The Holocene floodplain and alluvial deposits of the Rother valley and their bearing on the evolution of Romney Marsh

with a contribution by Robert G. Scazfe Introduction The Rother and its tributaries are intimately associated with the marshes of Romney and Walland, for these alluvial areas are an extension of the valley deposits out into brackish, estuarine and near-shore depositional environments. There has been much speculation concerning the role of this fluvial system in relation to the development of the marshlands, with attention often focussing on the possible courses of the Rother (or Limen) across the marsh together with the location ofits associated estuaries or inlets (Green 1968, Brooks 198 1 ) . However, relatively little is known about the Rother valley fill deposits. This paper is concerned with describing aspects of the geomorphology of the Rother valley and the fill deposits it contains. This study will focus on the upstream fill sequences, for those further downstream and out into the marsh are described in detail in the previous chapter. The Rother is the principal catchment in the south- eastern Weald and has a drainage area of some 700km2 with a relative relief of 174m. The river rises in a narrow ghyll in the High Weald south of Rotherfield at TQ 557287 and flows briefly south-east until it is joined by a right bank tributary south-west of Mayfield (Fig. 2.1). From this confluence until the Rother Levels and the marshlands, the underfit network is contained within a remarkably linear valley which has a pronounced west- east orientation and, with the exception of one bend between Etchingham and Robertsbridge, is essentially non-meandering. This linearity is also exhibited by some of the more important tributaries including the Tillingham and the Brede. Although small inliers of Upper Jurassic Purbeck Beds have been unroofed along the ridge which forms the south-western interfluve, most of the catchment has developed on interbedded Lower Cretaceous sands, silts and clays (Fig. 2.2) of the Hastings Beds Group, which consists of three important formations: the Ashdown Beds, the Wadhurst Clay and the Tunbridge Wells Sand. Weald Clay outcrops in the north-east of the catchment. Extensive folding and faulting has formed a series of periclines and synclines en echelon which give rise to the major ridges and vales of this part of the Weald. Positive structure-relief relationships exist, with the valley occupying a complex faulted syncline, bounded to the north and south by the Crowborough and Battle anticlines respectively. The Rother is poorly adjusted to the macro-structure of the Weald-Artois anticlinorium but is concordant to the secondary structures which are responsible for the predominantly west-east alignment of the valley. The linearity of the Brede valley is also a consequence of it being fault-aligned. The Rother valley is relatively broad with its sides occasionally mantled by head. It contains few river terraces and its bottom is floored by alluvium of which little is known. Although natural exposures and occasional boreholes have been noted in the geological memoirs, and the general distribution of alluvium mapped (Shephard-Thorn et al. 1966, Bristow & Bazley 1972), there have been very few investigations of the floodplain or valley fill deposits. The aim of this account is to provide some insight into the lithostratigraphy, geometry, sedimentology and chronology of the inland fill deposits and to examine some of the implications of these findings in relation to the development of the marshlands. Lithostratigraphic Investigations Sampling The alluvial stratigraphy of the Rother has been investigated by drilling 134 boreholes at twelve selected sites into the valley fill from Mayfield to Bodiam Castle (Fig. 2.1), using both gouge and exploratory augers. This has enabled the morphology and geometry of the alluvium to be investigated by constructing cross- profiles from the borehole data. Each cross-profile was sited where the valley walls are clearly defined and was made orthogonal to the valley axis. A maximium spacing of 20mm was generally employed between boreholes. The auger was kept as vertical as possible and the variations in the fill deposits recorded. Drilling was continued until bedrock was reached, usually recognised by a sudden increase in resistance to drilling. Estimations of the thickness of the deposits are considered to be within an accuracy of 15mm when using the exploratory corer, for this produced disturbed samples. All boreholes were surveyed in to Ordnance Datum using a Kern Quickset level. 3 2 Paul 3. Burrin The location of the cross-profiles is detailed as follows. The floodplain upstream of Mayfield tends to be relatively narrow and becomes increasingly fragmented within the tributary ghylls. Furthermore, as the valley sides become steeper, so the amount of colluvium in the fill deposits is liable to increase with an accompanying decrease in the alluvial component. The first section is therefore located in the upper Rother valley south-east of Mayfield where the floodplain alluvium is well defined. Twelve boreholes were sunk across the valley floor at the first station (Rl) from TQ604255 to TQ 604253, and a further twelve at the second (R2) from TQ605255 to TQ606253. The close spacing of the two Mayfield stations was arranged to examine the possibility of there being significant local variations in the alluvial fill stratigraphy. The next section (R3) is located 2.8 km downstream south-west of Bivelham Farm where nine boreholes (TQ63 1261 to TQ632259) provided subsurface data for the valley here. Eleven boreholes were sunk across the floodplain south-west of Stonegate (R4) from TQ657267 to TQ655264 with a further eight some 3 km downstream (R5) at Crowhurst Bridge (TQ 684264 to TQ 684262). A further section (R6) is north-west of Etchingham, where eleven boreholes from TQ708268 to TQ708265 have revealed the nature of the valley fill deposits. An additional five sections have been examined by subsurface investig- ations undertaken by Brown (personal communication) along the length of the large valley meander from Etchingham to Robertsbridge. Ten boreholes sunk from TQ 7 1 1264 to TQ 7 12267 across the entrance to the meander (R7) were supplemented by seven (R8) across the valley (TQ 716266 to TQ 714264) immediately upstream of the Limden confluence, and a further ten boreholes (R9) just downstream of the Dudwell confluence (TQ 718265 to TQ 716262). The next section is located 1 km downstream at the valley meander point of inflection where eleven boreholes have been sunk across the floodplain (R1 0) from TQ 7 19259 to TQ 724624. Twelve boreholes were also sunk across the valley floor east of Robertsbridge (R l 1 : TQ 743242 to TQ743238). The final section to be discussed in this paper is R 12 south-east of Bodiam from TQ 785255 to TQ78525 1, where eleven boreholes provide information concerning the valley fill sequence here. A series of generalised cross-profiles has been constructed by interpolation between the boreholes, which enabled variations in the valley fill lithostrati- graphy to be identified, both at-a-station and in a downstream direction. Borehole Records The section at R1 (Fig. 2.3) revealed a maximum fill of 5.8m overlying a polycyclic sub-alluvial surface cut into Wadhurst Clay with a minimum elevation of 39.6m O.D. Overlying and partly mantling the rockhead is a silty sand and gravel deposit which is irregular in thickness and extent. This is buried by fine-grained alluvium with a cross-sectional area of 422m2 in which three lithostratigraphic units can be identified. The deepest is an olive (5Y 5/4), olive-grey (5Y 412,414) and olive brown (2.5Y 414) clayey silt (unit 2) which extends from 39.6m O.D. to 44.0m O.D. Above this a mixed and variable unit is found (unit 3) which is usually a strong brown (7.5 YR 516) andlor light grey (2.5 Y N7/0) silty clay, mottled with dark brown (10 YR 3/3), dark yellowish brown (10 YR 414, 5/6), pale yellow (10 YR 313) and reddish yellow (5 YR 718). Sometimes these secondary colours can dominate the matrix. This unit extends across the 150m width of the valley floor and is found between c. 41.7m O.D. and 45.8m O.D., forming the present-day floodplain surface south of the channel between boreholes 3 and 8. The third unit (unit 4) occupies a smaller area than the two previously described. It has a width of only 54m, is restricted to the northern third of the present floodplain and has an elevation of between 44m O.D. and 45.9m O.D. As the valley sides are approached from the floodplain, the sediments appear to become increasingly coarser, consisting of silts, sands and grits which are here classified as colluvium. The alluvium at section R2 overlies a complex and irregular bedrock surface of Wadhurst Clay (Fig. 2.3) which is again partly veneered by a thin and somewhat localised grey (2.5 Y N7/0) and dark grey (2.5 Y N4/0) clayey silt with inclusions of decomposing wood and other organic macro-remains. The silt coarsens quickly with depth to become a silty sand and gravel (from 39.0m O.D. to 39.5m O.D.). A bluish and greenish grey (5Y 41 1,5 GY 4/ 1, 5/ 1) clayey silt (unit 1) was found in borehole 11 and extends from 40. lm O.D. to 41.lm O.D. This is overlain by an olive (5Y 5/4), olive-grey (5Y 412, 414) and olive brown (2.5Y 414) clayey silt (from 39.8m O.D. to 42. lm O.D.) in which burnt wood remains were found (unit 2). A strong brown and light grey clayey silt (from 41.3m O.D. to 44.2m O.D.) overlies these lower deposits with this unit (unit 3) forming the southern part of the contemporary floodplain surface (boreholes 1-5). North of borehole 5 the floodplain surface is directly underlain by a brown silt which extends from 42.3m O.D. to 44.8m O.D. and has a maximum thickness of 2.lm (unit 4). This sequence appears to dominate the upper Rother valley fills and can be traced as far downstream as Bodiam (R 12). In order to minimise repetition, the fine-grained fill deposits will be subsequently referred to as units 1 to 4 respectively as shown in Fig. 2.3. At R3 the floodplain has widened to 170m. The alluvium occupies a troughlike form in the valley rockhead (Fig. 2.3), has a maximum fill thickness of 3.8m, a mean fill thickness of c. 3.0m and a cross- sectional area of 505m2. Unit 1 ranges in elevation from 31.8m O.D. to 33.4m O.D. and was proven to overlie both the sub-alluvial surface and a silty sand and gravel deposit in the south east of the section. However, beneath the centre of the floodplain, the coarser sediment is overlain by unit 2 (boreholes 4 and 7) or unit 3 (boreholes 5 and 6). The former unit is relatively thin, extends from 32m O.D. to 33.5m O.D. (proven) or Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley 46- 44- 42 - 40 - MAYFIELO (R21 37l BIVELHAM FARM (R3) 30 [3 UNIT 1 WADHURST CLAY SSE 7 , , 20m WADHURST CLAY WADHURST CLAY NNW Fig. 2.3 Valley jill lithostratigraphy: sections R1 to R3. PI UNIT 4 I.",:UI ORGANIC CLAY BOREHOLE LOCATION P LOCATION OF POLLEN CORE OR SAMPLES m COLLUVIUM SlLTY SANDS & GRAVELS Fig. 2.3~ Key to Figures 2.3-2.5. possibly 34m O.D. (interpolated), and is buried by unit 3 which varies in elevation from 32.4 m O.D. to 34.9m O.D. and occupies the full width of the valley floor. The uppermost deposit is unit 4 and has been found immediately beneath the contemporary floodplain surface extending from 34m O.D. to 35.5m O.D. Subsurface investigations 2.9km further downstream at Stonegate (R4) proved a fill with a maximum thickness of 5.8m and a cross-sectional area of 1 195m2, overlying a broad, gently undulating, sub-alluvial surface of Wadhurst Clay with a minimum elevation of 23.4~1 O.D. (Fig. 2.4). A dark grey silt with fine sand and grit inclusions was found in borehole 7 from 23.4m O.D. to 24.6m O.D. Unit 1 was proven in boreholes 5-10 beneath the central and north-east parts of the floodplain surface, from 23.4m O.D. to 25.8m O.D. The rockhead is also overlain by unit 2 (boreholes 2 and 3) from 23.8 m O.D. to 27.9m O.D. and unit 3 (borehole 1) which has a range in elevation from 23.2m O.D. to 29.4m O.D. The present-day floodplain surface, with the exception of its south-western flank, consists of the uppermost elements of unit 4 (27.0m O.D. to 29.4m O.D.) and has a width of 270m. At R5, the floodplain width has narrowed to 180m. The alluvium has a maximum fill thickness of 5.8m and a cross-sectional area of 785mz. The underlying Wadhurst Clay rockhead has a polycyclic form (Fig. 2.4) and is overlain by different units - unit 1 as proven in boreholes 34 Paul 3. Burrin 2 and 6 (fkom 15.9m O.D. to 17.8m O.D.); unit 2 as shown in borehole 7 (from 16.3m O.D. to 19.5m O.D.) and unit 3 as identified in boreholes 3-5 (from 17.8m O.D. to 2 1.6m O.D.). These units are in turn buried by unit 4 which extends from 20.2m O.D. to 22.0m O.D. At R6 the sub-alluvial surface is cut into Ashdown Beds and has a troughlike form with a minimum elevation of 9.2m O.D. (Fig. 2.4). It is buried by alluvium those maximum thickness is 7.3m and which has a cross-sectional area of 1246m2. A silty sand and gravel overlies much of the rockhead except beneath the central floodplain (boreholes 6, 7 and 11) where it is buried by unit 1 (from 9.2m O.D. to 13.3m O.D.) and unit 2 (10.3m O.D. to 14.7m O.D.). In the north of this section (boreholes 8-10) the coarser materials are covered by unit 2, while unit 3 is very extensive at this location extending from 12m O.D. to 16.6m O.D. and forming much of the floodplain surface. Unit 4 (boreholes 4 and 5) extends from 14.0m O.D. to 16.6m O.D. and is found to the south of the present Rother channel which has been artifically straightened and embanked here. Downstream at R7 (Fig. 2.4) and beyond, the sub- alluvial surface has also been developed in Ashdown Beds and is both polycyclic and irregular in cross-profile. It is partly covered by silty sands and gravels (boreholes 1, 2, 4, 7 and IO), or by unit 1 (which extends from c. 7.7m O.D. to 13.lm O.D. and also buries the coarser deposits), or unit 2 (borehole 5), with this latter unit occurring from 7.7m O.D. to 14m O.D. These materials are overlain by unit 3 (extending from 1 1.6m O.D. to 15.7m O.D.) except in borehole 6 where unit 4 was proven. The latter is spatially confined to the centre of the floodplain with a width ofc. 80m, between 14m O.D. and 15.8m O.D. Half a kilometre downstream at R8 the floodplain has narrowed from 270m to 163m (Fig. 2.5). The sub- alluvial surface is trough-like and is buried by a fill with a maximum thickness of 7.5m and a cross-sectional area of 85 1 m2. Silty sands and gravels were proven in borehole 6, but the rockhead is generally overlain by unit 2 which extends from 8.lm O.D. to 14.4m O.D. Although the floodplain surface is underlain by unit 3 in boreholes 2 and 7 (fi-om 14.4m O.D. to 16m O.D.), unit 4 forms the floodplain in the central part of the valley, extending in elevation from 14.2m O.D. to 15.8m O.D. The internal variation of the valley fill is again evident some 300m further downstream at R9 (Fig. 2.5) where the floodplain widens to 258m, the maximum thickness increases to 8.lm and the fill cross-sectional area is 1228m2. The sub-alluvial surface is asymmetrical being deepest beneath the south-western part of the floodplain with a minimum elevation of 7.3m O.D. The silty sand and gravel has only been proven in boreholes 5 and 10 having a maximum thickness ofc. 0.5m. Unit 1 generally overlies the suballuvial surface with the exception of borehole 8 in which unit 2 is found to cover the rockhead. The former deposit occurs between 7.5m O.D. and 14.3m O.D., whilst the latter ranges from 8.5m O.D. to 14.3m O.D. Unit 3 appears to underlie much of the floodplain surface and occurs from 12.3m O.D. to 14.8m O.D., whilst unit 4 is shown to be spatially confined to the south-west of the section, ranging in elevation from 14.4m O.D. to 15.6m O.D. and only being proven in boreholes '1-3. At R10 (Fig. 2.5), the floodplain has increased in width to 306m whilst the maximum fill depth is 8.4 m and the cross-sectional area is 17 16m2. The sub-alluvial surface is polycyclic with a minimum elevation of 6.6m O.D. and is overlain by silty sand and gravel over much of its width. However, some boreholes (1, 4-5 and 11) failed to locate this coarser material and indicated that unit 1 (from 7.0m O.D. to 11.4m O.D.) overlies the rockhead in these locations, whilst unit 2 (ranging from 7.4m O.D. to 13.5m O.D.) is found to cover the sub- alluvial surface in borehole 11 and appears to be the thickest (c. 6m) of the four units at this site. Although unit 3 occupies virtually the whole width of the valley and ranges in elevation from 1 1.4m O.D. to 14.9m O.D., it is buried across the entire floodplain, except in the extreme east, by unit 4 which is c. 1.2m thick and occurs from 13.2m O.D. to 15.3m O.D., forming much of the present floodplain surface. At Robertsbridge (Rll), an undulating, polycyclic sub-alluvial surface, with a minimum elevation of 1.7m O.D. is partly covered by a thin (0.25m) silty sand and gravel. This is usually buried by unit 1 (boreholes 3-10 and 12) which extends from 1.7m O.D. to 8.0m O.D. A dark grey clayey silt, with inclusions of fine sand and grit, as well as interbedded greenish-grey clayey silt, peat, decomposing wood, charcoal and other organic inclusions, was found in borehole 12 between c. 5.8m O.D. and 7.6m O.D., apparently interbedded within unit one (Fig. 2.5). Unit 2 is found overlying the rockhead (boreholes 1,2 and 1 l), and extends the width of the valley floor ranging in elevation from 4m O.D. to 9.9m O.D. The upper two units (3 and 4) extend the full width of the valley, except unit 4 is absent in borehole 1 1 in the extreme north of the section. The former unit occurs from 6. lm O.D. to 1 1.4m O.D., whilst the latter ranges from 9.8m O.D. to 12.4m O.D. The final section discussed here is R12 at Bodiam Castle and it is evident that the valley fill sequences here begin to change significantly from those recorded above. Eleven boreholes have revealed (Fig. 2.6) that the sequences described for the valley fill upstream of this site are replaced by a more complex and variable suite of deposits. Bedrock was only proven close to the valley sides; beneath the centre of the floodplain the suballuvial surface lies at a depth in excess of I lm and was not reached. The deepest deposit identified is a blue-grey silty clay which appears to be similar in colour and texture to unit 1 and has a maximum elevation of c. - 3.0m O.D. This is overlain by peat (between - 5.8m O.D. and 1.2m O.D.) except in borehole 2 where a greenish-grey silty clay with lenses of sand and occasional small shell fragments is found (from c. - 5.5m O.D. to -3.0m O.D.). The peat contains abundant wood remains and macro-fossils including those of Corylus (hazel), although in boreholes 7, 8 and 11 it is Holocene Floodplain and Alluvial Fill Deposits of the Rother I'alley STONEGATE (R41 SW mOD CROWHURST BRIDGE (R51 S m 0.0. 22 - 20 - 18 - 16 - 1 ETCHINGHAM (AB) ETCHINGHAM (R71 16- 14- 10 12 - - 8- 6- WADHURST CLAY 2 3 4 WADHURST C 5 6 7 ASHDOWN BEDS SHDOWN BEDS 8 N Fig. 2.4 Valley Jill lithostratigraphy: sections R4 to R7. NE Paul 3. Burrin 6 ETCHINGHAM (R9) ETCHINGHAM (RIO) m0 D '" 1 16 WNW " ' "I- ASHDOWN BEDS 1 U 2 3 4 5 6 ASHDOWN BEDS ROBERTSBRIOGE (R11) S .. m00 4 12- 10 - 8 -( 6 4 2 0 - - - - D - 2 20m 3 4 5 ASHOOWN BEDS 6 Fig. 2.5 ValleyjlI lithostratigraphy: sections R8 to RII. ESE 7 P 7 9 l0 Q l1 10 N 11 Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley m LIf;yTCGLFYY Romano-Brttlsh horizonic.O,33rn 0.0. or 1.2m below ground Wealden blast-furnace slag= 1.80rn 0.0. or 0.45rn below ground PEAT 8 STRONG BROWN LAMINATED BLACK.GREY 8 m BLUE-GREY SlLTY CLAY Bodsarn Castle dock 8s at 7.5m O.D. BEDROCK UNDIFFERENTIATED DEPOSITS Fig. 2.6 Valleyjll lithostratigraphy at Bodiam (R12). detrital within a grey silty clay. The peat is overlain by grey, black and brown, laminated sands, silts and clays in boreholes 7 and 1 1 (from - 1.2m O.D. to 0.8m O.D.), while in borehole 8 a thin blue-grey silty clay (from - 0.75m O.D. to 0.35m O.D.) overlies the organic unit. More usually the peat is buried by a light grey mottled with reddish yellow (7.5 YR 716) and strong brown (7.5 YR 516) silty clay (from -0.5m O.D. to 2.5m O.D.) which extends the full width of the valley. The units comprising the rest of the lower Rother valley fill are described and discussed more fully in Waller et al. ( 1988). The main criterion for identifying apparent lithostrat- igraphic variations within the inland fill sequences is colour changes. It is recognised that this is a most unsatisfactory indicator as such differences might result from seasonal and permanent gleying as noted in other valley fills (Dury 1964). Evidence of gleying can be found in the alluvial deposits but the persistent spatial ordering of the alluvial stratigraphy (Figs. 2.3-2.5), the nature of the internal discontinuities and the variation in the inclusions found within the various units suggests that the colours identified reflect more than gleying alone. The reason(s) for this variation are unclear and require further investigation. Down Valley Variations The change in the units recognised at-a-station from the relatively ubiquitous, essentially inorganic, sequences recorded upstream from Bodiam and the more complex and variable sequences identified at and beyond this site is m BROWN SANDS. SILTS 8 CLAYS MADE GROUND m FLOODPLAIN .SOIL between these two associations requires further investigation, particularly the inter-relationships be- tween units 3 and 4 with the peat and post-peat deposits. Unit 2 has not been proven at Bodiam and it is unclear whether the bluish-grey clays consist of a single, or two units. Hence, it is difficult to ascertain the inter- relationships of these lower accumulations. The change in the valley fill associations is also reflected in the floodplain long-profile. Two components can be recognised: a lower relatively planar segment is found to intersect a floodplain surface which has a considerably steeper gradient and a negative exponent- ial form. The point of intersection seems to lie just upstream from Bodiam in the vicinity of Udiam suggesting that the coastal association is replaced at this point by the inland valley fill sequences. Similar floodplain longprofiles have been constructed for other Sussex rivers (Kirkaldy and Bull 1940; Jones 1981; Burrin 1983a, Burrin and Jones in press, Waller et al. ( 1988). A brief comment can also be made regarding the morphology of the suballuvial surface. The rockhead from Mayfield to Robertsbridge has a general gradient of 1:438 and does not appear to reflect lithological, stratigraphical or structural controls. There is no significant steepening in the vicinity of the Dudwell and Limden confluences suggesting that although there is a locally marked increase in both catchment area and discharge at this point, this has had little effect on the sculpturing of the sub-alluvial surface. Beyond Roberts- bridge the rockhead rapidly decreases in elevation over a important. An interface can clearly be recognised between the inland facies or association of valley fill sediments (comprised of units 1-4) and the coastal zone facies or association. This sedimentary change can be more clearly appreciated when downstream variations in the valley fill accumulations are plotted (Fig. 2.7). The complex nature of the sedimentary interface distance of almost 20km from c. 1.7m O.D. (Roberts- bridge) to c. -30m O.D. at Rye. The reason for this sudden steepening of the sub-alluvial surface is not clear. It does not appear to be the result of geological control but may reflect fluvial erosion or reexcavation of the rockhead during periods of lower sea levels, such as during the Devensian. Paul J. Burrin BIVELHAM (R31 - E - 0 c .- m 0 > - 40 20 O.D. -20 10 I STONEGATE (R41 15 Wadhurst Clay Fig. 2.7 CROWHURST (RS) ETCHINGHAM (R6) 20 UNIT4 m UNIT II7] UNIT 2 0 UNIT l 0 BEDROCK ROBERTSBRIDGE (R11) 25 Ashdown Beds Valley jll lithostratigraphy in a downstream direction (generalised) Alluvial Geometry At-a-station and downstream variations in the thickness and extent of the alluvium in the upper Rother valley suggest there are systematic trends in the data. In order to investigate this possibility the spatial variation in the valley fill geometry from Mayfield to Robertsbridge (Table 2.1) was examined statistically using correlation and regression analyses. The alluvium at these sites POST-PEAT m DE,OSlTS PEAT m BLUE CLAY BODIAM (RlZ) 30 that the valley fills not only increase in size systematically in a downstream direction but are also related to catchment area. Semi-log transformations (Table 2.2b) generally did not improve these results which is surprising given the negative exponential form of the long-profile, although in the case of AREA versus FPWl the correlation coefficient was improved very slightly. Log-log transformations (Table 2.2~) belongs to the inland valley fill facies which forms a distinctive component of the total valley fill. Hence, data from the Bodiam section, which marks the inland extremities of the coastal zone facies, was excluded from this analysis. The relationships between five variables were considered in the first instance to see whether the obvious increases in the width and depth of the alluvial tract in the inland valley sector could be related to catchment parameters. The five variables selected were: a. The maximum proven fill thickness (MAXF) recorded for each section; b. The width of the floodplain at each cross-section as surveyed in the field (FPWI) which is representative of the average floodplain width within that part of the valley; c. The width of alluvium at each cross-section when obtained from 1 :63 360 and 1:50 000 scale geological maps (FPW2); d. Distance downstream to each cross-section (LENGTH) as measured along the valley axis from the inland limit of the alluvium; e. Catchment area upstream of each cross-section (AREA) obtained from 1:50 000 scale topographic maps. Strong linear correlations have been found to exist between these variables (Table 2.2a) demonstrating improved the strength of the relationships between valley length (LENGTH) and floodplain width (FPWl) and also catchment area (AREA) and floodplain width (FPW2), but did not improve the strength of the relationships between maximum fill thickness (MAXF) and these catchment characteristics. However, high linear correl- ations exist between MAXF and both measures of floodplain width (Table 2.3) although the strength of these relationships was not improved by either semi-log or log-log transformations. The high correlations obtained in this preliminary analysis prompted the examination of whether other alluvial parameters could be related to the same easily measured geomorphological variables (FPW 1, F 1 W2, LENGTH, AREA). The following were selected for study: a. Alluvial fill cross-sectional area (AFA) calculated by measuring the area between the rockhead profile as interpolated between boreholes and surveyed flood- plain surface profile; b. Mean fill depth 1 (MF1) calculated by averaging the depth to rockhead recorded in boreholes; c. Mean fill depth 2 (MF2) obtained by dividing the cross-sectional area of the alluvial fill (AFA) by the surveyed floodplain width (FPWl) and d. Mean fill depth 3 (MF3) calculated as for MF2 but substituting FPW2. Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley Table 2.1 The alluvial geometry of the Rother valley (inland sector): the database SITE LENGTH AREA MAXF FPWl FPW2 MF1 MF2 MF3 AFA km km2 m m m m m m m2 NO. OF BOREHOLES Table 2.2 Correlation coejicients for alluvialjll variables in the upper Rother: (a) linear; (b) semi-log; and (c) log-log relationsh$s (a) LENGTH AREA (b) LENGTH AREA (c\ LOGLENGTH LOGAREA N = 11; rcrit (0.05) = 0.52 MAXF FPWl FPW2 LOGMAXF LOGFPW1 LOGFPW2 LOGMAXF LOGFPW1 LOGFPWP 0.80 0.81 0.74 0.78 0.90 0.87 MFl LOGMFl LOGMF1 0.94 0.95 MF2 LOGMF2 LOGMF2 0.96 0.96 Table 2.3 Correlation coeficients forjoodplain width and alluvialjll depths MAXF FPWl FPW2 LOGFPWl LOGFPW2 N= 11; rcrit (0.05) = 0.52 Despite the variation in rockhead morphology good linear correlations were found for AFA, MF1, MF2 and MF3 against down valley distance (LENGTH) and catchment area (AREA) (Table 2.2a). Semi-log transformations (Table 2.2b) impoved the coefficients for LENGTH against MF1 and MF2, While log-log transformations (Table 2.2~) yielded the best coefficients for LENGTH versus MF3 and AREA versus AFA, MFI, MF2 and MF3. Surprisingly good relationships were also obtained for MF1 versus FPWl and FPW2 (Table 2.3). MFI LOGMAXF LOGMFl MF3 LOGMF3 LOGMF3 0.76 0.83 AFA LOGAFA LOGAFA 0.91 0.94 Regression analyses were subsequently performed on those variables which were most strongly correlated. These analyses indicate that maximum fill and mean fill thickness increase systematically with respect to both distance downstream and catchment area as defined by the equations in Figs. 2.8-2.10. Similar relationships exist between floodplain width, alluvial fill cross- sectional area and the variables LENGTH and AREA. Of further interest are the correlation coefficients and regression equations obtained between measures of floodplain width and both maximum fill depth (MAXF) 40 Paul J. Burrin and mean fill depth (MFI). These results indicate that good predictions of maximum and mean fill thicknesses are available for the Rother valley fill by employing a range of equations involving catchment parameters or measures offloodplain width. Hence, it is possible to gain an indication as to the dimensions of subsurface parameters by measuring easily obtainable geomorphic variables. The strength of these relationships in the upper Rother valley suggest that, should similar relationships be proven for other fluvial systems, it may prove possible to estimate subsurface parameters from the measurement of surface variables, a finding which has potential value to applied geomorphologists, geologists and civil engineers. Whilst it has long been appreciated that floodplain dimensions usually increase in a downstream direction (Wolman and Leopold 1957; Bhowmik 1984), the finding that valley fills exhibit a similar trend is unexpected given the range of environmental factors that can control the development of such deposits. The effect of catchment lithological or structural variations does not seem to influence the dimensions of the valley fills. Similarly, hydrological variation does not appear to be an important consideration in the development of the valley fill upstream ofBodiam which is surprising in that the confluences of two tributary rivers, the Limden and Dudwell occur within a very short distance (c. 250m) of each other. This sudden increase in drainage area might lead to the expectation of an increase in the capacity for erosion and hence valley size beyond this point. However, it is not until the vicinity of Bodiam that the valley begins to widen and deepen dramatically and this appears to correlate well with the change in the valley fill sequences from the inland to the coastal or perimarine facies. This finding is in accordance with a similar study undertaken in the neighbouring Sussex Ouse valley (Burrin 1983a; Burrin and Jones in press). Sedimentological Investigations Sampling and Results The spatial variation in the sedimentology of the valley alluvium and the character ofthe contemporary channel bedload were investigated by particle size analysis. Attention was focused on three sites, at Mayfield (RI), Stonegate (R4) and Robertsbridge (R1 1). At R1, the channel bedload was randomly sampled within the vicinity of the cross section. Three boreholes were also selected in order to identify sedimentological variations within the valley fill at-a-station. The first of these boreholes (10) is adjacent to the present channel and allows the interaction of contemporary fluvial processes with the floodplain to be ascertained. Borehole 6 is located near the centre of the floodplain and is approximately mid-way between the channel and valley-side, whilst borehole 3 is close to the valley-side. Downstream variations were investigated from samples taken from boreholes adjacent to the channel at R4 and R1 1. Samples at all sites were taken at 50cm intervals depth and returned to the laboratory for standard pretreatments and dispersion procedures prior to particle size analysis by the hydrometer method (British Standards Institute, No 1377). Grain size distributions and momerlts analyses were calculated using a micro- computer employing the equations of Folk and Ward (1957). The channel bedload consists of fine and medium gravels with some fine to coarse sands (Fig. 2.11). This contrasts markedly with the valley fills which are dominated by silts and clays, with variable sand inclusions depending on location (Figs. 2.12 & 2.13). Although the fills show some internal variation in terms of the percentage sand or clay fractions present, all the 56 samples analysed had a significant and usually dominant mode within a very limited band (40 to 60 C#J or c. 4.05 to 4.604) within the coarse silt range (Figs. 2.12 & 2.13). This mode can account for over 50% of the distribution, but more usually accounts for 17% to 30% of the samples analysed. Only 18% of the samples were uni-modal (with the mode occurring in the coarse silt band described above); most (52%) were bi-modal with a secondary mode in either the fine sand or fine silt fractions, while 30% were polymodal. At R1 the sand content was found to rise significantly (up to 35'/;,) in samples taken close to the valley-sides or near the channel. There is also evidence of a general coarsening of the alluvium overlying the rockhead at all the sites investigated. Moments analyses revealed that the valley fill samples have mean grain sizes in the coarse silt range, are poorly or very poorly sorted with most being positively or very positively skewed. There is little variation in the sorting or skewness parameters at-a-station or in a downstream direction (Fig. 2.14) for the samples examined here. Although the kurtosis results varied considerably from very platykurtic to very leptokurtic, no significant sedimentological differences between the lithostrati- graphic units described above could be identified with the exception of a general coarsening in the alluvium immediately overlying the sub-alluvial surface. Sediment Sources The markedly different character of the present-day channel bedload in comparison with the alluvial fills, together with the spatial variation in the latter, merit further consideration. Although relatively few samples of alluvium have been investigated from the Rother valley, a considerably larger number of samples (over 300) have been analysed for sites in the adjacent Ouse and Cuckmere basins (Burrin 198 1, 1983a, 1983b). This allows comparisons to be made between the sediment- ological characteristics of these different valley fills. Although the Rother is able to transport sands and fine to medium gravel at high stages, the boreholes have demonstrated that these coarser sediments are not found within the inland valley fills. The silty sands and gravels found overlying the rockhead are finer than the present Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley MAXF 10- 0- 8 0 Log MFlLLl 1- r 0 8cogFpW1 r.0.74 5- I 3 . r20.88 LENGTH 25 FPW 2 400 0- , 0 LENGTH 25 0- 8 Fig. 2.8 Linear regressions for measures of alluvial geometry versus catchment characteristics. LogMFILL2 LENGTH 25 0 a 0 Log FPWl 6 - LENGTH 25 r.0.79 Fig. 2.9 Semi-log regressions for measures of alluvial geometry versus catchment characteristics. Log MFlLL3 2- Log LENGTH 3.5 LogAREA 0-, r- 0.76 I-, 6 Log AREA 6 3 dog AFA r.0.94 6 , 3 Log AFA 8- r.0.91 Log LENGTH 3.5 Log AREA 6 Log AREA 6 6- 8 2 0-8 3 Fig. 2.10 Log-log regressions for measures of alluvial geometry versus catchment characteristics. AREA 200 ARE& 6 Log LENGTH 3.5 Log AREA 6 Paul J. Burrin - - Total available sedimentological spectrum for catchment bedrocks Channel bedload GRAVEL SAND - - Typical valley- fill range ' SILT CLAY Fig. 2.11 Grain-size dislributions for catchment bedrocks, channel bedload and valleyjll deposits. MAYFIELD (RI) CHANNEL MARGINS % 100- 80- 60- 0 m depth --- 2.0 ..S.. 3.5 2 4 0 VALLEY SIDE % 100- 80- 60- 40- 20 - 0, 6 8 10 0 m depth ..... 2.0 MID VALLEY % 100- 80- 60- 40- 20- 07 4 6 0 8 10 0 Fig. 2.12 Grain-size distributions for selectedjll samples at-a-station. STONEGATE (R41 CHANNEL MARGINS % 100- 60 - 0 m depth - 3.0 ..." 6.0 2 4 ROBERTSBRIDGE (R11) CHANNEL MARGINS % 100- 80- 60- 6 8 10 0 m depth - 3.0 --- 5.0 .S... 7.0 2 4 0 6 m depth --- 2.0 0 . ... 3.0 2 8 10 4 l Fig. 2.13 Grain-size distributions for selected valley $11 samples in a downstream direction. 6 0 8 10 Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley MAYFIELD M1 O J 6 VpS,M,S VYS VN , S ,V: STON EGATE M2 M3 M4 V? , M , VF M1 M2 0 0 6 VPS MS VWS V,N $ , Vf' Vr , M , ROBERTSBRIDGE M3 M4 M, M2 M3 - - M* 0 4 6 VPS MS VWS V N $ , "1 1 - 2- - c P $ E bedrock 3- 4 - 5- 5.8 Fig. 2.14 Moments characteristics for selected boreholes. channel sediments and frequently appear to be either the result ofthe weathering of the rockhead or, more usually, an older fluvial deposit. The lack of gravels within the bulk of the valley fills suggests that the coarse channel deposits are a more recent input to the fluvial system indicating a change in the sediment supply. Their presence in the present river indicates that denudation of the local rocks can produce these clastics. As the sources of these materials (presumably from the more arenaceous facies of the Hastings Beds Group) should have remained available during the Holocene it is surprising that these gravels are not found within the valley fills. Similar findings have been reported from other valleys in south-east England (Burrin 1983a). This raises the question as to why such coarse materials were not formerly available to the Rother when they are currently a dynamic constituent of the fluvial system. A related and equally noteworthy observation is the general coarsening of the valley fill close to the channel and adjacent to the valley-sides. This trend has been found at many of the stations investigated and is recorded in the borehole logs. This finding is almost certainly the result of hillslope processes increasing the supply of coarser materials to the valley floor. The near- channel valley fill sedimentology similarly reflects the interaction between the coarser sediments being transported by the river and the floodplain. Interest- ingly, no gravels have been located in the boreholes close to the channel at any station, suggesting little channel migration at the sites investigated since these coarser materials become available to the fluvial network. This is initially surprising in that there is good evidence at many sites ofsignificant bank erosion at the present time. This, together with the more recent input of gravels to the fluvial system, suggests that river metamorphosis may be taking place as the Rother adjusts to a new V : unlt 3 ""lt 2 ""It 1 grey sllt bedrock L 1 grey slit li! VP V ! M VL sedimentological (and perhaps hydrological) regime. The reason for this may well be related to the question posed above and will be discussed further subsequently. The alluvial deposits in the centre of the valley have a lower sand component (Fig. 2.12) and this may be because of their position away from the valley sides. Conversely, this fining might also reflect processes of floodplain sedimentation - e.g. backswamp or flood- basin deposition. If the spatial variation in these sediments is indicative of such activity, then lateral migration of the channel by autogenic processes has almost certainly been limited. This view is also offered support in that many parish boundaries follow the course of the channel and there are few locations where there is an obvious deviation of one from the other, which indicates channel migration. Furthermore, if the Rother had migrated freely across the floodplain such differentiation of the alluvium would be unlikely because fine-grained overbank-derived accumulations generally take many years to develop sizeable thicknesses (Bridge and Leeder 1979). The presence of such sequences implies little or no reworking by the river, indicating a stable channel. However, these processes do not explain the change in sediment supply described above nor do they account for the high coarse silt fraction found in many of the samples analysed at this station irrespective of location. This latter feature is an interesting phenomenon in that it is a persistent characteristic of valley fill deposits inland from the coastal zone elsewhere in the Weald (Burrin 198 1, 1983a). Two contrasting explanations have been forwarded to explain this. Detailed sedimentological analyses of these alluvial deposits have shown that they are not only dominated by coarse silt, but are poorly or very poorly sorted, positively or very positively skewed, with a tendency to be leptokurtic. As 44 Paul 3. Burrin these are classic properties of wind-blown sediments or loess (Catt 1977, 1978, 1985) it has been suggested (Burrin 1981, 1982, 1983a, 198313) that these alluvial deposits contain reworked loess with variable inclusions of locally derived sediments. Conversely, Gallois (1982) has noted that silts can form a significant proportion of Wealden bedrocks and hence, the alluvial silts may simply have been derived from selective reworking of these deposits (Fig. 2.1 1). While this explanation might also account for the lack of gravels in the valley fill, the reason(s) why selective sorting should have occurred and then subsequently ceased are not considered or discussed by Gallois (1982). Furthermore, this explan- ation does not account for the following findings. First, although the bedrocks contain silts, they do not appear to have such high fractions in the coarse silt band which is usually dominant in the alluvium. For instance, the Wadhurst Clay which is perhaps the most similar of the bedrocks to the valley fills with respect to sorting, skewness and kurtosis (Table 2.4), has a coarse silt content of less than 17% in the critical range of 40-60p, whilst the alluvium has values of up to 55% in the same interval. There is to the author's knowledge no documented process other than aeolian sorting that can cause this selective enhancement of coarse silts as found in the valley fills. Second, the bedrocks exhibit a wider range of moments characteristics than is found in the valley alluvium. Samples of the Hastings Beds Group (excluding gravels) which form the main catchment bedrocks have mean grain sizes ranging from fine sands to clays (Table 2.4), and a more variable sorting, skewness and kurtosis than that of the alluvium. If the argillaceous lithologies within the more arenaceous formations are analysed, there is a similar diversity in the moments. This suggests that a similar range should also be found in the alluvium if it has been derived from selective reworking of the bedrock silts, especially in this small catchment where there is unlikely to be a drastic modification of the sediments supplied because of relatively low energy levels within these upstream tracts. Third, comparison of the coarse silt mineralogy between the alluvium and the bedrocks reveals there are significant differences between them. Although attempts to identify variations between these two populations using X-ray diffraction and infra-red spectrophotome- try proved unsuccessful (Burrin 1983a), analyses and comparisons of the heavy mineral suites (Allen 1949, 1967, Millar 1984) indicate that differences exist (Table 2.5). Statistical analyses undertaken here of the heavy mineral suites present in the valley fill coarse silts from units 2-4 at Robertsbridge indicate that these sediments contain significantly higher inclusions of the semi-stable mineral staurolite, and the less stable minerals epidote and hornblende. Chlorite was also found in two of the alluvial samples. Conversely, the bedrocks contain significantly higher percentages of the ultra-stable minerals zircon and garnet. This suggests that the alluvial coarse silts have suffered less weathering than the bedrock deposits and by implication are presumably younger. Conversely, the higher contents of stable minerals in the bedrocks indicate longer periods of weathering and diagenesis as would be expected in older rocks. The mineralogy of the alluvial coarse silts is not only different from the bedrocks but also compares favourably with that described for loess deposits from Kent (Catt et al. 1974) and Essex (Eden 1980). SEM micrographs also demonstrate that the alluvium is dominated by coarse silts (Burrin 1983b) with granular differences existing between the coarse silts of the alluvium, some of which show signs of wind transportation, and the silts of the Hastings Beds which are of a marine origin (Millar 1984). Weathering and reworking of the coarse silts found within the alluvium is also evident from SEM analyses (Burrin 1983b) for many of the coarse silt grains have adhering fine silt and clay particles, some of which are probably the result of inter-granular attrition and the weathering of coarser particles. These combined geomorphological, sedimentological and other considerations suggest that there is a loessic component in the Rother alluvial fills and, by extension of the argument, in other Wealden valleys (Burrin 1983b). It is also incongruous that 4m or more of loess can sit unconformably on the Chalk at Pegwell Bay, Kent (Pitcher et al. 1954) yet no similar materials can apparently be found in the Weald. That the presence of loess in the valley fills is difficult to identify is not surprising in that the sediments no longer form a composite mantle covering rocks which are very different in character. Rather, they have been extensively reworked with former weathered mantles (such as solifluction deposits) and soils to produce a heterogenous 'cover- loam' which has subsequently been further eroded and transported. Much of this cover- loam has evidently been removed from the valley sides, thereby explaining its apparent scarcity in contempor- ary soils as reported by the Soil Survey (Catt 1978), only to be redeposited within the valley bottoms as a type of 'river brickearth' . Hence, it is possible to produce a working hypothesis which can explain the nature of the valley fill and present-day channel sedimentology. A blanket of wind- blown dust (loess) perhaps of several metres thickness formerly covered the Rother catchment and probably much of the Weald (Fig. 2.15). This would have effectively sealed off the underlying bedrocks from continued erosion until sufficient loess had been removed by denudation. Loess is particularly susceptible to stream erosion and consequently would have been reworked by fluvial and other processes along with older weathered mantles and soliflucted deposits. This heterogeneous, dominantly fine-grained, sediment would have been transported by hillslope and fluvial processes, with some of it becoming deposited within the valley fills. Once sufficient loess had been removed from the interfluves, coarser bedrock materials would once again become available for transportation and depo- sition as is now found within the present Rother and other Wealden drainage networks. Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley Table 2.4 Moments of the Lower Cretaceous High Wealden Bedrocks M1 Sample Wadhurst Clay Tunbridge Wells Sand (arenaceous facies) Tunbridge Wells Sand (argillaceous facies) Ashdown beds (arenaceous facies) Ashdown Beds (argillaceous facies) (mean) 6.3-7.3 2.9-3.8 4.G 6.5 2.9-3.8 3.6-7.1 M2 (sorling) poor moderate-very poor poor moderate-very poor moderate-very poor M3 (skewness) positive-negatively skewed negatively skewed-symmetrical negatively skewed-symmetrical negatively skewed-symmetrical negatively skewed-symmetrical M4 (kurlosis) very leptokurtic platykurtic-mesokurtic mesokurtic-very leptokurtic platykurtic-mesokurtic platykurtic Two samples ofwadhurst Clay and two from the Ashdown Beds argillacaceous facies had open-ended distributions and consisted ofover 40% clay. Table 2.5 A comparison of non-opaque heavy mineral assemblages for the Rother valley Jill at Robertsbridge with catchment bedrocks (based on original data in Allen 1949,1967 and Millar 1984) Mineral ANATASE RUTILE ZIRCON TOURMALINE GARNET BIOTITE APATITE BROOKITE STAUROLITE KYANITE EPIDO TE HORNBLENDE SILLIMANITE HYPERSTHENE CHLORITE Figures in percentages Alluvium 1 2 3 Bedrocks 4 5 6 7 * = trace (10.1 Oh) 1: Unit 4 (0.5m); 2: Unit 3 (1.01~-2.0m) 3: Unit 3 (3.01~-3.5m); 4: Unit 2 (4.4m-4.6m) 5: Ashdown Beds; 6: Wadhurst Clay; 7: Tunbridge Wells Sand; 8: Hastings Beds (general) 8 Minerals in italic type have a statistically different content between assemblages as indicated by the Mann- Whitney U test (p=0.1) The origin and age of this WeaIden loess remain less clear. Catt (1977, 1978, 1985) has discussed the significance of loess in the British landscape and identified potential sources for this aeolian material (Fig. 2.15). Loess deposits in many parts of eastern England appear, on the similarity of their textures and mineralogy, to have been derived from glacial outwash and other detritus on the exposed floor of the North Sea and a similar source area is probable for the brickearths of Kent (Weir et al. 1971) and Essex (Eden 1980). This region is also supported as a potential source area by palaeoclimatological reconstruction (Lamb and Wood- roffe 1970, Catt 1977, Lill and Smalley 1978) and the fact that aeolian deposits fine towards the southwest (Perrin et al. 1974). However, there are also other possible sources for Wealden loess, including parts of the North European Plain where there are extensive coversands and loess (Zeuner 1959), and local sources such as the weathered products of Palaeogene strata (Smart et al. 1966). A Late Devensian age (c. 18 000 46 Paul J. Burrin DECREASE IN MODAL SlLT SIZE; INCREASE IN FLAKY SlLT MINERALS (MICA & CHLORITE) SlLT IN NORTHERN ENGLAND MAY BE DERIVED FROM LOCAL DEVENSIAN (WEICHSELIAN) GLACIAL MATERIALS (LILL & SMALLEY 1978) LOESS DERIVED FROM GLACIAL OUTWASH IN THE NORTH SEA BASIN (CATT 1977,1978) LOESS DERIVED FROM EUROPEAN GLACIAL DEPOSITS (LILL & SMALLEY 1978) Fig. 2.15 The distribution of loess in England and Wales (modified from Catt 1977). B.P.) for much of the loess in England and Wales is generally accepted (Catt 1977), although whether the Wealden loess is of this age is uncertain. These findings have implications for the development of Romney Marsh in that it is clear that sediment supply from the Rother catchment to the developing marshland was dominated by fine-grained, essentially silt-rich sediments. This suggests that coarser deposits found within the lower valley and extending out into the marshland are most unlikely to have been supplied by the fluvial system. Such deposits most probably reflect the influence of estuarine and shallow marine processes in the development of these lithostratigraphic sequences. It is also likely that some of the silts found within the lower Rother valley and the marshlands have been fluvially derived and contain a former loessic component, although these sediments have sub- sequently been redeposited within very different alluvial environments. Consequently, the original nature and character of these sediments is more difficult to identify, although there are similarities in the grain-size distributions of some of the silts and clays found in the lower valley and marshland (Waller et al. 1988) and those described here for the upper valley fills. The readily available supply of such fine grained sediments within the Rother catchment, together with rapidly rising sea levels, have combined to produce ideal conditions for rapid and prolonged episodes of sedimentation within the lower valley and out into the marshlands during the Holocene. Consequently, it is not surprising that significant thicknesses of materials have accumulated within both these areas. However, shallower, but nevertheless sizeable, fills have also accumulated in the fluviatile sectors of the valley beyond the range of the marine influence. This demonstrates that other causal mechanisms have been responsible for alluviation here and it is on these that attention is now focused. "'len and 'lant Macrofossil Investigations b Robert Scaife The predominantly inorganic nature of the fills inland from Bodiam has largely excluded the use of radiometric analyses to provide an absolute chronology, although samples of the organic deposits from Robertsbridge await radiocarbon dating. Consequently, palynological analyses have been employed in an effort to gain some insight into the palaeoenvironmental conditions and age of the deposits in the upper Rother. These results, together with archaeological and historical evidence, can also be used to provide some indication as to the probable causes and rates of valley fill sedimentation. Methods and Results Standard pollen analyses were undertaken on a core Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley from the Mayfield (Rl) site (Fig. 2.3) and the organic peaty clays interbedded within unit 1 at Robertsbridge (R1 l, Fig. 2.5). Samples of 3-4 grams were subjected to normal pollen preparation procedures (Faegri and Iversen 1974, Moore and Webb 1978) although micromesh (lop) sieving was also undertaken in order to remove a remaining clay residue. Absolute pollen frequencies were calculated by using the addition of known quantities of exotic pollen (Garrya) and were found to be generally high throughout, ranging from 4 7 the analysis presented here (for reviews see Scaife 1980, 1982, 1987; Waton 1982). These studies have, however, tended to concentrate upon valley mire organic peat sequences. Pollen analyses of inorganic valley fill sequences in Sussex, although problematical in their interpretation, have yielded valuable information (Scaife and Burrin 1983, 1985). The apparent lack of pollen in the inorganic deposits at Mayfield is disappointing in that it has been shown (Scaife and Burrin 1983) that pollen data can be derived from such 60,000 to 370,000 pollen grains per gram. A sum of 400 pollen grains was counted at each level (excepting the uppermost and lowest levels where 300 and 200 repectively were counted) excluding modern spores and Lower Cretaceous palynomorphs. These data are presented in Figs. 2.16 and 2.17 where pollen has been calculated as a percentage of their total at each level and spores as a percentage of the sum of total pollen plus spores. The pollen spectra of the organic clays at Robertsbridge are dominated throughout by arboreal and shrub pollen in approximately equal proportions (45% TP). Pollen of herbs is present but in much lower proportions from 1.8% TP at 795cm to 13.5% TP at 765cm. Arboreal pollen is dominated by Tilia, Quercus and Alnus, with lesser quantities of Betula, Pinus, Ulmus and Fraxinus. Pollen of Tilia and Fraxinus are likely to be under-represented in the pollen sum because of their comparatively poor pollen production and dispersal with respect to the other arboreal taxa present. Examination of well preserved Corylus type has shown that this group, although possibly containing Myrica gale, is dominated by Corylus. This is substantiated by the finding of macrofossils. Lesser quantities of rosaceous taxa, Hedera, Salix and Viburnum are present and are also regarded as being under-represented in the pollen spectra for reasons of entomophily and lesser pollen production. Herbaceous pollen are relatively diverse although percentages are low, with most taxa occurring only sporadically. Plantago lanceolata (to 1% TP), Gramineae (to 15% TP) and cereal type (to 1.5% TP) are the predominant types. Spores of Pteridium, Dryopteris type (to 15% TP+ spores) and Polypodium are consistent. Considerable numbers of Lower Cretaceous paly- nomorphs were also recorded, attaining their highest values of 22-33% TP + palynomorphs between 745-750cm. Pollen is poorly preserved in the samples taken from the Mayfield section. At present satisfactory results have only been obtained from the basal lOcm of the profile (3.8m). First indications show that the pollen spectra of these levels are broadly similar to the organic clays at Robertsbridge and have a mid-Holocene assemblage comprised of mixed deciduous woodland taxa, including Quercus, Tilia, and Alnus. Pollen Stratigraphy There are now a substantial number of pollen spectra available from southern England with which to compare materials. The reason for the deficiency at R1 may reflect oxidation and destruction of the former pollen content or be the result of very rapid sedimentation of inorganic deposits at this site. The sequence from Robertsbridge falls between the two extremes of organogenic accumulation and inorganic sedimentation. The relatively high organic content here has resulted in higher absolute pollen frequencies than found within the inorganic fills at both Mayfield and in other Wealden valleys (Scaife and Burrin 1985). The relatively high silica content and the presence of Lower Cretaceous spores and pollen (e.g. Cicatricosisoorites brevilaesuratus, Classopollis torosus, Con- cauisporites s@., Todisporites sp., Abietinaepollenites sp., and Paruisaccites radiatus) attests a substantial secondary input from surrounding lithologies. These Cretaceous inclusions demonstrate the significant amount of reworking of the loessic sediments with weathered derivatives from the local bedrocks. Nevertheless, the generally fine preservation of the Holocene pollen indicates that the greater proportion of the pollen is undoubtedly of sound stratigraphical position. The dominance of Corylus pollen (occasionally in clusters) and its macrofossil presence in these deposits suggests that hazel was present in extremely close proximity and possibly growing on the floodplain in damp but open woodland conditions. Alnus might also be expected to be growing in such floodplain environments as found in other past and contemporary situations in southern England (Scaife 1980, 1982), and as indicated at 3.8m at Mayfield. The relatively high values of alder at R1 1 (up to 19% TP at 750cm) illustrate this to some extent but pollen percentages are usually considerably higher when alder carr woodland is locally dominant. Thus, it is more likely that Alnus was either only sporadically present locally, or that the pollen represents more substantial growth at some distance from the site. Alnus pollen is readily transported aerially but consideration must also be given to pollen derived from floodwaters during overbank flows (Peck 1973). Salix may also have been a floodplain constituent here or upstream. Hence, the vegetation community appears to have been dominated by Coylus, with Alnus and Salix, located within a relatively dry floodplain habitat. This view is further supported by the presence of Humulus type pollen and the macrofossil evidence of Corylus nuts, leaves and twigs (currently being studied). The significance of Tilia and its subsequent decline in South-East England during the middle Holocene has long been known (Godwin 1940, Birks et al. 1975) and 48 Paul J. Burrin numerous pollen spectra now illustrate the generality of this phenomenon (Scaife 1982). This has also been found in the analysis of other Sussex floodplain peats and inorganic deposits (Thorley 197 1, Brooks 1983, Scaife and Burrin 1983, 1985, Waller et al. 1988). At Robertsbridge, and probably for the period represented by the basal part of the sequence at Mayfield, similar domination of the interfluves by Tilia also occurred. From the relatively high pollen percentages seen here (R1 1) it can be suggested that it formed dominant or at least CO-dominant woodland with Quercus and Fraxinus close to the floodplain. Pollen analyses of peat sequences adjacent to, or on, Lower Cretaceous sandstone lithologies elsewhere (Scaife 1980) have shown that soils developed on these bedrocks were ideally suited to its growth and possibly dominance. Conversely, the low frequencies of Betula and herbs, (notably Gramineae) indicate the relative absence of non-wooded areas either of natural or anthropogenic origin at least close to this site. The presence of Plantago lanceolata, cereal type and Gramineae are evident and indicate the probable impact of anthropogenic activity within the pollen or fluvial catchment. The minerogenic nature of the bulk of the valley fill sediments and similar lithostratigraphy to sites in other Sussex valleys where palynological research has demonstrated the role of man in causing or exacerbating sediment inwash to valley floors, suggests that the deposits may have resulted in response to similar causes, i.e. anthropogenic activity within the catchment by prehistoric societies. The finding of burnt wood remains in the valley fill at Mayfield (section R2) suggests that fire may have been utilised to assist the clearance of the vegetation cover. Similarly, the presence of charcoal in the organic sequence at Robertsbridge also attests the use of fire by these prehistoric communities. This site therefore, along with that at Mayfield (Rl) currently under investigation, Sharpsbridge in the Ouse valley (Burrin and Scaife 1984) and Stream Farm in the Cuckmere valley (Scaife and Burrin 1985) have yielded pollen spectra which show that considerable inorganic sediment accumulation took place during the middle to late Holocene (F 111). This appears to have been in response to prehistoric anthropogenic processes with resultant valley side erosion and sediment inwash to the valley bottoms. These findings demonstrate that the previous and long-held belief that the Central Weald was covered by virgin forest and was avoided as such by prehistoric man would appear to be totally unfounded. The fossil pollen record and inorganic inland fill stratigraphies found throughout the Ouse, Cuckmere and Rother valleys testify to at least the eastern High Weald being utilised by prehistoric man in a manner that caused serious environmental degradation and which has hitherto been largely unrecognised (see also Scaife and Burrin 1985). The ages of these fill sequences have yet to be determined by radiocarbon dating and the results of such tests for the plant macrofossils are awaited. While pollen analysis is no longer accepted as a dating technique in late Quaternary studies there are sufficient data now available to state that the pollen sequence described here is at its earliest of middle Holocene (Atlantic) date. This probability is based upon the prevalence of mixed deciduous woodland and/or dominance of Tilia, which did not decline significantly in this part of the Weald until c. 3700 B.P. (Waller et al. 1988). Furthermore, the presence of cereal pollen (a single grain at 800cm but consistently from 775cm) and Plantago lanceolata throughout, allows a Neolithic (c. 5500 B.P.) or post-Neolithic date to be postulated. Although some current debate has been focused upon the presence of pre-Ulmus decline cereal pollen (Edwards and Hirons 1984), such occurrences have been sporadic where they have been thought to occur. Here, continuous cereal pollen upwards from 775cm to 745cm may be regarded as resulting from Neolithic or post-Neolithic agricultural activities. This is almost certainly corroborated by the relatively low Ulmus pollen frequencies recorded here. Pollen percentages of Ulmus prior to its decline (at 5040 f 80 B.P. in the Brede tributary - see Waller et al. 1988) usually attain values of 10-15% throughout southern England during the Atlantic (F 11) (Scaife, in press). Here, Ulmus values are significantly lower throughout, suggesting a post-Ulmus decline date for the basal pollen spectra (Figs. 2.16 and 2.17). It can be concluded therefore from the pollen data of the relatively organic sequence at Robertsbridge, that the local environment was characterised by a mixed deciduous woodland environment (Quercus, Tilia, Fraxinus and some Ulmus) with areas of dominant Tilia on the interfluves. The floodplain ecosystem, was dominated by Corylus but with areas of Alnus and Salix (carr woodland) of damp character (but dry in comparison to more usual valley mire situations). From the presence of woodland of this character and of some anthropogenic indicators it can be suggested that the date of this 55cm organic clay sequence can be attributed to a Neolithic or later phase. The pollen from the basal levels at Mayfield would appear to indicate a similar age. It is also pertinant to note that the organic materials at Robertsbridge overlie c. 2m ofinorganic deposits. Whilst the age of these underlying sediments remains enigmatic, they pre-date the organic inclusions and are therefore presumably of a Neolithic or earlier (Mesolithic?) date. Sedimentation Rates and River Channel Changes The probable age of the organic materials within the valley fill allows some tentative conclusions as to valley fill sedimentation rates at this site. It is evident from the stratigraphy that at least 7.45m of sediment overlie the organic deposit which appears to date to the Neolithic period (c. 5000 B.P.) . Assuming this to be the earliest age for this inorganic accumulation and near constant sedimentation, then a rate of 1.04mm per year has l 790 - T + -- - nno- -. i 0 0 0 1- CM nnnnnnnnnnnn UUUU 30 %Total Pollen 6 0 0 0 0 0 0 Fig. 2.16 Palynology of the peaty clay at Robertsbridge (RII): 0 0 0 0 trees and shrubs. uuuuuuuuuuuu p0u501;loo,U , y o u I . . . . . . . . . I 0 0 0 0 0 0 0 0 0 0 1 0 0 0 % Total Pollen -1- 40 %Pollen+ Spores -1- % Pollen + Palynomorphs-~- +Unident. 0 50 % Total Pollen Fig. 2.17 Palynology of the peaty clay at Robertsbridge (RII): herbs and summary. U 100 -1 0 500 1000 Pollen/gram 50 Paul 3. Burrin occurred at this site. This, however, may prove to be a serious underestimation of the actual rates of accumul- ation for there is evidence from the stratigraphy that sedimentation was not constant but was interrupted by episodes of incision and erosion. It is also possible to estimate sedimentation rates at other sites within the Rother valley because of artifacts buried within the alluvium. An archaeological excavation at Bodiam (Lemmon and Hill 1966) discovered a Romano-British site (TQ 783251) buried by floodplain sediments south of the river. The excavations revealed eight occupation levels, the lowest of which is 1.8m below the present ground surface or c. 0.33m O.D., which rested on a light bluish-grey clay in which an ebony-black trunk or plank was located at c. 2.3m. Also found embedded in this clay were oak and other charcoals, as well as macro-remains of yew, alder and oak. If this Romano-British occupation level is extrapolated a few metres downstream to the line of the cross-section presented here (Fig. 2.6), the stratigraphy appears very similar to that described above. Artifacts suggest an occupation date 0fA.D. 100 which, assuming constant sedimentation, indicates an alluviation rate of 0.95mm per year since that date. It is also interesting to note that Lemmon and Hill (1966) suggest that the Rother valley at this site in the Roman period could not have been an open estuary or a significant tidal channel because the occupation levels would have been prone to flooding; rather, a navigable river with firm banks, subject only to tidal variations of current is preferred. If so, the laminated deposit located in the centre of the valley at Bodiam (Fig. 2.6) may indicate the position of the Rother at this time, for such laminated sediments are more likely to have been deposited in a relatively low energy environment subject only to slight variations in flow velocities. A further point of interest raised by these authors concerns the apparent relationship ofBodiam Castle to the river. The castle was built c. 1386 at which time "it was provided with a dock which could be entered at high tide by vessels offour feet draught. This dock is now 25ft O.D. (Liverpool)" (Lemmon and Hill 1966, 101). This elevation is 4-5m above the floodplain surface which, assuming the elevation of the dock is correct, suggests it must have been located well above the elevation of both the then floodplain and the channel. It also suggests the Rother was approximately in its present position in order to allow the vessels to reach the dock. This contrasts with the Small Hythe facilities which were apparently used by the fleet of Henry V111 and which are located at or about the present floodplain level. The use of Small Hythe as a port or dock at this time is in accordance with knowledge concerning the route used by the Rother in the lower valley in historic times (Rendel 1962, Eddison 1985). Some indication as to sedimentation rates and valley development in the Rother can also be gained from consideration of the industrial archaeology associated with former iron-working in the Weald (Straker 1931, Cleere and Crossley 1985). Although there is evidence of iron working from the Iron Age and Romano-British periods, the peak of this activity occurred in the mid- sixteenth century (Straker 193 1, Hammersley 1973). Two significant residual components from this industrial period remain prominent: reservoirs or hammer ponds and waste products. Dams or bays were built at a number of sites in the Rother valley (Fig. 2.1) in order to create small reservoirs or hammer ponds (see Cleere and Crossley 1985 for a complete list). Many have long since been drained and hence it is theoretically possible to obtain some indication as to sedimentation rates by considering the thicknesses of materials behind the bays. Straker (1931) indicated that 3m or more of sediment had been proven behind some dams, but this almost certainly includes the underlying valley fill thickness. A recent study (Burrin 1985) has discussed some of the problems in attempting to estimate valley sedimentation rates (or catchment denudation rates) utilising data derived from hammer pond sites and these difficulties need to be resolved before reliable figures can be be obtained. The waste materials, however, appear to be of greater value in calculating historic rates of alluviation. Cinder and slag waste deposits from the furnaces and forges are very durable materials and are easily recognisable in the field. Typically, forge cinder is blue- black in colour, rounded or globular in shape, with a very smooth, hard surface, a specific gravity of 2.5-3.8 and usually has a high iron content. Blast furnace slag often appears much like bottle glass and bears a resemblance to obsidian. It exhibits black, blue or green colours and is often vesicular. The solid forms have a specific gravity of 2.8-3.0, while the vesicular variety is lighter, more brittle and has a lower iron content. The location of these materials with repect to the valley fills can provide some indication as to sedimentation rates at selected sites. These materials have not been found within the valley fill deposits in the upper valley between Mayfield and Robertsbridge. Rather, they are usually located either within the channel (within the banks or strewn along the floor), or on the valley floor. For example, the floodplain surface to the north of the river at Bivelham (R3) is littered with this debris and similar oocurrences can be found in some of the tributary valleys. This suggests relatively little floodplain sedimentation at these locations in the upper valley. Furthermore, the sedimentation rate calculated above for the Roberts- bridge site may be further under-estimated if the Rother floodplain surface has altered little in elevation since Tudor times. These findings can be contrasted with sedimentation in the lower valley at Bodiam (R12) where slag was found in the floodplain sediments at a depth of 0.45m in borehole 2 (Fig. 2.6). Assuming the slag dates from the mid-sixteenth century then a sedimentation rate of c. 1.lmm per year has occurred here. It appears, therefore, that floodplain sediment- ation within the inland valley sectors had apparently ceased or become of limited significance away from the hammer ponds by the Tudor period, but in the Holocene Floodplain and Alluvial Fill Deposits of the Rother Valley downstream valley tracts sedimentation was continuing well beyond this time (see also Eddison 1988). Similar conclusions regarding valley fill development away from the coastal zones have also been drawn from research in the neighbouring Ouse and Cuckmere valleys (Burrin 1983b). Conclusions This paper has described aspects of the geomorphology of the Rother valley and its alluvial deposits from Bodiam inland to the headwater areas. The valley fill lithostratigraphy has been investigated and two major sedimentary associations defined: (a) a shallower inland association which extends as far downstream as Bodiam, where it is replaced by (b) a thicker and more complex perimarine or near-coastal association. The geometry of the inland association has been examined and a number of relationships described between catchment character- istics and parameters defining the scale of the fills. A contrast has been made between the present channel and the valley fill sedimentology. The alluvium is frequently dominated by coarse silts and it has been suggested from geomorphological, sedimentological and other considerations that the alluvium contains a reworked loessic component. The significance of this reworking is reflected in both the sedimentology of the alluvial fills and the pollen spectra which contain high values of Lower Cretaceous palynomorphs. These 5 1 findings indicate the Rother was supplying essentially fine-grained material to the marshland during the Holocene. Palynological, archaeological and historical evidence provides some indication as to the age of the fills and allows estimations of alluviation rates to be calculated. The inland sequences as evidenced from the Mayfield (RI) and Robertsbridge (R1 1) sites have probably developed largely as a response to the impact of prehistoric cultures within the valley. The clearance of the former forest cover exacerbated soil erosion and increased sediment loads within the fluvial system which aided valley fill accumulation. This appears to have been particularly prominent since the Neolithic period. The inland fills seem to have developed largely by the 16th century, but sedimentation in the lower valley and the marshlands appears to have continued throughout the historic period and into modern times. Acknowledgements Thanks are extended to all those who have helped with this project including David Lawes for his painstaking patience in drawing and modifying the diagrams, as well as Simon Wood, Steve Brown and a host of students for assistance with fieldwork. The research was funded by grants from the L.S.B. Leakey Memorial Trust and Goldsmiths' College. References Allen, P. 1949: Wealden Petrology: The top Ashdown Pebble Bed and the top Ashdown Sandstone. Quart. 31. geol. Soc. 104, 257321. Allen, P. 1967: Origin of the Hastings Facies in NW Europe. Proc. Geol. Ass. 78, 27-105. Bhowmik, N. G. 1984: The hydraulic geometry of floodplains. J. Hydrol. 68, 369-40 1. Birks, H. J. B., Deacon, J. and Peglar, S. 1975: Pollen maps for the British Isles 5000 years ago. Proc. R. SOC. Lond. B189, 87-105. Bridge, J. S. and Leeder, M. R. 1979: A simulation model for alluvial stratigraphy. Sedimentology 26, 616-644. Bristow, C. R. and Bazley, R. A. 1972: Geology of the County uroundRoya1 Tunbridge Wells. (Mem. geol. Surv.). Brooks, A. 1983: In Sussext Enuironment, Landscape and Society, (University of Sussex) 118. Brooks, N. P. 1981: Romney Marsh in the early Middle Ages. In: Rowley, T. (editor) The evolution of marshland landscapes. (Oxford) 74-94. Burrin, P. J. 1981: Loess in the Weald. Proc. Geol. Assoc. 92, 87-92. Burrin, P. J. 1982: The coastal deposits of the southern Weald. Quaternary Newsletter 38, 16-24. Burrin, P. J. 1983a: The character and evolution ofJloodplains with speciJic reference to the Ouse and Cuckmere, Sussex. (Unpublished Ph.D. Thesis, University of London). Burrin, P. J. 1983b: The nature of Wealden floodplain alluvium and some possible implications. Geologisches Jahrbuch A 7 1, 237-263. Burrin, P. J. 1985: Holocene alluviation in south-east England and some implications for palaeohydrological studies. Earth Sufice Processes and Landforms 10, 257-271. Burrin, P. J. and Jones, D. K. C. inpress: Environmental processes and fluvial responses in a small temperate zone catchment: a case study of the Sussex Ouse Valley, South-East England. In Gregory, K. J., Thornes, J. B. and Starkel, L. (editors) Fluvial Processes in the Temperate zone during the last 15,000 years (Wiley) . Burrin, P. J. and Scaife, R. G. 1984: Aspects of Holocene valley sedimentation and floodplain development in southern England. Proc. Geol. Ass. 95, 81-96. Catt, J. A. 1977: Loess and Coversands. In Shotton, F.W. (editor) British Quaternav Studies, Recent Advances (Oxford) 222-229. Catt, J. A. 1978: Thecontribution ofloess tosoilsinlowland Britain. In Limbrey, S. and Evans, J. S. (editors) The effect of man on the landscape: the lowland zone. CBA Res. Rep. 2 1, 12-20. Catt, J. A. 1985: Soil particle size distribution and mineralogy as indicators of pedogenic and geomorphic history: examples from the loessial soils ofEngland and Wales. In Richards, K. S., Arnett, R. R. and Ellis, S., Geomorphology and Soils (Allen and Unwin) 202-2 18. Catt, J. A., Weir, A. H. and Madgett, P. A. 1974: The loess of eastern Yorkshire and Lincolnshire. Proc. Yorks. Geol. Soc. 40, 23-39. Cleere, H. and Crossley, D. 1985: Wealden Iron-making. (Leicester) Dury, G. H. 1964: Subsurface exploration and chronology of underfit streams. US Geol. Surv. Prof. Paper 452-B. Eden, D. N. 1980: The loess of north-east Essex, England. Boreas 9, 165-177. Edwards, K. J. and Hirons, K. R. 1984: Cereal pollen grains in pre-elm decline deposits: implications for the earliest agriculture in Britain and Ireland. Journ. Archaeol. Sci. 1 1, 7 1-80. Eddison, J. 1985: Developments in the lower Rother valleys up to 1600. Arch. Cant. 102, 95-1 10. Eddison, J. 1988: 'Drowned Lands': Changes in the course of the Rother and in its estuary and associated drainage problems, 163551737, In this volume, chapter 12. Faegri, K. and Iversen, J. 1974: Textbook of pollen analysis. (Oxford, Blackwell). 52 Paul J. Burrin Folk, R. L. and Ward, W. 1957: Brazos river bar: a study in the significance of grain size parameters. 3. Sedim. Petrol. 27, 3- 26. Gallois, R. W. 1982: Loess in the Weald. Proc. Geol. Assoc. 93, 316. Godwin, H. 1940: Pollen analysis and the forest history ofEngland and Wales. New Phytol. 39, 370400. Green, R. D. 1968: Soils of Romney Marsh. Soil Survey Gt. Britain, Bull. 4. (Harpenden). Hammersley G. 1973: The charcoal iron industry and its fuel 1500-1 750. Econ. Hist. Rev. 26, 593413. Jones, D. K. C. 1981: Southeast and Southem England. (Methuen). Kirkaldy, J. F. and Bull, A. J. 1940: The geomorphology ofthe rivers of the southern Weald. Proc. Geol. Ass. 51, 8-149. Lamb, H. H. and Woodroffe, W. A. 1970: Atmospheric circulation during the last Ice Age. Quat. Res. 1, 29-58. Lemmon, C. H. and Hill, J. D. 1966: The Romano-British Site at Bodiam. Sussex Archaeol. Coll. 104, 86-102. Lill, G. 0. and Smalley, I. J. 1978: Distribution ofloess in Britain. PTOC. Geol. Assoc. 89, 5745. Millar, M. R. 1984: A heavy mineral study of thejne-grained sedimenfsfrom Flandrian alluvium in the Weald (Unpublished MSc. Thesis, Polytechnic of North London). Moore, P. D. and Webb, J. A. 1978: An illustratedguide to pollen analysis (Hodder and Stoughton). Peck, R. M. 1973: Pollen budget studies in a small Yorkshire catchment. In Birks, H. J. B. and West, R. G. (editors) Quatema~y Plant Ecology, (Oxford, Blackwell) 43-60. Pcrrin, R. M. S., Davies, H. and Fysh, M. D. 1974: Distribution oflate Pleistocene aeolian deposits in eastern and southern England. Nature 248, 3204. Pitcher, W. S., Shearman, D. J. and Pugh, D. C. 1954: The loess of Pegwell Bay and its associated frost soils. Geol. Mag. 91, 308-14. Rendel, W. V. 1962: Changes in the course of the Rother. Arch. Cant. 77, 63-76. Scaife, R. G. 1980: Late Devensian and Flandrian palaeoecological studies in the Isle of Wight. (Unpublished Ph.D. Thesis, University of London). Scaife, R. G. 1982: Late-Devensian and early Flandrian vegetation changes in southern England. In Bell, M. and Limbrey, S. (editors) Archaeological Aspects of Woodland Ecology, BAR International Series 146, 57-74. Scaife, R. G. 1987: A review oflater Quaternary plant microfossil and macrofossil research in southern England; with special reference to environmental archaeological evidence. In Keeley, H. C. M. (editor), Environmental Archa~ology: A Regional Review, vol. II (English Heritage). Scaife, R. G. and Burrin, P. J. 1983: Floodplain development in and the Vegetational History of the Sussex High Weald and some Archaeological Implications. Sussex Archaeol. Coll. 12 1, 1-10. Scaife, R. G. and Burrin, P. J. 1985: The Environmental Impact of Prehistoric Man as Recorded in the Upper Cuckmere Valley at Stream Farm, Chiddingly. Sussex Archaeol. Coll. 123, 27-34. Shephard-Thorn, E. R., Smart, J. G. O., Bisson, G., and Edmonds, E. A. 1966: The geolo~ of the country around Tenterden. (Mem. geol. SUN.). Smart, J. G. O., Bisson, G. and Worssam, B. C. 1966: Thegeology ofthe country around Canterbury and Folkestone. (Mem. geol. Surv.). Straker, E. 1931: Wealden Iron. (London, Bell). Thorley, A. 1971: Vegetational history in the Vale of the Brooks. In Williams, R. B. G. (editor) Guide to Sussex Excursions (Inst. Brit. Geogr. Spec. Publ.) 47-50. Waller, M., Burrin, P. J. and Marlow, A. 1988: Flandrian sedimentation and palaeoenvironments in Pett Level, the Brede and lower Rother valleys and Walland Marsh. In this volume, chapter 1. Waton, P. V. 1982: Man's impact on the chalklands: some new evidence. In Bell, M. and Limbry, S. (editors), Archaeological aspects of woodland ecology, BAR International Series 146, 75-91. Weir, A. H., Catt, J. A. and Madgett, P. A. 1971: Post-Glacial soil formation in the loess of Pegwell Bay, Kent. Geoderma 5, 13149. Wolman, M. G. and Leopold, L. B. 1957: River floodplains: some observations on their formation. US Geol. Suru. Pro$ Paper 283-C. Zeuner, F. E. 1959: The Pleistocene Period. (Hutchinson).