2012:178 - Dundrum Castle, Dundrum, Down

Excavations were undertaken between June and August 2012 at Dundrum Castle in County Down.  The castle, a State Care Monument, situated on the summit of a prominent hill overlooking the small coastal town of Dundrum and the adjacent tidal inlet of Inner Dundrum Bay, has a long and complex history.  The surviving standing remains consist of an inner and an outer ward.  The inner ward dates to the Anglo-Norman period and contains the remains of a large circular tower and a gatehouse, whilst a small 17th-century house is located in the outer ward.  Artefactual and place-name evidence suggests that the Anglo-Norman castle was, almost certainly, built upon the site of a high-status, Early Christian period settlement.  It has been suggested that proto-historical evidence from the Ulster Cycle of Tales affiliates the site with the story “Bricriu of the poison tongue”.  Following the decline of the Anglo-Norman Earldom of Ulster in the 14th century, the castle was occupied by a branch of the Magennis family, although occasionally it was temporarily possessed by other prominent Gaelic figures and various representatives of the English Crown.  During this later medieval phase the upper floor of the circular tower within the inner ward was significantly altered.  It is uncertain whether the outer ward dates to the Anglo-Norman period or the later Gaelic phase of medieval occupation.  Following the Nine Years’ War (1594 – 1603) the castle and its associated estate was held by Edward Cromwell, then Governor of Lecale, whose son subsequently sold it to the Blundell family.  After being garrisoned during the wars of the 1640s, local tradition maintains that the castle was slighted by Parliamentarian forces in the mid 17th century (for the earliest published record of this tradition cf. Harris 1744, 15).  There is no evidence to suggest that the castle was occupied after that date, although its associated manorial estate continued to be farmed by tenants of the Blundell family.  During the late 18th century the castle and its estate passed into the hands of the Hill family and formed part of the Downshire Estates.  The immediate environs of the castle were landscaped at this date in order to create a picturesque setting for the ruin.  Dundrum Castle was placed into State Care by the Marquis of Downshire in 1954, prompting an extensive programme of conservation at the site before it was fully opened to the public in the early 1960s.

Prior to 2012, only a limited amount of archaeological investigation had taken place at Dundrum Castle.  Dudley Waterman had excavated a number of narrow trenches within the inner ward during the 1950s with a view to establishing the relative sequence of the upstanding remains.  His excavations revealed that a rich stratigraphic sequence survived within the inner ward and also produced a wealth of artefactual material ranging in date from the Early Christian period to the 17th century (Waterman 1951; 1958).  Waterman also supervised the clearance of the cistern beneath the circular tower during the winter of 1959-60 (Waterman 1964) and conducted unpublished excavations in the outer ward in 1960.  In 2009 limited excavations were carried out at Blundell’s House, the 17th-century building located across the southern wall of the outer ward (Macdonald 2011a; 2011b; Excavations 2009, No. 257, AE/09/36), whilst in early 2012 resistivity and magnetometry surveys were conducted by the Centre for Archaeological Fieldwork, in partnership with GeoArch Ltd, in both the inner and outer wards, as well as across the extra-mural terraces located immediately to the south and west of the outer ward.

The 2012 excavations were conducted in two phases. Initially, the excavations had an evaluative character and were filmed for an episode of the television programme Time Team, which was first broadcast in February 2013.  Following the departure of the cameras, the excavation took on a research character and continued under the joint direction of Philip Macdonald of the Centre for Archaeological Fieldwork and Liam McQuillan of the Northern Ireland Environment Agency (DOE).  In total six trenches were opened, five in the inner ward and a sixth within the outer ward.  The excavation, commissioned by NIEA, had two aims.  The first aim was to promote Dundrum specifically, and Northern Ireland in general, as a tourist destination through exposure of the castle on an internationally broadcast television programme.  The second aim was to refine and expand the known archaeological sequence of the site with a view to improving both the public presentation of the monument and informing the future management strategy for the site.

 

Trench One

Excavation of two of the trenches within the inner ward consisted of partially reopening, and then extending, trenches dug by Waterman.  The most significant of these, our Trench One; (Waterman’s Trench 2, cf. Waterman 1951, 19-20, fig.1) was located against the curtain wall in the southern part of the inner ward and its excavation resulted in a partial revision of the site’s sequence.  Waterman had identified a series of dumped deposits which he suggested probably represented an earth-and-rubble bank that predated the construction of the curtain wall and that he, and subsequent authors, identified as representing the remains of a ringwork-like defence.  Waterman suggested that this feature probably dated to the ‘campaigning’ phase of the Anglo-Norman conquest of Ulster led by John de Courcy in the last quarter of the 12th century (Waterman 1951, 19, 23-24; 1958, 64-65; ASCD. 1966, 207; see also Sweetman 1999, 36-37).  The 2012 excavations demonstrated that these deposits were actually a series of dumps used to level the ground surface within the inner ward, and that they post-dated the building of the curtain wall, but predated the construction of the large circular tower.  Underlying these levelling deposits was a soil horizon that sealed the construction cut for the curtain wall and a negative feature of uncertain purpose.  The latter feature was only visible in the east-facing section of Waterman’s trench and did not extend into the 2012 extension of the trench, hence the uncertainty over its character.  Construction of the curtain wall of the inner ward can be confidently attributed to John de Courcy and dated to the final decades of the 12th century.  The two features were cut through a layer of possibly redeposited gritty loam that overlay a sandy clay loam which, in turn, directly overlay the shattered surface of the bedrock.  Provisional analysis of the small number of pot sherds recovered from these early deposits suggests that the upper layer may date to the ‘campaigning’ phase of de Courcy’s Ulster enterprise, whilst the lower deposit is probably a pre-Norman soil horizon.

At the southernmost end of Trench One, and cut through one of the levelling deposits, was a negative feature that appeared to represent an attempt to destabilise or slight the curtain wall.  Waterman didn’t identify this later feature, misinterpreting its surface as the ‘toe’ or outer edge of his putative earth-and-rubble bank (Waterman 1951, 19 (see above)).  Within the footprint of his original narrow trench the edges of the later feature and the construction cut for the curtain wall did coincide – hence his confusion.  However, the extension of Waterman’s trench dug in 2012 demonstrated that they are separate, and stratigraphically unrelated, features.  The space between the edge of the later cut and the curtain wall had been backfilled with redeposited natural subsoil.  Stratigraphically cutting this fill, and extending throughout Trench One and also physically truncating the earlier levelling deposits was a horizontal discontinuity which possibly represents the creation of a level surface upon which the large circular tower within the inner ward was built.  This tower is plausibly dated, on architectural grounds, to the late 12th or early 13th century (McNeill 1980, 7; Donnelly 1997, 84-85), providing a somewhat imprecise terminus ante quem for the apparent attempt to slight the curtain wall and the levelling of the ground surface within the inner ward (McNeill subsequently revised his dating of the tower, plausibly attributing it to Hugh de Lacy during the period AD 1205-1210; cf., 2003, 98).  Overlying the horizontal discontinuity was a rubble-rich deposit of silty clay loam which may have been another levelling deposit.  Laid on to this deposit was a thin spread of mortar that represented the remains of a floor, probably of medieval date.   Unfortunately, post-medieval spade cultivation had removed all other traces of the building associated with this floor.  Overlying the truncated surface associated with the spade cultivation were only superficial and modern deposits.

 

Trench Three

The second of Waterman’s trenches to be partially re-opened and then extended in 2012 was located against the masonry addition to the curtain wall in the eastern part of the inner ward.  This was our Trench Three; Waterman’s original trench was not numbered (Waterman 1951, 22-23, fig.1, fig.2. E-E).  Within his trench Waterman had exposed the east-facing element of an apparently circular, dry-stone revetment that retained a tip of stone rubble.  Waterman stated that he could trace the topographic expression of part of this feature across the inner ward and his published plans indicate that it had a diameter of either 12m (1951, fig.1) or 11.3m (ASCD. 1966, fig.133).  Initially, Waterman expressed no view of the date or purpose of the revetment, apart from noting that it predated the masonry addition to the curtain wall.  He subsequently speculated that the revetment may have been intended as a foundation for the late 12th- to early 13th-century circular tower, which had never been used because a change of plan had caused the tower to be built elsewhere within the inner ward (ASCD. 1966, 208).  As built, the base of the circular tower has an external diameter of 14.9m (pace Waterman 1951, 15) which intimates that any tower atop the revetted structure would have been much smaller.  Ann Hamlin, in the text of her visitor guide card to the monument, noted that the dry-stone building technique was suggestive of a pre-Norman date (Hamlin 1982).

The re-excavation of the trench confirmed the validity of many of Waterman’s observations.  Underlying the topsoil was a thin deposit of dark brown, clay loam that apparently represented a cultivation soil of post-medieval date.  The truncated surface created by the cultivation sealed a series of levelling deposits of dumped stones and redeposited natural subsoil the earliest of which ran under the masonry addition to the curtain wall, suggesting that it was either deposited to level the interior of the inner ward or to form a level surface upon which the structure associated with the masonry addition was built.  Underlying the earliest of the levelling deposits was a deposit of sandy clay, which contained no finds, physically overlay the bedrock and butted up against the outer face of the revetment.  The outer-face of the revetment was built directly upon the uneven surface of either the natural mineral subsoil or the bedrock and consisted of several courses made up of angular slabs of local greywacke.  Its lowest, faced course consisted of large stones of varying sizes arranged so that their upper edges formed an even surface upon which to build the remaining courses of the structure.  Physically retained by the faced courses was a rubble fill which had been carefully keyed into the outer face-work of the structure.  Careful inspection of both the faced courses and rubble infill suggested that both had been earth-bonded with a dark brown clay loam.  What had not been explicitly recorded by Waterman was that the rubble fill in turn retained a layer of redeposited natural subsoil that within Trench Three directly overlay bedrock.  In fairness to Waterman, this interpretation was not obvious and was not recognised in 2012 either until investigations into what appeared to be natural subsoil within the interior of the revetted structure in Trenches Four and Six demonstrated that, in places, it overlay a truncated soil horizon.  Waterman himself had noted that his interpretation of the character of the revetted structure would remain provisional until its interior had been trenched (Waterman 1951, 23).

 

Trenches Four and Six

The revetted structure was investigated in two other trenches, Four and Six, both located within the central part of the inner ward.  The two trenches were conjoined - the western edge of the structure passing through the eastern edge of Trench Four, whilst part of the interior of the structure was investigated in Trench Six.  Within Trench Four the west-facing revetment wall had a similar character to the east-facing part of its circuit exposed in Trench Three, except rather than retaining a dump of rubble it retained a thick layer of redeposited subsoil.  This layer of redeposited subsoil was excavated in two narrow sondages within Trenches Four and Six and was found to overlie the truncated surface of a thin, buried soil horizon, which itself overlay either the shattered surface of the bedrock or a thin natural mineral subsoil.  That the buried soil horizon was truncated suggests that, prior to its construction, the footprint of the platform was deturfed.  Presumably, the removed turves were used as part of the construction of whatever structure was built upon the platform.  Excavation revealed that the lowest course of the west-facing revetment wall was mostly laid directly onto the exposed surface of the bedrock, however, in places where the uneven surface of the bedrock was deepest, it was possible to observe that it was laid on to a foundation deposit of small shattered stones that filled a foundation trench, which cut through a trampled silty clay soil. This, in turn, directly overlay the shattered surface of the bedrock and extended throughout the western part of Trench Four.  It is reasonable to assume that this soil is a continuation of the buried soil that was preserved beneath the revetted structure, however, the decision to preserve the revetment wall in situ meant that it was not possible to demonstrate this.

Although only two short sections of the revetted edge of the structure were exposed during 2012 (in Trenches Three and Four), the excavations demonstrated that it was not perfectly circular, but rather had a slightly elliptical plan with its shortest axis of 11.2m aligned approximately east-west and its longest axis aligned approximately north-west (estimated length 12m).  Rather than being part of a building the revetted structure is interpreted here as representing part of a platform.  Any evidence relating to the surface of this platform, such as its use to accommodate a building or other structure, was removed by subsequent activity.  Stratigraphically, this truncation was represented by the horizontal discontinuity associated with the post-medieval spade cultivation, however, the construction of the medieval building, represented by the masonry addition to the curtain wall noted above, would probably have been the episode responsible for the loss of this evidence.  The disposition of both the surface of the bedrock and the Anglo-Norman levelling deposits within Trenches Three, Four and Six demonstrated that the platform was located upon the summit of the hill.  No direct dating evidence for the revetted structure was recovered, although the earth-bonded quality of its construction suggests it is pre-Norman in date.  A large number of soil samples were taken from the truncated soil horizon it was built upon and it is intended to process them with a view to recovering material suitable for radiocarbon dating.

Butting up against the west-facing side of the revetment wall in Trench Four, and overlying the silty clay soil that the revetment wall’s foundation was cut through, was a sequence of dumped deposits of redeposited subsoil all of which contained displaced slabs that suggested their deposition coincided with the demolition of the revetted platform.  Although stratigraphically pre-dating the construction of the platform, the underlying soil horizon appeared to have been heavily trampled - presumably during the ‘occupation’ of the platform.  Consequently, finds recovered from this horizon are of limited value for dating purposes, providing only a terminus post quem for the demolition of the platform and not its construction.

A silty loam soil, associated with two localised deposits of silt that appeared to have been washed into the area of the trench, overlay the dumps of redeposited subsoil.  Cut into these soils was the base of a hearth which was truncated by a horizontal discontinuity apparently associated with the demolition of whatever building the hearth furnished.  Overlying this surface was a series of levelling deposits of dumped stones used to level the ground surface within the inner ward.  Unfortunately, Trench Four was not extended as far as the circular tower in order to ascertain whether these deposits pre-dated the construction of the tower, as was the case with the levelling deposits encountered in Trench One. However, it is reasonable to assume that they are medieval in date.  The upper levelling deposit was truncated by a discontinuity associated with spade cultivation that extended throughout Trenches Four and Six.  As with the stratigraphic sequence in Trench One, overlying this truncated surface were only superficial and modern deposits.

 

Trench Two

Our Trench Two was located within the inner ward, immediately adjacent to the passage through the gatehouse.  That the gatehouse was built over a reduced length of the curtain wall indicates that it was not the original entrance into the inner ward.  The original entry was, almost certainly, located in the eastern angle of the inner ward adjacent to the bridge-pit which was cleared out in an unrecorded episode of excavation prior to 1955 (ASCD. 1966, 207, fig.133).  On architectural grounds the gatehouse is conventionally dated to the second half of the 13th century (ASCD. 1966, figs.133-134; Donnelly 1997, 85).  Unfortunately, Waterman’s excavations within the gatehouse only uncovered a sequence of disturbed deposits and he found it impossible to corroborate this date on archaeological grounds (Waterman 1951, 20).

Excavation of Trench Two revealed evidence for a timber building which partially extended into the south-western part of the trench.  This consisted of a clay floor and an associated post-hole which respectively overlay and cut into a truncated soil horizon that contained only animal bone and shell.  The form of the post pipe associated with the post-hole suggested that the timber building had been dismantled following a period of occupation.  Although no direct dating evidence for the building was recovered during the course of the excavation, given the trench’s location immediately adjacent to, and on the path of, the passageway of the gatehouse it is reasonable to suggest that it was demolished prior to constructing the gatehouse.  Overlying the horizon associated with the building was a sequence of rubble deposits of uncertain date that were probably created during either one or two episodes of demolition or slighting of the gatehouse.  Again, as with the other trenches within the inner ward, the most recent of these deposits had been truncated by post-medieval spade cultivation.

 

Trench Five

Only a single trench was excavated in the outer ward (Trench Five).  It was located over part of one of two large annular, sub-rectangular anomalies identified in the magnetometry survey of the castle.  The anomaly was recorded in the north-western quarter of the outer ward in an area of well-preserved spade-cultivation ridges located on a steep slope.  Excavation revealed that this anomaly was caused by the well-preserved remains of a limekiln, almost certainly of medieval date.  Underlying the cultivation soil was a layer of hillwash that sealed the partially demolished kiln.  Investigation of the kiln revealed that it was a slightly elliptical-shaped stone structure that survived to a depth of at least 2m.  The absence of any topographic expression within the lower ward of the presence of such a large structure reflects the large amount of soil that has either accumulated or been deliberately deposited in the eastern half of the outer ward.  That the medieval ground surface within the lower ward was at least 2m below the modern ground surface is corroborated by the presence of a defensive loop in the eastern section of the curtain wall at a significant depth below the present ground surface.

The partial excavation of the kiln estimated an internal diameter of between 4m (east-west) and 3m.  The bowl of the kiln was lined with clay-bonded blocks of roughly squared greywacke, laid in irregular but discernible courses, which had been subjected to repeated high temperature firings, resulting in some cracking of the stones and heat modification of the clay bonding.  Due to the small size of the area investigated upslope of the kiln’s superstructure it is not possible to interpret the excavated stratigraphy with certainty.  A series of dumps of redeposited subsoil were laid against the back of the kiln on the upslope side of the structure.  Although the constraints of time meant that it was only possible to carry out limited investigation of these deposits, it is considered likely that they formed part of a ramp that was built up against the back of the kiln to provide access to the top of the kin in order to load it.

Excavation of the kiln’s bowl revealed that its fills consisted of a 1.4-1.5m deep dump of quarried stone associated with medieval pottery, which overlay a 0.6m deep deposit of heat-modified clay and burnt stone blocks that almost certainly represent the demolished remains of the upper part of the kiln’s superstructure.  Underlying this deposit was a greyish-brown sandy clay that probably represents an accumulation of soil that occurred after use of the kiln was abandoned.  This deposit was dug to a depth of 0.4m before the excavation was halted due to a lack of time.  The absence of post-medieval and modern sherds from the dump of quarried stones, which contained several sherds of medieval date, suggests that the limekiln is medieval in date and presumably relates to a major phase of construction of the castle.  Unfortunately, not enough of the bowl’s fills were removed to uncover the base of the kiln.  Consequently, it is unknown how many flues or stokeholes with which it was furnished.

 

Concluding Remarks

CAF in partnership with NIEA intend to return to Dundrum Castle for a further season of excavations in the summer of 2013.  The current targets for this work include a completion of the excavation of the limekiln within the outer ward; a cutting against the curtain wall of the lower ward with a view to establishing its date; a trench across the extra-mural terraces located immediately to the west of the lower ward and to undertake a cutting across the rock-cut ditch which surrounds the inner ward.

Centre for Archaeological Fieldwork, School of Geography, Archaeology and Palaeoecology, Queen’s University, Belfast, BT7 1NN and Liam McQuillan, Northern Ireland Environment Agency: Built Heritage, Waterman House, 5-33 Hill Street, Belfast. BT1 2LA