an irregular, almost meandering course, KEY as if tracing a natural watercourse (? Rmnenesea. see Holocene fine alluvium below), in contrast to the long, straight segments seen to Holocene 5 km the southeast of Snargate. sands/gravels The Rumenesea Wall between Snargate and Old mpre-Holocene Romney (Fig. 8.1) pursues a sinuous course just to the northeast of, and in harmony with. an also sinuous watercourse, identified with the Rumenesea (Brooks Z;rg. 8.1. The Walland and Romney Marshes showlng the 1988). Now called the Yoke or New Sewer, it has for Xzrnzenesea andRhee Walls andplaces meritroned ln the text. some centuries been one of the larger, managed drains of Tlir Rumenesea 11es between the Jlizlls l~rrtIS om~ttedfor the area. c/ai.rh: A critical property of this watercourse is that its meanders rapidly and systematically roughly treble in wavelength southeastward to values near Old Romney in Old Romney; elsewhere it appears merely as a 'probable excess of 400 m (Allen 1999, fig. 3). In terms of empirical sea wall'. Exactly how Green identified walls is nowhere geomorphological evidence, as was explained, this means explicit. However, as his survey occurred largely before in the case of the natural channel that its width and the highly mechanised arable farming became widespread in discharge of water through it increased rapidly toward the the area, many banks may have been visible at the time as southeast. In form, the waterway very closely resembles upstanding features. Where these structures had already the trunk channels of mature salt-marsh creek networks been removed or had become naturally degraded, he may that expand into the heads of a small tidal inlets or have based his mapping on the survival of ramps (Allen estuaries, and hence is thought to be of tidal origin (Allen 1997,6, illus. 5), that is. consistent and fairly sharp changes 1999). Indeed, the waterway has an almost axially central in ground height across the line of the former bank. Green place within one of the larger occurrences of Green's (1968, 17, fig. 18) had recognised that 'abrupt contrasts (1968, fig. 10) 'land type with creek ridges'. The sinu- in levels andlor sedimenta~patterns' could arise after osities of the Rumenesea Wall, however, although clearly the building of a sea wall. As is fully explained elsewhere in phase with those of the watercourse, are of a smaller (Allen 1997), this is because the erection of a bank, firstly, amplitude and are locally impinged on by that channel in alters the l~ydraulicsof a tidal marsh and, secondly, halts its prescnt-day form. Consequently, it may be argued that siltation on the enclosed area while allowing the continuing the Rumenesea was (1) an active tidal system prior to the build-up of sediment on the portion that remains active. construction of the Wall, (2) already well-developed by The course of the Rumenesea Wall was recently the time the Wall was built, and (3). although constrained confirmed instrumentally on the basis of a consistent to the northeast by the Wall, coatimially or intermittently southwestward increase in ground level across a con- active up until the imposition of the man-made regime tinuous, narrow zone several metres wide that could be that finally changed it into a wllolly freshwater drain. On traced, with but one short break. from the coastal barrier the other hand, Eddison (2000, 84) suggests that 'the at New Ronlney as far as Snargate (Allen 1999, fig. 1). channel ...was originally dug to provide soil for the bank' Levelling was done at 44 rando~nlychosen sites roughly and that later 'the sea broke in, enlarged the eastern end 200 m apart. The measured southwestm-ardrise in elevation of the channel and greatly magnified its curves'. As a dug The Runzenesea Wall, Romney and Wallnnd Alarshes: A Cornrnerztary 123 channel among the unconsolidated sediments of Romney the centre of the Marsh, to emph into the sea at Hythe i11 Marsh. honever, the watema? may be expected to ha\ e the far northeast (Fig. 8.1). This is groundless speculation, been straight. Further evidence on the character and role supported by neither the soil patterns and geomorpho- of the Rumenesea comes from a recently completed, logical features of the area (Green 1968, soil map) nor the detailed analysis bq Dr A.F. Pearson and Professor J.F. long, high-resolution, stratigraphic profile colnpleted by Potter of the distribution and geological provenance of Long et al. (1998) on a north-south line across the middle building materials in the churches of the Marshes (Potter. of the Marsh (see also Spencer et al. 1998). personal communication, 2001). Mapping the occurrence No late Roman occupation/activity sites are known from of beach cobbles and boulders of Ashdown Sandstone Romney Marsh proper. One among many possible explan- and of flint. they found that these materials occurred in ations (Rippon 2000) is 'tidal flooding', so long as this is northw est-southeast belts roughly centred about the course understood to mean the replacement of salt marsh, already of the Rumenesea. Their suggestion that boats brought subject to immersion by 100-200 tides annually, by stone at various tiines inland along this channel is not environments (e.g. low marshlmud flatslsand flats) lower incoinpatible with the geomorphological e~idencefor the still in the tidal frame, on account of a steeply increased scale of the waterway (Allen 1999, 8). which in its natural rate of sea-level rise andlor a steep decline in the state could at Old Romney have been about 30 111 across. availabilih of suspended sediment (Allen 1990; French reducing to a width of about 10 m at Snargate. Rafts and 1993). Eddison (2000, 46) favours such a dro~v~ling, boats of the order of 10 m in length could have been writing that 'the Roman land-surface is generally buried operated in such a channel, but there remain issues of beneath a metre or so of later sediments'. Rippon (2000, dating before it can be said nhen the regime was either 138) also favours deep burial, for according to him tidal or fresh. 'palaeoemironmental work has confirmed that around c. AD 300 there was a widespread episode of marine flooding (Long et a/. 1998)'. No sudden. widespread marine transgression is claiilled by Long et al. (1998, 45), who Conditions Antecedent to the Wall wrote that 'By c. 1700 cal. yrs BP intertidal conditions Beyond agreement that salt-making occurred on Romney had returned to most of Romney and Walland Marshes'. Marsh proper during Roman times, at such widely a reference to what they had shown to be the gradual dispersed sites (Fig. 8.1) as (Isaacson 1846). retreat, west\$-ard and southward across Romney Marsh Ruckinge (Bradshaw 1970), and Snave (Reeves 1995), proper, of peat-forming conditions. begun in the north no final view can be given on the environmental character and east more than a millennium earlier. and evolution of the Marsh before the building of the 111 support of their conclusions, both Eddison (2000, Rumenesea Wall. There was some later Iron Age activity 42) and Rippon (2000. 138) point to early Roman sites on in the area, but of the now many known Ro~nallsites. none Romney Marsh proper now buried many decimetres below post-date the early third century AD. present ground level. Although archaeological materials Most of the Marsh in the early Roman period is can achie\:e deep burial in coastal lowlands because they suggested by Eddison (2000, 52, fig. 14) to have been ended up in creeks or large drains (e.g. Hawkins 1973), co~nposedof a 'lagoon' filled with 'tidal sand and some of the deeply buried finds noted from the Marsh mudflats'. on the basis of activityloccupatiori sites appear to be genuinely related to subsequent episodes of supposedly restricted to the northwestern margin below sediment accretion outside of creeks, for extensive the Wealden scarp, and to the inner fringes of the coastal occupation surjbces were seen (e.g. Isaacson 1846). The sand-gra\:el barrier in the southeast. Although so far known sites of deep burial are, however, few in number restricted to modest areas. recent field investigations and, crucially. share the same. particular context. They all chiefly by Reeves (1995; personal communication, 1998) occur around the inner fringes of the narrow coastal barrier. show that sites of this date occur much more widely and in the Dymchurch-St. Mary's Bay area (Fig. 8. l), where plentifully (Reeves 1995, fig. 5.3; Allen 1999, fig. %A). storm ovenvashing and breaching may be expected to For example, a loose cluster of thirteen sites occurs over have spread raised plu~nesof sediment, and created short- a substantial area in Orgarswick parish (Fig. 8. l), in the lived creeks, that ranged inland for a little way. Sites middle of Eddison's supposed lagoon. Several sites near further inland still - the great majority known from Newchurch lie within the outcrop of Green's (1968, soil Romney Marsh (Reeves 1995, fig. 5.3; Allen 1999, fig. map) calcareous soils, again coincident with much of the %A)- have come to light through field-walking and other postulated lagoon. surface searches (Reeves 1995, personal communication, On the other hand, Rippon (2000, 60, fig. 67A) states 1998). Consequently, they lie within normal plough depth. that at the time in question 'the whole of Romney Marsh The totality of the evidence at present available does not was in fact a tidal saltmarsh'. One difficulty with this suggest that deep burial is typical of early Roman proposal, however, is that a proto-Rolher is carried through occupationlactivity sites in Romney Marsh proper. 124 ,John Allen

Age and Function of the Wall paralivel!. late structure in the area, cutting parishes and estates, although it has come to sene as the administrative Charter evidence reviewed first by Brooks (1988) and boundary of Walland Marsh (Teichman Denlille 1936, later by Eddison (2000) and kppon (2000) leaves little 7). Eddison (2000, 67, figs 27, 29) subsequently repeated doubt that Romney Marsh proper \+-aswidely and probably this mapping. briefly noting that the colonisers 'must have quite densely populated by the 10th or l lth century AD; started in the north-east and moved progressively south- and that the process of settlement had begun by at least \vestwards'. Eddison and Draper's (1997. fig. 2) two the seventh century. The wording of these charters implies .cordons of connected embankments' in the Walland that chiefly arable cultivation and little pasturing were Marsh, sin~ilarto those of Green (1968, fig. 14), later being practised. On the coast la>-fishing communities. become simply 't\\-o embankments' (Eddison 2000, 67). Recent archaeological discoveries vindicate these Each cordon embraces a number of land-claims of various, documentan sources. At least 23 Saxon acti\.ity/occu- chiefly later medieval dates, and probably records a major pation sites are known from the Marsh (Allen 1999, fig. change in natural conditions which temporarily halted 8B). largely as the result of field-walking in a number of further reclamation and forced the repair and general restricted but scattered areas (Ree\-es 1995: personal upgrading of the outermost seabanks in a series of communication, 1998). Substantial sites existed on juxtaposed enclosures of different ages. elevated tracts. such as Sandtun on a dune-covered spit The Rumenesea Wall lies to the northeast of the Rhee near Hythe (Macpherson-Grant and Gardiner 1998: Wall (Fig. 8.1) and is the most northeasterly bank of any Eddison 2000, 58-60), at the former entrance to an east- substance in the Walland and Romney Marshes (Green west tidal system larger than. but resembling geomorpho- 1968. soil map: Allen 1999). It was first proposed in 1996 logicall!-. the Rumenesea (Allen 1999, 8). Most sites lie. that the Wall was of Saxon date, and that it was 'the ho~vever,in Eddison's (2000) 'lagoon' or n-here Rippon earliest sea defence in the area'. the 'defining monument (2000) envisages late Roman 'flooding'. Eddison (2000, of Romney Marsh', and the definitive boundary between 10. 64) expresses the belief that there is 'no relationship the Walland and Romney Marshes as settled landscapes between the sites occupied in the Saxon period and those (Allen 1996, 275, 276). In support of this contention, previously occupied by the Romans' and 'no known reference was made to an eclectic range of partly published relationship between Roman and Saxon landscape'. e~idence:changes in land elevation, soil and stratigraphical Rcccnlly. however. it has been found that at least three patterns. archaeological finds. charters, churches, place substantial early Roman and Saxon find-spots coincide. names, and parish and field patterns. The later. detailed in thc nidely dispersed Marsh parishes of Snave, Old ground suney (Allen 1999. fig. l) yielded no evidence Romney and Newchurch (Reeves, personal communic- against this view, and confirmed beyond reasonable doubt ation, 1998). Other occupationiactivity sites of these the persistence from New Romney to Snargate of Green's periods lie only tens to a hundred metres or so apart, for (1968) largely 'probable sea wall'. Together m ith the crample. in Orgarsnick parish. That both early Roman plausible segment between Appledore and Snargate. it and Saxon occupation debris should occur within plough appeared to form a 'bridging bank' (Allen 1997, illustration depth, and in some instances lie mixed together in the soil 12D) rooted in the high coastal barrier (New Romney) at at the same place, strongly points to the persistence of one end and the Wealden scarp (Appledore) some 12 km landscape features over a period of several centuries, and away at the other (Allen 1996, 276: 1999, 13). Hence the suggests how little, if at all, siltation modified the greater Wall. supplemented by the coastal barrier and perhaps part of Romney Marsh during the earlier first millennium. other seabanks in the north or east (see Green 1968, soil The question naturally arises as to how the change in map), at a stroke made possible the transformation of the land-use between early Roman and Saxon times was intertidal landscape of Romney Marsh proper, and achieved. Given that the underlying trend of relative sea provided high ground in which further land-claims levcls continucd to be upward during the late Holocene, spreading toward the southwest could be rooted (Allen the transforlnation of a tidally-influenced coastal lowland 1996, fig. 1; 1999). into a permanently settled landscape would have demanded Although regarding it as a 'minor bank', and assigning that any existing natural barriers be supplemented by to it no name. Eddison (2000, 83) later claimed the engineered banks. To be effective these must in the first Rumenesea Wall as 'probably constructed in Saron times instance either form a continuous structure encircling the as the defining boundary between the reclamations on choscn area or be rooted at each end in natural high ground Romney Marsh proper and the extensive salt marshes (Allcn 1997, 16-17, illus. 12). There is little consensus which the11 stretched away to the south-west'. It is thereby on the prosecution of such reclamation or land-claim in left unresolved as to whether the Wall was the single, the Walland and Romney Marshes definitive monument which allowed the start of permanent Like Elliott (in Lewin 1862) before them. Eddison and settlement or \\-hether. like the cordons of later banks to Draper (1997, fig. 2) appear to root marsh enclosurcs of the southwest, it merely consolidated the southwest~vard later medie\:al date in the Rhee Wall (Fig. 8. l), but this limit of a group of prior land-claims. The latter is perhaps earthwork, a 13th-century bivallate canal, is a conl- unlikely. for the only enclosures so far identified to the The Runzenesea K.tirll, Xonzney and Jf7alln~~tlAIIarsheLs: A Co~ilnzentary 125 northeast of the Wall lie remote from it, in the Ncwchurch environments that existed there subsequent to the gradual area (Fig. 8.1). Here Reeves (1996, fig. 2) concluded that I\-estward retreat of the pcat marshes identified by Long el a sequence of modest 'innings' (see also Rippon 2000. ul. (199%).that is. in the important inte~alfrom the late fig. 56) had spread southeast\vard. The regional slope is Bronze AgeJIron Age up to Saxon times. from south do\\-11 to north, however. and consideration At prcsent. onl~.circumstantial el-idcnce dates the should be given to the possibility of a staged northward Rumenesca Wall 10 Saxon times. Although the bank is advance of settleincnt from the vicinih of the coastal nowhere upstanding over its tisible extent; enough may barrier toward a bank. which may itself have been settled. be presencd locally ~vithinthe sun-king ramp to allow engineered along a diminished tidal waternay discharging excavations in search of organic material dateable by near Hythe. Rippon (2000) also endorses views pre\-iously accelerator mass-spcctroinetn (AMS). Excavations and a published on the crucial rolc played by the Runleilesca detailed ground suney are also requircd in the area Wall. but prefers the name 'Yoke Sewer Wall' (surely a betwcen Snargate and Appledore, where the Wall appar- name for a freshwater drain) for the structure. Writing entlj. was either subsuined into, or lay parallcl with, the 'that this importailt landscape feature was in Fact the Rhee Wall. In conjunction with thc ell\-ironmental studies origi~lalboundan. between Romney Marsh proper and proposed, esca\:ations are also needed on activity/ Walland Marsh', he too depicts the Wall as thc root of occupation sitcs in Romne!- Marsh proper. at locations further iiledie\-a1 land-claims to the southwest (Rippon \\-here only early Roinan material is kno~vnand also n.here 2000. 157, fig. 67B. C). early Roman and Saxon rcinains coincide. Thc archaeo- logical richness and potential of the area are clear from the work of Rec\.es (1995, 1996). and the rcsults should provide critical evidencc on \\-hat environmental and Concluding Discussioil cultural changes took place in the Marsh in thc centuries The Rulnenesea Wall merits serious considcration as the between early Roinan abandonment, itself puzzling, and most significant eartllrvork on the Romney and Walland the inflow of Saxon peoples. Fieldwalking in the so-far Marshes. but clarification of its status demands new unexplored. greatcr part of the Marsh should strengthen research priorities and a shift of emphasis in the area. the e\-idence for the extent of both Roman and Saxon A better understanding of the Rumcnesea itself ~vould actilities and settlenlent. result from a chronologically-constraincd. stratigraphical and environmental study which aimed to establish the scalc Aclino~vled~c~n~enfs and character of this watcnvay o\,er its kno1+-11length. The I am indebted to Andrcn Pearson and John Potter for the northern tidal waterway merits a similar attack. Studies of opportunit~to makc reference to thcir study of the fora~niniferalassemblages in dated scdiments from pro\ enance and distribution of buildings materials in the Romney Marsh proper l\-ould clarif\. the issuc 01 the marshland churches. in advance of its publication.

References Eddison, S. (ed.) 1995. Ronzney .Zln,:sh: the Debatable Groiilrcl Allen. S.R.L 1990. Salt-marsh growth and stratification: a (Oxford University Committee for Archaeology 41). Oxlord. numerical model with special reference to the Severn Estuary. Eddison; J. 2000. Ro17117eyh4a~sh: Survival on n Frontier. south-west Britain, 'bfarine Geology 95. 77-96. Tempus. Stroud. Allen, J.R.L. 1996. The sequence of early land-claims on the Eddison, J: and Ilraper, G. 1997. A landscape of medieval Walland and Romney Marshes. southern Britain: a pre- reclamat~on:Walland Marsh. Kent, Lundscupe Histor,. 19; lim~narqhypothesis and some implications. Proceedings of 75-88. the Geologists ' Associntion 107, 27 1- 80. Eddison, S. and Green, C. (eds) 1988.lioniney~bliirsh: Evolzition, Allen; J.R.L. 1997. The gcoarchaeology of land-claim in coastal Occz~pution,Reclirnration (Oxford IJniversity Committee for w~etlands:a sketch Prom Britain and the north-west European Archaeology 20). Oxford. Atlantic-North Sea coasts, Archueological Jozimal 154. 1- Ixddison, S.. Gardiner, M. and Long A. (eds) 1998. Roninej' 54. ,$far.sh. Bnvironniental Chnnge uuwci Hz~niunOcc~~palion in Allen. J.R.L. 1999. The Kumenesca Wall and the early settled cr Coustnl Lowland (Oxford University Committee for landscape of Komney Marsh (Kent). Lnndscupe Ilistory 21. Archaeology 46). Orford. 5-18. French. J.R. 1993. Numerical simulation of vertical mal-sh Bradshaw, J. 1970. Ashf'ord area, Ruckinge. Archueologiu growth and aillustment to accelerated sea-level rise, north Cantin~o85, l. Norfolk. IJK.. Eurth Surjuce Processes and Ltrridfornzs 18, Brooks, N. 1988. Komney Marsh in the early Middlc Ages. in 63-8 1. S. Eddison and C. Green (eds) Roniney ~lfcrrsh:Evolrltion, Grecn, R.1) 1968. So11.r of Ronlney .I/crrsh (Soil Survcy of Occzlpution, Reclunzation (Oxford Committee for Archac- Great BI-itain.Bullclin 4). IIarpenden. ology 111, 90 104. Oxlord. Hawkins, A B. 1973. Sea level changes around South\\-est 126 John Allen

England. in D J Blackman (ed ), Marzrie Archaeology Reeves. A 1995 Romne\ Marah the field-walk~ngev~dence, (Colston Papers 23), 67-87 Brlstol in J Edd~son(ed ) Rorntrej March the Debatable Ground Isaacson, S I 1846 The discovery of Roman urns and other (Oxford Un~verslt)Colnln~ttee for Archaeology 41), 78-91 anclent remalns, at D~mchurchIn Romney Marsh.,lrchaeo- Oxford logzi~31, 487-88 Reeves, A 1996 Earthwo~ks survey, Romney Marsh, Archaeo- Lewin, T 1862 The Invas~onof Br~talnby Jzd1111sCaesar logza Cant~ana126, 61-92 London Rippon, S 2000 Thr Transformatzon of Coastal Jletlands Long, A Waller, M, Hughes, P and Spencer, C 1998 The Oxf 01d Holocene deposit~onalh~story of Romney Marsh proper, In Spence~ C Plater A and Long, A 1998 Holocene barr~er J Edd~son,M Gardiner and A Long (eds) Ronzney Marsh estuar\ exolutlon the sed~mentaryrecord of the Walland En~~lronnientalChange and H~itlianOCCIL~LI~IO~Z zn a Coastal Marsh I eglon In J Eddlson. M Gardlner and A Long (eds) Lo~tland(Oxford Unlvers~t) Commlttee fo~Archaeology Romnev Marsh Environmental Change and Human Occu- 46) 45-63 Oxford patzon zn a CoastalLowland (Oxford Un~vers~tyComm~ttee M'~cpherson-Grant. N and Gardlner, M 1998 Pottery from for Archaeology 46). 13-29 Oxford Sandtun, West Hythe, Kent. dledleval Archaeologv News- Te~chlnanDerville, M 1936 The Lrvel and L~berty ofRornney letter 18, 7 \larsh zn the County of Kent. Ashford and London