Paxton House, Paxton Glen, Waterworks

This site was surveyed as it provided an example of a domestic industry to be added to the National Inventory.

Paxton Glen, Waterworks

(NT95SW 120.00) see also sites NT95SW 120.01-120.14

1. Background

The provision of a domestic water supply to Paxton House (built 1758-63, with significant work of 1811-12 under George Home and the Edinburgh architect Robert Reid) would have been a major reason for the choice of site for the house. The house sits adjacent to the Paxton Glen where there are several natural springs, as well as an accessible water table. The Nabdean and Paxton Burns have created the natural glen in which the Paxton House water works are situated, the burns providing a motive water supply for pumping. The archives offer tantalising references to the mechanical means by which water (especially drinking and bathing water) was supplied to the west wing of the house in the early 19th century. Water was variously supplied to the House by hand, horse-driven pump, waterwheel, ram pump (or hydram) and electric pump. (1) The installation of water closets meant the building of a cistern (now gone) in the stable block (GD267/4/1). Paxton House continued to use the waterworks site (with several developments), until a period of drought in the 1960s saw a move to connect to mains water. (2)

The impounded water was used in the waterworks at NT9325 5216 as well as for ornamental purposes in an area above a weir at NT93379 52246 (NT95SW 15.04). The weir with its substantial dam (originally crossed by a Chinese style bridge, now gone which appears on the ‘Plan of the Policy at Paxton House’, known as ‘The Blackadder Map’ by John Blackadder, c.1810) created an ornamental ‘lake’, which would be viewed from the carriageway as it crossed the Linn Bridge (NT95SW 20).This body of water is now heavily silted up, but the intention is still visible. This area also provided supplies for the ice-house situated under the S arch of the Linn Bridge on the S bank of the Linn Burn (NT95SW 15.10).

This site was recorded as it is under threat from the elements, is unique and provides an example of an unusual water management system for the National Record. Paxton House Trust is tentatively looking into bringing parts of the system back into use as part of the ‘visitor experience’ of the House and policies.

2. The Site (NT95SW 120.0)

The overall form of the site as far as water management is concerned may date from the 1811/1812 works carried out by George Home and utilising already existing spring water collection apparatus at the waterworks site.

There are various elements to the surviving waterworks in Paxton Glen (moving W to E):

2.1 NT9305 5228: Pond (NT95SW 120.10), Dam and Sluice Gate (NT95SW 120.1, NT93054 52289)

A ‘dam’ and ‘sluice’ are depicted and named on the Ordnance Survey 1st edition, 6-inch map (Berwickshire, sheet XVIII, 1862). The pond created by the dam is fed by the Nabdean Burn although the pond has now silted up, having been in use up until at least the first quarter of the 20th century. The pond measured 20.5m (north/south) by 20.0m (west/east) on the 1st edition map and covered an area of 328 square metres (4046 square feet). On the current Ordnance Survey map, the west and south line of the pond is defined by the south bank of the Nabdean Burn as it runs through the now silted-up pond area.

There is a one- arch masonry bridge over the Nabdean Burn which acted as a dam to the reservoir which served Paxton Glen (DP048814 and DP048815). This bridge continues to allow access from the park in front of Paxton House to the dene of the Linn Burn (the name of the Burn after the confluence of the Paxton and Nabdean Burns) and to the walled garden on its N side. The N arch of the bridge incorporates a sluice on its upstream side (DC53603, A, DP048808-DP048812). There are two overflow channels (DC53603, B and C, DP048815 and DP048816) built into the bridge on either side of the sluice. The N Channel (DC53603, B) measures 0.6m in width, 0.5m in height and 5.05m in length. The South channel (DC53603, C) measures 0.6m in width, 0.5m in height and 5.0m in length.

The sluice gate arrangement consisted of two masonry abutments (doubling as a walkway 0.65m in width) and a lintel supported by the bridge behind (DP048805). The walkway ‘capstones’ (DP048806) show disturbance and do not fit together as if they have been dislodged and replaced poorly (possibly a result of the August 1948 floods). The sluice gate itself (DC53603, D) consisted of a cast-iron ‘plunger’ valve measuring some 2.3m in length (DP048804).

The sluice valve would have been opened and closed using a simple hand-operated worm drive with a horizontally set ‘wheel’ handle (now gone). The valve would have sat on a seal in the form of a cast-iron ring which in turn sat on a wooden casing, or box, which would have formed a barrier to water flowing under the lintel when the valve was in a closed or down position (DP048806 and DP048808). There are fragments of this wooden box arrangement in situ (DC53603, E; DP048808). There is little evidence surviving of a superstructure which held the sluice valve in place above the wooden sluice box. Presumably the sluice valve could be opened and closed from the walkway created by the sluice abutments. Water could also flow over the sluice gate forming a weir if there were high water levels. Remnants of the valve, seal and some wood fragments (possibly from the superstructure or from the sluice box) are scattered around the sluice-gate site.

The water would then pass under the bridge at NT93055 52287 through an archway measuring 1.5m in width and 1.35m in height. The bed of the Burn beyond the bridge (heading downstream) has been revetted with flat stones to NT93058 52283 in order to reduce scour (DC53603, F).

2.2 Sluice Arrangement (NT95SW 120.2, NTT9315 52246) and conduit (NT95SW 120.11, NT 93115 52245)

The now canalised burn continues to the remains of a second sluice arrangement (DC53602, A and B). There are the remnants of two sluice gates and their dams at NTT9315 52246 and NT 93115 52245. Each gate measures 0.2m in width (DP048819). The sluice gate superstructure at NT 93115 52245 is denuded (DC53602, B), presumably through flood damage (in 1948) and lack of maintenance. The masonry, grooved, guide-stones for the sluice-gate structure appear to have been washed downstream to NT93116 52241 (DC53602, C). This second gate and dam (in a closed position) would have formed a small ‘basin’. The sluices and their dam walls were originally of masonry with later concrete additions.

The water impounded between NTT9315 52246 and NT 93115 52245 appears to have been channelled to the feeder pipe (NT95SW 120.11, NT93114 52245) which supplied the enclosed underground conduit (DC53602, D). This conduit in turn supplied the waterworks/pumping station some 200m to the E (see NT95SW 120.07-NT95SW 20.10 and NT95SW 120.12-NT95SW120.13).

The course taken by the enclosed lade or conduit (NT95SW 120.11) [the word Scots word, ‘cundy’ is also used locally and means a drain or conduit] which carries burn water, probably follows a rough alignment from NT93114 52245 to NT93113 52244 to NT93119 52229 to the water wheel pit area at NT932 521. There is also a feed to the ram pump house (NT95SW 120.09) at NT93268 52170.

2.3 The Dene, ‘laundry’ area (NT95SW 120.3, NT9311 5223 centred)

Carrying on downstream through the area known as The Dene (4) beyond the enclosed conduit or lade entrance, there is an area traditionally described as ‘the laundry’. (5) It appears that the bedrock has been fashioned to form a weir (see DP048820 and DC53602, E) to allow the management of water at a domestic level. There is a cut channel (see DC53602, F) with a ledge alongside the N burn edge which could perhaps accommodate receptacles for collecting or otherwise managing water use.

On the S side of the Nabdean Burn at NT9311 5223, there is a drystone revetment wall in the bank which allows for a ‘work’ area adjacent to a shallow depression in the stream bed. The shallow depression itself (DC53602, G) measures 1.20 m in length by 0.9m in depth. It may have begun life as a natural depression caused by boulder attrition, although the depression is wide and smooth which could suggest human intervention to create a laundry ‘basin’ (see DP048822, DC53602, H), the pond performing as an area for rubbing and agitating cloth. Adjacent to this depression the rock of the stream bed may have been worked to provide squared off edges, again suggestive of an arrangement to channel water to this shallow depression area.

The stream continues under the bridge at NT 93122 52231 (NT95SW 120.04; see DC53602 I, DP048821 and DP048824). This bridge was rebuilt after the floods of August 1948 (3) and incorporates remnants of the old bridge which is depicted on the 1st edition OS map. The bridge deck is constructed of riveted plates, brick and concrete (poorly executed at burn bed level: see DP048821, DP048823). The bridge also incorporates rubble masonry on its upstream side, at its S end and on its downstream side (DP048825). The parapets are of rubble masonry.

2.4 (NT931 521): Cisterns (NT95SW 120.6) and pump house (NT95SW 120.5)

This group of structures (NT95SW 120.6, NT93250 52166) includes a pump house (NT95SW 120.05; NT93144 52166) with a well adjacent and a cistern below (DP048826). The lower cistern has two compartments: one to collect sediment, the other (with a galvanised perforated ‘rose’ for filtration) to collect the water for moving onto the pumping area. This water was then channelled down the 3-inch cast iron pipe (2-inch bore) carrying the spring water above and roughly along the inferred line of the enclosed lade E towards the main waterworks pumping area (NT9325 5216). This 3-inch pipe was picked up again at the main pump site at NT93243 52158 when a small trench was dug by Mr Home Robertson in 2008. The well adjacent to the pump house was locked on the date of visit.

The pump house is of rendered brick construction and is of 20th century date (DP048827and DP048828). The relationship between the pump house and the lower cistern is not clear. It may be that this pump house primed the siphon effect from the well adjacent to the pump house to move the water to the cistern and then onto the pumping area at NT9325 5216.

2.4 The Waterworks area (NT9325 5216)

2.5.1 The main cistern area (NT93250 52166);

The Ordnance Survey 2nd and 3rd editions of the 25-inch map (Berwickshire, 1909; 1926, sheet XVIII.9) show a roofed structure in this area, This structure (presumably the covered cistern) does not appear on the Ordnance Survey 1st edition 25-inch map (Berwickshire, 1862, sheet XVIII.9) despite the general consensus being that the enclosed reservoir or cistern and the associated wheel pit and other structures were in place by 1811 with work being carried out in 1855. It may be that the area was covered at some point after 1862 and that the apparatus on the site did not merit being mapped at this scale.

There are three springs collecting in cisterns in this area (DC53601, A, B and C). The cistern area consists of (from W to E) a brick cistern (DC53601, A; DP048848) measuring 0.60m by 0.65m into which a spring runs from the slope above, a second (older) brick cistern of poorer quality (DC53601, B, DP048852, DP048853) fed by a second spring with two compartments with two pipes with ‘rose’ terminals feeding water to the larger cistern adjacent (see DC53601, DP048829, DP048830). The two compartments with a dividing ‘wall’ impeded the settling of sediment in the compartment feeding the adjacent main masonry cistern to the E (DC53601, D).

There is a third brick cistern at NT93244 52165 (DC53601, C) which may have been fed from the bank at NT93247 52167 where there was a galvanised lining set into the bank wall (DC53601, S and see DP048854). The spring which emanates from NT95SW 120.06 appears to have been fed into the dressed masonry cistern (DC53601, D).

The original, dressed masonry cistern (DC53601, D, NT93249 52166 and DP048830) was originally plaster-lined in order to decrease the possibility of contamination from surface water. These capstones have pecked undersides as have the interior side masonry blocks to act as a key for the plaster. It has two inlet pipes, one from the brick cistern at NT93244 52165 (DC53601,E) and one feeding in from the bank into which the main cistern is built (originating at NT93144 52166, DC53601, B; see 2.4 above). The outlet pipe is at the NE end. The outlet opened out on to a cemented area and the water was guided by roughly made cement troughs into the waterwheel pit (see DC53601, F; DP048845). This appears to date from the insertion of a second tank or sump (now gone – see DC53601, G). The cement base on which this metal second cistern or sump sat (NT93253 52168, DC53601, G and DP DP048849) bears a date of ‘1941’ and traces of the metal of the tank sides are visible at ground level.

The flat concreted area (see DC 53601, H) between the metal tank and the wheel pit shows the outlines of pump base plates and fixings and presumably overlies any earlier pump installation evidence (at NT93253 52168 and see DP048840). Later pumps were presumably scrapped by the 1960s when the house was connected to mains water.

2.5.2 Waterwheel pit area (NT95SW 120.7) c. NT9324 5216)

The Wedderburn Papers of the Home family (National Archives of Scotland (NAS): GD267/7/16) note that on the 22 June 1816, one George Purves, Millwright,

“Received from Mr Home of Wedderburn one pound fifteen shillings by agreement for keeping up and supporting the mill machinery trows sluices and pillars for supporting the trows for raising the water to the house of Paxton for one year from Martinmas 1814.”

This system was installed as part of additions to Paxton House 1811-1812 which included water closets ‘and apparatus’, sheet lead for cisterns (NAS: GD267/4/1) and masonry work.

The waterwheel pit (NT93247 52168, DP048842 and DC53601, I) is well constructed of dressed masonry which has been partially dismantled to 0.75m at its highest point. Mr J Home Robertson has found dressed masonry in the wheelpit, as well as a pinch bar abandoned by those who demolished the structure. The waterwheel would have been positioned north/south and was fed by a pipe or enclosed lade from the Nabdene Burn some 200 m to the W (DC53601, J). The wheel would have been narrow (the pit is 0.8 m in width) and would not have exceeded 2.5m (8 feet 2 inches) in diameter, although the demolition of the waterwheel pit has removed the evidence of axle bearings at either side of the pit. It is presumed that an open launder or flume connected the enclosed lade outlet in the bank into which the pit is built to (possibly) an overshot waterwheel. The reference to ‘trows’ in documents (trows were troughs or open conduits; NAS: GD/267/7/16) suggests open lade-like constructions which could perhaps describe the arrangement for channelling water from the enclosed lade outlet from the burn in the bank to the water wheel (a distance of up to 5.5m, see DC 53601). The waterwheel would have, in turn, operated the pumps.

A ‘water engine’ is referred to as early as 1809 (6), although this would suggest developments before 1811 when the water closets and other ‘apparatus’ was installed in the House. The pumps ‘at the waterwheel’ had work carried out on them in 1855 and the cistern at the stable (now gone) is referred to in 1889 (7). The types of pumps in not known. There may have been a combination of lifting and sucking pumps which would have been put together and installed by a plumber. A forcing pump may have been in use as water was needed to be sent uphill. These would have consisted of wooden cased pumps with metal pistons and valves sealed with leather. (8)

There are examples of similar waterworks such as that described by Celia Fiennes in 1695 who noted that Broadlands House, Hampshire (the original manor house, now gone) had a ‘water house that by a wheel casts up the water out of the river just by and fills the pipes of the house...’ .(9) The Aster Tower at Blenheim (1706-09) had an ‘engine’ pumping water to a cistern at the top of a hill which than ran by gravity to a cistern at the gatehouse. An interesting example with parallels to Paxton House is at Houghton Hall in Norfolk where in 1730, a wheel was turned by a horse-engine pumping well water through a lead conduit into a large cistern which was apparently in operation until the 1920s. (10)

2.5.3 Horse engine (NT95SW 120.08, NT93257 52168)

The horse-engine (or horse-gin, DC53601, K, DP048837, DP048834 and DP048835) pre-dates the pump house adjacent to it (NT95SW 120.12, see DC53601), as the pump house would have impeded any horse or human movement around the shaft and mechanism. It has an overall diameter of 2.5m, excluding the platform on which the horse would have walked in order to turn the engine. The surviving mechanism in situ consists of a cast-iron bearing bar. There are fragments of mechanism found by John Home-Robertson (in 2009) which may have come from the horse-engine (see DP048856-DP048858). There is a shaft bearing-block on the adjacent concreted area (NT95SW 120.06 and DC53601, L). The shaft alignment can be seen (DP53601) running from the horse engine but it is unclear when it was abandoned although there is oral testimony to the fact that it was in used around the time of the First World War (1914-1918). (11)

2.5.4 Ram Pump House (NT95SW 120.09, NT93268 52171)

The ram pump (or hydram; DC53601, M) was ordered from John Blake Ltd water engineers (now part of Allspeeds, Accrington, Lancashire) in 1897 (MS/6189/2). It is a Blake’s ‘B’ ram pump which used dirty water to pump clean water. A ram pump works on the principle of a head of water at a little height (in this case dirty burn water collected in a feed tank) used to elevate some of the volume of the water (or in the case of Paxton House, a volume of spring water) to a greater height (see MS/6189/3 and MS/6189/4).

The hydram at Paxton Glen has been partially dismantled, with the ‘bell’ shaped air vessel typical of this type of pump with all its internal valves, back pressure valve and waste valves and casing having been removed (presumably for scrap or reuse). The brick-built pit-house (or sunken pump house which gives extra fall from the feed tank at NT93524 52164, see DC53601, N), the waste water injection socket, the brick feed tank (DC53601, O) and (presumably) the drive pipe between the pump and the feed tank survive (DC53601, P). The power water feed pipe, allowing water into the feed tank for the pump (at NT93524 52164), would have joined to the enclosed lade from the Nabdean Burn via the surviving pipe and socket flange joint ring, as the power feed pipe needs to almost level with the tank it supplies (DC53601, Q). The small cistern which would have collected the spring water to be pumped to the House from the pit-house has been removed but the spring feed pipe survives (see MS/6189) although it is unclear which spring fed it or how the pipe-work might have been arranged. It is unknown when the ram pump was abandoned.

2.5.6 Pump House of the 1940s (NT95SW 120.12; NT93258 52168), pipe bridge (NT93256 52174) and well (NT95SW 120.13)

This pump building (NT95SW 120.12, see DC53601 R, DP048835, DP048836, and DP048838) is of rubble construction with a slate roof (now gone) and sits adjacent to the horse engine (NT95SW 120.08). There is an E facing window facing, the door faces N onto the pipe bridge (the superstructure of which is now partially demolished). There is a surviving base plate on which the pump sat which says ‘BERESFORD STOURPORT’. This building contained a pump which pumped water to the walled garden (NT95SW 15.07) and to the main house (NT95SW 15.00), presumably (and latterly) from the well (DC53601, T; NT95SW120.13), although the relationship between the well and the pump house was not investigated as access into the well could not be arranged.

The relationship between the pump house and the 1941 cistern base (NT95SW 120.06) is not clear. The spring water collected in this 1941 cistern or sump (not clear whether linked to run off or for collecting spring water) may have been pumped up to the house using the pump in this pump house as the horse-engine, and waterwheel, were probably no longer in use.

3.0 Dating

It is clear that there are several dateable phases despite the constant repairs and re-modelling that characterises this site. The waterwheel pit and the masonry cistern probably date from the early 19th century. There was possibly an earlier incarnation of spring collecting apparatus prior to the setting up of the waterwheel and pumps in 1809 and the extensions to the house in 1811.

The cisterns are all from different dates, some newer than others such as DC 53601 A and C predating B (based on quality and wear and tear on the brickwork) which may predate the ram pump (1897).

The ram pump with its definable features dates to 1897 from independent documentary evidence. The later developments in the 1940s are represented by the helpfully dated cistern or sump base of 1941 at whose S end masonry wall line survives, suggesting earlier structures. The area adjacent to this has pump base plate remnants may date from an earlier period (pre-1941) as the shaft drive from the horse-engine (in use until at least 1918 and pre-dating the ram pump) is accommodated.

The well, according to Mr Home-Robertson, dates from the early 1950s. (12)

References

(1) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 29, 30.

(2) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 30.

(3) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 33. The ‘Dene’ was a husbandry enclave used by the Home family. Currently, the surviving cottage on the site is being renovated.

(4) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 34

(5) Information from Mr John Home-Robertson, August 2009.

(6) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 30. George Home journal 7th November 1815; ‘The first thing to be pointed out to Mr Stevens [drainer] is the cistern that collects the water for supplying the house that in the course of his operations he may do nothing to diminish the collection of water there’; ‘ A water engine is referred to in 1809 and later in 1855, pumps at the waterwheel that cost £1-6s-0d. The water works needed a steady eye and managed on a daily basis. House hold accounts 20 Feb 1889: “I told him that after Whits. he [Mr Waite the gamekeeper] was to take charge of the Water Wheel, & water works; & Mr Bardner told him he was also to watch the Stable tank so as to get it emptied before overflowing. He will get help of the two Policy men for this.”’

(7) Jefferson-Davies C., and Snow, E., Creating Paxton (Paxton Trust, 2008), 30.

(8) Nicholson, P. The New Practical Builder (London, 1823), 408-409.

(9) Girouard, M., Life in the English Country House (Yale University Press, 1978), 250.

(10) Girouard, M., Life in the English Country House (Yale University Press, 1978), 251.

(11) Information from Mr John Home-Robertson, August 2009.

(12) Information from Mr John Home-Robertson, August 2009.