'Scouring the Conduit Head at Woodnesborough' Investigations into Convent Well near Sandwich

127 ‘Scouring the Conduit Head at Woodnesborough’: Investigations into Convent Well, near Sandwich keith parfitt and helen clarke While working on the publication of twentieth-century excavations at Sandwich’s Carmelite Friary (Whitefriars), the authors decided to test the accuracy of a document referring to a spring which was acquired by the friary in 1306. The record states that Thomas Shelving, a Sandwich wool merchant and holder of a manorial estate in the nearby parish of Woodnesborough, assigned to the prior and Carmelite friars of Sandwich a plot of land in Woodnesborough, 12ft x 12ft, with a spring there, to enclose it and make an underground conduit through his lands to their house (Calendar of Patent Rolls 1301-1307, 440). A site marked as Convent Well on large-scale Ordnance Survey maps of the Sandwich region since the nineteenth century (Figs 1 and 2) was the starting point. It lies some 600m north-east of Woodnesborough parish church, at the foot of a ridge of high ground running towards Marshborough. Members of the Dover Archaeological Group (DAG) visited the site in March 2014 but found no surface traces of a well, not even the presence of damp ground, although two loose pieces of dressed stone were discovered in the undergrowth nearby. Local enquiries provided little further information beyond the fact that the site was known locally as Conduit Well. Despite this unpromising start, the current landowners were enthusiastic about finding the lost medieval feature, and invited DAG to return and investigate the site. This was carried out between March and June 2014 and excavation was rewarded by the discovery of the remains of a small medieval building buried in a field bank, with portions of its walls still standing to a height of more than 1m (Parfitt 2014). Inside the little structure was a well-shaft, capped-off but intact and still containing water from a spring that filled it from the bottom. This was Convent or Conduit Well itself, in fact the conduit-head of Sandwich Whitefriars’ medieval water supply, the starting point for the conduit that carried water to the friary at Sandwich for more than two hundred years, and then continued to provide drinking water to part of the town for further centuries. It consisted of a well-house, standing above a stone-lined shaft sunk to capture water from a free flowing spring issuing from the base of the ridge. The lower part of the shaft served as a cistern and settling-tank to hold the water there until it reached a certain depth, when it could flow through a hole in the wall of the well-shaft into a conduit which led the KEITH PARFITT AND HELEN CLARKE 128 Fig. 1 Ordnance Survey map of 1877 showing position of Convent Well in relation to Sandwich town, South Poulders Stream and The Delf. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 129 Fig. 2 Detail of 25 inch Ordnance Survey map for 1873, showing position of Convent Well (centre, top) in relation to Woodnesborough Parish church and Shelving Court (both bottom left). KEITH PARFITT AND HELEN CLARKE 130 water some 2km to a conduit-house in the friary. From there the water would have been piped on to where it was needed in the precinct (Fig. 3). The well was protected by a small stone structure, originally roofed and looking like a tiny house. A few examples of these small structures still survive in the grounds of medieval monasteries, and are called either spring-houses or wellhouses. In this case, ‘well-house’ seems more appropriate and that is what the structure protecting the well-shaft will be called in this paper. Fig. 3 Schematic diagram showing the arrangement of a typical monastic water supply system (after Bond 1993). ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 131 the investigations At about 10m aod and overlooking Sandwich and the South Poulders marshes, the site (NGR TR 31090 57264) is situated in the western corner of a large arable field and at the bottom of an overgrown lynchet bank which on its south-west side is almost 2m in height (Fig. 2). Along the north-west side, the bank slopes down steeply as the ground level falls away. The top of the bank is followed by a long-established trackway running between Woodnesborough and Each End at Ash. The natural geology of the area consists of silty clay, belonging to the Thanet Formation. Cartographic and field evidence indicates that the medieval well had been sunk where a natural spring issued from the base of the hill. Archaeological trenching in the area showed no signs of a pond or mere around the mouth of the spring, but it uncovered evidence of activity long before the medieval period. This took the form of a surface scatter of prehistoric struck flints and a Neolithic polished stone axe, and a Bronze Age pit radiocarbon dated to 1611-1442 Cal bc (Univ. of Belfast Accelerator Lab., 25926) excavated in the field immediately to the east of Convent Well. The lynchet bank, into which the medieval well-house had been set, probably represents plough soil which had accumulated at the foot of a long, north-east facing slope. There is another such bank defining the upper edge of the adjacent field about 100m further up the slope to the south-west. The implication must be that the bank had developed sometime prior to the erection of the well-house but the natural spring here must have been in existence long before the region came under the plough. Water issuing from this and two or three other springs in the area would originally have fed into a stream that flowed towards the coast. Today, this is represented by the South Poulders Stream which runs north-eastward from the well site. Still important for local drainage, the stream traverses the marshlands of the South Poulders, skirting the northern edge of Mary le Bone Hill, (on which may have stood a thirteenth-century chapel (Ogilvie 1960)) which blocks a direct route to Sandwich. Currently, the stream enters Sandwich through a culvert under the medieval town rampart at the Butts, where it connects with the Delf Stream and finally empties into the main River Stour (Fordham and Green 197 3, 102). Convent Well: the conduit-head for Sandwich Whitefriars The remains of the well-shaft (Figs 4, 5 and 6) were solid and showed no evidence of movement or subsidence even though it must have been difficult to build on such a wet site. Excavation showed that the structure had been set into a wide construction pit, cut back into the existing lynchet bank around the mouth of the spring. Apart from obvious mid/late twentieth-century modifications (see below), the well-shaft and its superstructure appeared to be of one build, with no evidence of any subsequent repairs or alterations. There was nothing to suggest that it was any later in date than the fourteenth century, as implied by the documentary evidence, but the only physical evidence for that date is a single fourteenth-century potsherd found in the well’s construction-pit (see below). KEITH PARFITT AND HELEN CLARKE 132 Internally, the well-shaft itself was 0.84m square (2ft 9in.) and over 1.60m (5ft 3in.) deep. More than 1m of fill had accumulated in its lower part, but excavation showed that very little medieval material was present. It was not possible to excavate the shaft right down to its base due to the amount of water flowing in through the basal cobbles, but it is unlikely to have been much deeper than the lowest point reached. The shaft had been very carefully constructed, its builders presumably working in constantly flowing water at the lower levels. At the bottom, layers of laid flint cobbles set in light grey silty clay provided a foundation for four rough, large, horizontal slabs of undressed ragstone, one along each side of a square to form a solid base for the stone shaft above. They were c.0.10m thick with their top at c.8.72m aod. Resting on the ragstone slabs, the main well-lining consisted of five courses of neatly cut and laid blocks of Caen stone, bonded with grey clay and reaching a height of c.1.20m (9.95m aod). There was little sign of water erosion, and also no indication that water-inlets had been cut through the blocks, so the well must have been filled by water from the spring flowing through the cobbles at the base of the shaft. A small area exposed on the south-eastern side indicated that the top of the shaft had been packed around externally with interleaved layers of clay and chalk and Caen-stone rubble (Fig. 5, Contexts 15, 16) before being capped with further large, horizontal slabs of ragstone, very similar to those used at the base. Internally, the Fig. 4 General site plan of the excavated well ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 133 Fig. 5 Section through the well. KEITH PARFITT AND HELEN CLARKE 134 upper slabs overhung the sides of the shaft by between 0.03 and 0.05m on the three surviving sides, and in themselves provided a solid base upon which the walls of the well-house were set. At the same level and extending back from them towards the edges of the construction pit, flint cobbles set in clay (Fig. 5, Contexts 14, 25) would have provided a firm external surface for the builders of the superstructure to work from. On the south-eastern side, these cobbles had been surfaced with a thin layer of crushed chalk (Context 13). Although there were no water-inlet holes in the sides of the well-shaft, the damaged north-east wall of the shaft showed clear evidence of an original outlet. About halfway up the shaft wall, a Caen-stone block set slightly east of centre, was drilled with a roughly-shaped circular hole about 0.10m in diameter. The base of the hole lay at about 0.70m below the top of the shaft, giving an invert level of 9.29m aod for this water-outlet (Fig. 7). The adjacent block to the north-west may also have been provided with a similar aperture, although this stone had later been removed to allow the insertion of a modern drainage pipe (see below). Nevertheless, a loose block of the appropriate size and shape to fit the gap was found in the filling of the trench associated with the modern pipe. This block contained a similar sized aperture, roughly chiseled through one edge, leaving little doubt that it was the missing stone from the shaft wall (Fig. 8). Without doubt, these two perforated Fig. 6 The well-shaft fully excavated, looking south-west. Scale, 50cm. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 135 stones represented the original outlet for the accumulated spring water which would have discharged into the conduit leading away to the friary. There was no surviving evidence to suggest that either hole had originally housed an outflow pipe of any sort. The presence of a thin, horizontal, ridge of accumulated ‘lime scale’ on all four walls was indicative of the medieval water-level within the shaft and must mark the level of settled water over many years. It occurred at 9.33m aod, almost halfway up the side of the one undisturbed outlet-hole. The top of the well-shaft had been protected by a small, square, masonry wellhouse (Fig. 5 and Figs 9, 10 and 11), contemporary with the well-shaft, measuring internally 1.11m across (3ft 8in.) and with the walls between 0.38 and 0.50m thick. Surviving up to a height of 1.35m (4ft 5in.), the walls were best preserved on the south-western side where they had been set into the lynchet bank. Forward of the bank, the north-east wall was completely missing, but a scar in the masonry of the north corner was just sufficient to indicate the former presence of a return wall here. It is most likely that there had not been a continuous wall on this side, but two short spurs extending from the corners as far as a central opening for access, probably at least 0.50m wide (cf. that at Beaulieu abbey, below). Fig. 7 Medieval water outlet hole and adjacent modern pipe, lo oking north-east. KEITH PARFITT AND HELEN CLARKE 136 Nothing remained of the well-house roof, but there was some evidence that it had been of stone; for example, a surviving, neatly-shaped, Caen-stone springer was set at 45 degrees into the western corner of the building at 1.13m (3ft 9in.) above floor level (Fig. 10). There was also the scar of a now missing one at a similar height in the south corner. In addition, a loose piece of Caen stone, probably originally from a window mullion but reshaped into another springer, was found in soil near the north corner, and two or three fragments of groin-vaulting were found in the fill of the well-shaft. The walls of the well-house itself had been founded on the large ragstone slabs surrounding the top of the well-shaft, which left a ledge 0.17-0.20m wide on three sides. This effectively formed the floor of the building, at 10.02m aod. Around the inside, the basal course of the well-house walls consisted of large Caen-stone blocks, very similar to those used in the shaft below but not extending through the thickness of the wall. Above and behind them, the main walls of the building were largely constructed by mortared flint cobbles, with some re-used Caen-stone ashlar (mostly in the south-west wall). The interiors of the three surviving corners of the building at the south, west and north were turned in yellow-pink medieval Fig. 8 Loose Caen-stone block with probable water outlet hole. Scale, 50cm. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 137 Fig. 9 View of the excavated well-house, looking south-west. Scale, 1 metre. brick. The external faces of the walls, however, were roughly finished, indicating that they were not intended to be exposed, probably because they would have been mostly enclosed by the lynchet bank. External facing bricks remained only KEITH PARFITT AND HELEN CLARKE 138 in the north corner, at the foot of the bank. Traces of rendering showed that all the interior wall surfaces of the well-house had originally been plastered (Fig. 10). Outside the north-west wall, the fill of the well’s construction-pit where it had Fig. 10 Detail of west corner of the well-house, showing plastered internal face and in situ springer. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 139 been cut into the existing lynchet bank (Context 22) produced a few fragments of medieval brick and peg-tile, small pieces of Caen stone and a single potsherd, broadly dated c.1300-1375 (pers. comm. Nigel Macpherson-Grant). Post-medieval maintenance and decay Over 1m of waterlogged grey silts and rubble (Fig. 5, Contexts 7, 23) filled the lower part of the well-shaft to just below the medieval outlet-hole. Excavation of these deposits yielded nothing of certain medieval date, apart from a few fragments of Caen stone derived from the superstructure of the building. The rubble included large quantities of nineteenth-century red brick, together with broken vessel glass, a tin can, soles of three shoes, and a decorated clay-pipe bowl broadly dated c.1880- 1930. The general impression was that this was deliberate infill, deposited during the second half of the nineteenth or early twentieth century, perhaps specifically intended to reduce the overall depth of the then open well-shaft. Significantly, however, the infill material had not been allowed to block the original water-outlet, showing that the spring water could still flow out as before. Other post-medieval activity at the site was indicated by a series of roughly laid ragstone blocks, boulders and several bricks resting on the north-western slope of the lynchet bank, by the north corner of the well-house (Fig. 4). After careful cleaning it could be seen that these represented the slumped and partially collapsed remains of a crude flight of steps, c.1m wide, that formerly led down to the well from the trackway running along the top of the bank (Fig. 11). Many of the stones showed signs of adhering mortar indicating that they had been re-used. Comparison of the mortar with that used in the well-house, however, showed that the mixes were completely different. The presence of two large maroon-red bricks, probably of late eighteenth- or early nineteenth-century date, implies a late post-medieval date for the steps, perhaps c.1800. Thus, although apparently not part of the original medieval arrangement, they suggest that access to the well was needed until relatively recent times. The soil sealing the steps contained much modern material indicating that they had been visible until the twentieth-century sealing of the site (see below). Although a little rubble, apparently from the superstructure of the well-house, was discovered within the well-shaft and in deposits surrounding the building itself, it was not sufficient in quantity to represent all the missing portions of the structure. Thus, it would seem that a certain amount of original material must have been removed and taken elsewhere. Whether this was the result of regular clearance of the well-shaft in order to maintain the flow of water as the wellhouse collapsed, or whether stonework had been specifically taken away for use in another structure, remains uncertain. If the latter, the amount of material recovered would not have amounted to a great deal of useful raw material, so maintenance work perhaps seems more likely. Twentieth-century capping and concealment The only significant modification made to the medieval structure itself occurred sometime during the mid to late twentieth century, when a new underground pipe was installed to convey the still flowing spring water away to the north-east. Part KEITH PARFITT AND HELEN CLARKE 140 Fig. 11 General view of the excavated well, looking west. Scales, 1metre and 50cm. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 141 of the north-east wall of the well-shaft had been broken out to allow the insertion of the new six-inch outflow pipe. This work seems to have included the removal of a Caen-stone block containing an original medieval water-outlet hole (see above). The damaged wall was then made good with concrete blockwork. Interestingly, the base of the newly inserted pipe was set only 0.06m higher than the surviving medieval water-outlet, at 9.35m aod, implying that there had been minimal change in the working levels of the system over the centuries (Fig. 7). Externally, the new outflow pipe had been set into a broad, gravel-filled trench (Fig. 4) between 1.50 and 2.00m across. The trench itself was at least 0.50m deep with steep sides, although its base could not be exposed due to the water flow. The pipe-line itself comprised a series of oddments with the length of new pipe leading out from the well-shaft being of black pitch fibre, a material mainly in use between the 1950s and 1970s. Further to the north-east it consisted of lengths of ceramic land-drain, all different in design but all bedded on and enclosed by the same gravel deposit. This new pipework must have superseded the earlier conduit leading away from the well and presumably destroyed any original medieval pipework or masonry culvert. Following the insertion of the new pipe, two heavy concrete paving slabs had been used to cover the well-shaft. Finally, the sealed shaft, its enclosing well-house, and the eighteenth/nineteenth-century access steps were buried under tons of soil, rubble and general rubbish that included plastic, glass bottles, car tyres and an old domestic water tank, all tipped down the bank from the west. An interesting find amongst this material was a small lapel badge of the National Union of General Workers. This can be dated to the period 1916-1924 but it was certainly earlier than some of the other objects that had been included in the du mp. With this soil and rubble covering in place, all surface traces of Convent Well were obscured and collective local memory of the site seems to have faded quite quickly. After a few weeks re-exposure in 2014, the well site was back-filled for safety and to preserve it for the future. Immediately prior to backfilling, John Simpson, Water Quality Scientist with Affinity Water, tested the water from the spring (report supplied Dec. 2014), deducing that it was from a non-chalk aquifer. The water was rather harder than that from chalk, slightly more acidic and with much higher levels of iron. There was also a relatively high lead level, probably reflecting localised human contamination, as this metal is never found in natural raw water from the region. A high bacterial count quite possibly resulted from the disturbance caused by the excavation work. interpretation of the excavated structures The conduit-head and the conduit (Figs 4, 5 and 9) A medieval conduit-head consisted of a well-shaft, a cistern, a settling tank, and a well-house, together forming a structure where water from a spring or springs was gathered before being channeled out to its monastic destination (Fig. 3). At Woodnesborough the well-shaft served several purposes. The lower portion of the well-shaft acted as a settling-tank for purifying the water which came in from the spring and the upper part, lined with Caen-stone blocks, was the cistern. The stone KEITH PARFITT AND HELEN CLARKE 142 lining preserved traces of an outlet-hole and a lime-scale ‘tidemark’ marking the level of the water. There have been almost no excavations of conduit-heads, so that of Woodnesborough is of exceptional interest. Coincidentally, one of the few others to have been excavated was on the outskirts of Canterbury some 25 years ago, and although it was published as the ‘conduit house’ for St Augustine’s Abbey, it was certainly what would be called a ‘conduit-head’ in modern terminology (Bennett 1988). It was erected on a spring-line some distance outside and north-east of the precinct and collected water from several springs into a cistern or reservoir before piping it down to St Augustine’s. Some glazed ceramic pipes were discovered inside the conduit-head and a lead pipe is shown leading out south-westwards to the abbey (Bennett 1988, fig. 4). Built in the thirteenth-century, it was only slightly earlier than the Whitefriars’ conduit-head at Woodnesborough and, similarly, continued in use until the nineteenth century. Partially set into the lynchet bank, the well-house probably had a stone-vaulted roof and a narrow opening for access on the north-eastern side. It had been expertly constructed in stone, some of which was dressed and carved Caen stone. This may have been brought from Sandwich Whitefriars, which was in the course of construction during the late thirteenth century, and was probably still being built when the friars acquired the Woodnesborough spring. Some reused carved stones were also found lying loose in the adjacent soil and rubble, and in the infill of the well-shaft. Fifteen of the most elaborate carved pieces have been preserved in the owner’s garden at South Street Barn. The structure was 1.11m square internally, with solid walls apart from a single opening in the north-east. This was perhaps c.0.50m wide and may have been a doorway, although nothing has been found to suggest that it was originally closed by a door. However, it must have had a door sometime after its construction, for in 1534 the Sandwich treasurers’ accounts record that the town paid for a lock at the conduit-head (KHLC Sa/FAt 30). The one springer stone that remained in situ showed that the walls had been 1.13m high, and that the roof was steeply vaulted to a height of no more than 2m above the floor. It was, therefore, a tiny structure, much too small to allow ready access by local villagers and their buckets; rather, it was designed to protect the well-shaft from damage, and its spring water from pollution. The scanty excavated evidence makes it difficult to be more precise about what the Woodnesborough example looked like, but parallels may be sought in a number of fourteenth-century well-houses still standing on medieval monastic sites in England. Three survive at Mount Grace Priory, North Yorkshire, one of which was investigated archaeologically when a precinct wall was being consolidated in 1958 (Fig. 12; MA 1959, 306). It was described as ‘a small rectangular building of fine ashlar with a gabled roof of stone and an arched opening in its W. wall’. The Site Manager of Mount Grace Priory has kindly provided the following dimensions: 2m x 1.27m in area, with walls 2m high and a steeply sloping roof, the ridge at 3.17m above floor level’. A couple of years later another was found south of the first (MA 1962-3, 317), and even further south there was a third one, known as St John’s well, which was in use until 1901 (Coppack and Aston 2002, fig. 72). ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 143 Haughmond Abbey, Shropshire, has a similar fourteenth-century well house, ‘small and simple … with a gabled roof’ (Hope and Brakspear 1909, 308) as does Alnwick Abbey, Northumberland (Aston 2000, 102, colour plate 21). Canons Ashby Priory (Northamptonshire), Monkton Farleigh Priory (Wiltshire) and Valle Crucis Abbey (Denbighshire) all have well-houses of the same date and type (Bond 1989, 87), but a thirteenth-century structure recorded from Beaulieu Abbey, Hampshire shows what may be a chronological difference. This lies in its circular plan, 12ft (3.31m) in diameter and its domed roof, for the dimensions of its doorway, 5ft 3in (1.60m) high and 22in (0.55m) wide, are much the same as the known fourteenthcentury examples (Hope and Brakspear 1906, 178). The Mount Grace examples illustrated in Fig. 12 are roofed with stone slabs, and that is true of some of the other survivors. It could also have been the case at Woodnesborough, but would have been unusual in the Middle Ages in east Kent where stone was a rare and costly commodity. Nevertheless, stone reused from Sandwich Whitefriars was used in the construction of the well, so stone slabs may have been acquired from the same source. No remains of lead, ceramic or wooden pipes were found outside the north-east wall of the well-shaft where the water would have flowed from the shaft and into Fig. 12 Fourteenth-century well-houses at Mount Grace Priory, N. Yorks. (photos Helen Clarke). KEITH PARFITT AND HELEN CLARKE 144 the conduit, although one would assume that there must have originally been something of this type here. At Beaulieu Cistercian Abbey hollowed elm trunks were used (Bond 1989, 87) but at Woodnesborough the evidence was probably destroyed when the new outlet pipe was inserted during the twentieth century. The water flowed out of the well-shaft at 9.29m aod, then through a conduit of some sort to the conduit-house in the friary in Sandwich. The ground level there was between 2.20 and 2.65m aod, giving a drop of around 7m, sufficient to enable delivery of the water through a gravity-fed system, with a fall of approximately 1 in 300. The route, however, is unlikely to have been direct because an intervening area of higher ground, in the form of Mary le Bone Hill, would have interrupted the flow. The conduit must have been c.3km long, not unusual for monastic water systems in the Middle Ages when distances from water sources to religious houses were often between 3km and 5km, as at London Charterhouse or the Benedictine abbey of St Werburgh’s in Chester. The estimated fall of 1 in 300 for Sandwich was comparatively good, for in some other cases it was hardly sufficient to keep the water flowing. The St Werburgh’s conduit, for example, had a fall of only 3m over a distance of 3km (1 in 1000; Bond 1993, 50). Thomas Shelving’s donation of the spring and land in Woodnesborough specified that the conduit conveying the water to the friary should run underground through his land, but there is no physical evidence to support this. Had it been an underground conduit, the water would have been channelled through pipes of wood, ceramic or lead. Monastic houses are said to have favoured lead pipes, even though they were known to oxidise over time producing the danger of lead poisoning (Bond 1993, 49). They were, however, easier to seal against leakage, and that may have been considered sufficient to justify their use. But it is difficult to imagine their being used over the often long distances between water source and delivery point as lead was a costly material and skilled plumbers had to be employed to make, lay and repair the pipes. Beaulieu abbey, for example, had lead water pipes which were welded together, but they were needed to convey the water for only hundreds of metres rather than kilometres as was the case for the Sandwich water system (Hope and Brakspear 1906, 178). Perhaps ceramic pipes, similar to field drains, were used to traverse long distances in the countryside and lead pipes introduced into the system once the conduit reached the religious house or the town. In the case of the conduit from Woodnesborough to Whitefriars, pipes of any sort may only have been used once the water was inside the town or in the friary itself. Most of the way it may have flowed through an open ditch that eventually became the South Poulders Stream. Such open water systems were not unusual in the Middle Ages, and Sandwich town itself benefited from one when, in the twelfth century, streams from the Lydden valley, south-east of the town, were diverted to flow into the 4km long Delf Stream. The engineering involved then was probably financed by Christ Church Canterbury who had a small property at the west end of the town (Clarke et al. 2010, 36-7). Once the water arrived at Whitefriars it would have been be piped to a conduithouse somewhere in the precinct. The only archaeological evidence for such a thing is a feature excavated by Cecil Hogarth in 1972 who wrote a brief account of his work in 1973 (MA 17, 155). He found a ‘7ft-square (c.2m-square) brick ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 145 plinth’ in the centre of the cloister which, he suggested, was ‘the floor of a twopost roofed structure’. Although the excavator does not say so, it may have been the foundation of a conduit-house, but he makes no mention of pipes leading to or from it. Although a number of brick and stone-built culverts have been recorded within the monastic precinct, these all appear to represent foul-water drains, rather than part of a freshwater distribution network. Prior Wiberts’s plan of Canterbury Cathedral Priory in the twelfth century shows conduits criss-crossing the site (Bond 1993, fig. 5.3) and at Sandwich we should expect there to have been a less complex version of the same thing. The later history of the Whitefriars’ water system After the record of Thomas Shelving’s gift in 1306, there is no further documentary evidence for the conduit from Woodnesborough until the fifteenth century when there are indications that the town council had become interested in it. As mentioned above, the Delf was Sandwich’s main water supply and as it was an open ditch through the town as well as through the countryside it became appallingly polluted and needed constant maintenance. The townspeople must have been envious of the friars’ clean water, but it was not until 1483 that the town council decided that it could afford to lay on piped water for the inhabitants. The council brought in masons from Maidstone to build a conduit leading to a brick cistern in the friary precinct (KHLC Sa/FAt 8). By 1485 there were two pipelines through the town, one carrying water to ‘Davis Gate’, now The Barbican, and the other to a conduithouse (called ‘roundhouse’ in the document) in ‘Fishmarket’, now Market Street (KHLC Sa/AC1 f.305v). There is nothing to suggest that the water came from any source other than the Woodnesborough spring. One might think that there was a danger of the water drying up during dry weather, but there are no documentary mentions of this. In any case, the friars would always have had first call on what water there was as the town’s supply was next along the line. Although the water may have been led to Sandwich in open ditches, the documents show that the pipes in the town were made of lead. For example, in 1491 a plumber from Canterbury was employed to repair the conduit; this involved breaking up the pavement and replacing it after the pipes had been soldered together again (KHLC Sa/FAt 10). Repairs were needed frequently and fell heavily on the public purse, so it is rather surprising that by the 1530s the town had taken over responsibility for the Whitefriars’ stretch of conduit as well as the conduit-head itself, which was ‘scoured out’ in 1534. The takeover by the town may tell us something about the state of the finances of the friary shortly before it was dissolved in 1538. Apparently, the friars were then reduced to such dire straits that they were begging to be taken over by Henry VIII because otherwise they would have to sell the tiles and lead from their church ‘for they have nothing else’ (Letters and Papers of Henry VIII, xiii: 2, no. 1058, p. 452). Some of the friars may have gone out into the world before the Dissolution, for in 1537 Edmund ‘Friar’ was paid 18s. for mending the pipes outside Woodnesborough Gate (KHLC Sa/FAt 32). In the same year the whole responsibility was handed over to a sub-contractor, James Hall, who was to be paid 40s. annually ‘to keep in good order [the conduit] from Davis Gate to the conduit-head at Woodnesborough’. But this soon proved to be KEITH PARFITT AND HELEN CLARKE 146 an ill-judged move to privatization, for in no more than five years, the decay of the conduit was attributed to the same James Hall (KHLC Sa/AC3, f.90). The town must then have resumed responsibility, and there are records of expenditure for cleaning and repairs until the end of the century. After the Dissolution, the site of Whitefriars was sold to Thomas Arden of Faversham who held it until his notorious murder in 1551. The site then went through various hands up to the present day (Boys 1792, 176-7; Deighton 1994, 326, fn21). It seems that the friary buildings were soon demolished and not replaced by a dwelling of any sort, so the original water system was presumably no longer needed. We do not know what happened to the conduit leading into the friary precinct, but the Woodnesborough conduit-head was still accessible by steps around 1800 and not abandoned until the early decades of the twentieth century (see above). The branch of the conduit leading into the town was probably still in use until the late nineteenth century when a modern piped supply for the town was laid from a new corporation waterworks, also established at Woodnesborough in 1894 (Vaile 2015, 112-14). acknowledgements The authors are most grateful to Gary and Tracey Hall for allowing excavation on their property and their enthusiastic encouragement; to Barbara Owen, the Site Manager of Mount Grace priory for her assistance and to C.J. Bond for kindly providing the original of Fig. 3. Above all, we thank those members of Dover Archaeological Group whose persistence led to the rediscovery of Convent Well. bibliography Aston, M., 2000, Monasteries in the Landscape, Stroud. Bennett, P., 1988, St Augustine’s Abbey conduit house, Archaeologia Cantiana, 106, 137- 141. Bond, C.J., 1989, ‘Water management in the rural monastery’, in R. Gilchrist, and H. Mytum (eds), The Archaeology of Rural Monasteries, BAR British Series 203, 83-111. Bond, C.J., 1993, ‘Water management in the urban monastery’, in R. Gilchrist, and H. Mytum (eds), Advances in Monastic Archaeology, BAR British Series, 227, 43-78. Boys, W., 1792, Collections for an History of Sandwich in Kent, Canterbury. Clarke, H., Pearson, S., Mate, M. and Parfitt, K., 2010, Sandwich – the ‘Completest Medieval Town in England’: a Study of the Town and Port from its Origins to 1600 (Oxbow, Oxford). Coppack, G. and Aston, M., 2002, Christ’s Poor Men: the Carthusians in England, Stroud. Deighton, E., 1994, ‘The Carmelite friary at Sandwich’, Archaeologia Cantiana, 114, 317- 327. Fordham, S.J. and Green, R.D., 1973, Soils in Kent II, Sheet TR35 (Deal) (Soil Survey record No. 15). Hope, W.H.St.J. and Brakspear, H., 1906, The Cistercian abbey of Beaulieu, in the County of Southampton, Archaeological Journal, 63, 129-186. Hope, W.H.St.J. and Brakspear, H., 1909, Haughmond Abbey, Shropshire, Archaeological Journal, 66, 281-310. MA 1959, ‘Medieval Britain in 1958’, Medieval Archaeology, 3, 306. MA 1962-3, ‘Medieval Britain in 1961’, Medieval Archaeology, 6-7, 317. ‘SCOURING THE CONDUIT HEAD AT WOODNESBOROUGH’ 147 MA 1973, ‘Medieval Britain in 1972’, Medieval Archaeology, 17, 155. Ogilvie, J., 1960, ‘Mary-le-Bone Hill’, Archaeologia Cantiana, 74, 141-50. Parfitt, K., 2014, ‘Convent Well at Woodnesborough rediscovered’, KAS Newsletter, 100, 10-11. Vaile, M., 2015, ‘Late Victorian Sandwich – polluted water, sickness, Council inertia and the case of the deceased unnamed mayor: a public health study’, Archaeologia Cantiana, 136, 101-115.