It was third time lucky for the members of the Longtown & District Historical Society when they learned in 2015 that their Longtown Castles Project had been awarded a Heritage Lottery Fund grant after two previous applications had failed. They now had the money they needed (£78,300) to work with professional archaeologists to excavate, study, and publish two scheduled monuments in their parish: one a motte and the other a stone-built castle sited on one corner of a massive rampart, from within which ‘ancient Roman remains’ had been unearthed in 1869.
Work began in May 2016 at castle number one, Ponthendre Motte. Led by Tim Hoverd, Herefordshire Archaeology (the county council’s archaeological service) worked with local volunteers to carry out a full survey of the site, while Adam Stanford of Aerial-Cam, whose drone photographs often feature in this magazine, took multiple digital images from which to create a 3D terrain model of the motte and the surrounding landscape. Using software to strip back trees and vegetation, this showed in crisp detail the circular mound located on the spur of a ridge on the south bank of the Olchon Brook, close to its confluence with the River Monnow. Measuring up to 44m in diameter and 10.5m in height, the motte is surrounded by a ditch and bailey enclosed by a rampart, all overlooking a stone bridge over the Olchon Brook and a cluster of small farmsteads.
The results of the geophysical work (resistivity, gradiometry, and ground-penetrating radar) – here and at Longtown – were, to quote the book that has just been published as a result of the project (see ‘Further reading’ on p.45), ‘largely uninformative and frankly disappointing: there were no obvious indications of buried buildings and those anomalies that were detected were amorphous and could not be resolved into anything recognisable’. What could excavation reveal to fill in the story of the two castles – when were they built, by whom, what activities took place there, and what relationship did the two have to each other?
Two three-week seasons of excavation took place simultaneously at both sites in July 2016 and July 2017. Over the 30 days of excavation, some 79 volunteers took part, supervised by five professional archaeologists. Trench 1 at Ponthendre was placed on top of the motte to look for evidence of a structure. Nothing was found – no post-holes or foundation trenches. If there had been a tower or keep it must have been built on sill beams placed on the ground that have left no trace. Another trench placed to pick up the remains of buildings in the bailey again found nothing – no buildings or hearths. A third trench placed to sample the defences yielded good evidence of the way the natural contours had been used to create a ditch and rampart, 6m in height, but with only one post-hole where a line of them might have been expected as evidence of a stockade. Overall, the results suggested that the motte had a single phase of construction and was never completed or occupied, that no keep was ever built, and that a timber palisade might have been intended but was never constructed.
Half a mile away at Longtown, an existing gap through the rampart was cleaned up to reveal three distinct phases. The first, 0.55m high, was built with blocks of turf; charcoal fragments from the original ground surface gave a carbon date of 355 to 117 BC. It is possible that the rampart was begun in the Iron Age, but the square shape of the rampart as a whole is more suggestive of a Roman fort, so either this was old charcoal that had lain on the turf for a long period, or it was derived from a mature tree. The second phase raised the rampart to a height of around four metres and was made up of layers of earth, subsoil, and degraded sandstone and mudstone, mostly derived from digging the ditch around the rampart. No dating evidence was found for this phase, but the third phase, which added another 0.3m of soil, contained a few sherds of 13th- to 15th-century pottery, suggesting that Phase 2 dates from an earlier period.
Trench 2 was intended to sample some lumps and bumps on Castle Green, and in the first year of excavation this produced 19th-century demolition rubble connected with the building of the nearby school. This rubble overlay the remains of a medieval road (complete with potholes!), and a possible simple timber-and-daub building associated with mid-13th-century pottery. A considerable quantity of charcoal and a complete horseshoe, as well as a flat stone that had been spattered with spots of molten iron, were all suggestive of a blacksmith’s forge.
The ditch alongside the medieval road appeared to follow the line of an earlier cutting, and from this, in the final week of the excavation, came the first evidence of Roman activity in the form of a big piece of Dressel type 20 amphora of the kind used from the 1st to 3rd centuries AD to ship olive oil from southern Spain. Soon, much more Roman pottery began to emerge in the form of 29 sherds, including 11 fragments of Severn Valley tableware, then a thick layer of burnt clay and charcoal from a kiln or oven overlying six or seven short oak planks that had been carbonised by the heat from the kiln/oven. Carbon dates for the wood gave the planks a 2nd century BC date range, but the deposit below the planks contained Roman pottery and the charcoal from this layer was dated to the period AD 24 to 125. The planks must therefore have been sawn from a mature tree that was already of a good age when it was felled, and the planks used here either as a bed for the Roman oven or as a duckboard along a wet patch of ground.
The Longtown Castle enclosure does, then, appear to have begun as a Roman fort, probably built during the period AD 50 to 78, when the people of south-east Wales (called the Silures, according to Tacitus) put up fierce resistance to Roman invasion. The fort probably went out of use at some time in the 2nd century AD, because the earthen ramparts were not replaced in stone as they were at longer-lived forts.
Elsewhere across the Castle Green there was evidence of a good deal of later medieval and early modern dumping on what had remained an open space after the 13th century. This helped to explain the disappointing geophysical results: the Roman and earlier medieval deposits are too deep to show up and are covered by multiple layers of debris. Below the dumped material, there is a good chance that earlier archaeological remains are well preserved.
Delving into documents
What, then, can be said about the relationship, if any, between the two medieval structures? For answers to this question, the project turned to historical records. Before the Norman Conquest, the border between England and Wales south and east of Hereford was fought over constantly, often by claimants to the English crown seeking to push eastwards with the support of Welsh rulers. To counter this, Edward the Confessor granted land on the border to many of his Norman friends, and some of them built the first castles to be seen in Britain. The earliest was built by Osbern Pentecost in the eastern part of Ewyas (later to be called Ewyas Harold), and another, still known as Richard’s Castle, was built by Richard FitzScrob. Clearly the symbolism of the motte and bailey was not lost on the local population who (according to the Anglo-Saxon Chronicle) saw them as ‘an injury and an insult to the king’s men thereabouts’.
Matters came to a head in 1053 when Ælfgar, Earl of East Anglia, was convicted of treason and exiled to Ireland. There he raised an army and landed in Wales to be joined by Gruffydd ap Llewelyn. Together they marched on Hereford and ransacked the town. Edward the Confessor responded by recruiting an army under the command of Harold Godwinson, the future Harold I of England, which assembled at Gloucester and marched into Wales in 1055. They camped ‘beyond Straddele’, meaning Ystrad Dwr, or what is now the Golden Valley, the valley of the River Dore. An old Roman fort would make an ideal place to camp, and Longtown Castle is in the right location to be described as ‘beyond Straddele’. The members of the Longtown Castle Project team are firmly convinced that their Roman fort was the site of Godwinson’s short-term encampment, and that the second phase of rampart-building was his work, heightening the remains of the earlier rampart to the scale of an Anglo-Saxon burh defence.
Godwinson was eventually successful: his army prevailed at the battle of Glasbury-on-Wye some months later, after which Gruffydd agreed to swear allegiance to King Edward. In 1057, this part of Herefordshire and adjacent parts of Gloucestershire were added to Godwinson’s already huge earldom of Wessex. Godwinson was an energetic defender of his realm and extended it into the Welsh kingdom of Gwent, but to no avail: his fate was to die at Hastings on 14 October 1066.
William the Conqueror appointed three of his most trusted supporters to the three earldoms that he created along the border with Wales, centred on Chester, Shrewsbury, and Hereford – important commercial centres with good river and road connections. William FitzOsbern, Earl of Hereford and High Steward of Normandy, was given all of Godwinson’s extensive territory and he set about building a chain of castles to control the main road and river routes into and out of Wales. Prominent among them were Chepstow, Monmouth, Berkeley, Gloucester, Wigmore, and Ludlow castles, but these are just the largest and best-known examples: even today the countryside between England and Wales is rich in mottes built by the Norman invaders as they sought to enforce their rule on the existing population. In the words of the Anglo-Saxon Chronicle, ‘the Normans built castles throughout the land and they oppressed the poor folk and wearied all England with their erections’.
Seven mottes were constructed or reinforced in a chain along the border between the Welsh kingdom of Gwent and the Ewyas valleys: the pre-Conquest motte at Ewyas Harold was joined by Rowlestone, Llancillo, Walterstone, Tre-fedw, Pen-y-Clawdd, Ponthendre, and Longtown. Why was the Ponthendre motte constructed when the more easily defended Longtown, with its Roman fort ramparts and its Anglo-Saxon defences, was a short distance away and in a better strategic position?
Various suggestions were considered and rejected by the Longtown Castle Project team. One was the theory proposed by Alfred Watkins, author of The Old Straight Track (1925), that Ponthendre was not a motte at all but a prehistoric surveyor’s platform, constructed for the laying out of ley lines. Another was the idea that Ponthendre might be a ‘spare’ motte, for use in case any of the others were captured. A third dated the motte to the Anarchy of 1135 to 1153, the civil war between Stephen of Blois and the Empress Matilda. Another suggested that the motte was a make-work scheme to keep the conquered peasantry out of mischief, and another said that an uppity tenant decided to build his own motte at Ponthendre without permission and suffered the consequence.
None of these castles is securely dated and there are no documentary records to help us understand the building sequence, but their regular spacing at intervals of 1.5 miles suggests an integrated plan of defence and simultaneous construction. The exception is Ponthendre, a mere half mile from Longtown – it looks anomalous and superfluous, unless the Ponthendre earthworks were built before the motte at Longtown was established. Perhaps, the team members concluded, the Longtown site was already occupied and defended by forces hostile to the Normans. If so, the Normans faced the choice of attacking a well-defended site and perhaps suffering heavy casualties, or they could build a new motte at Ponthendre and wait for an opportunity to clear the fort at Longtown – after which the Ponthendre motte would no longer be needed.
The kingdom of Gwent did not long survive the Conquest: it is clear from the Domesday survey that it had capitulated to Norman control some time before 1086. It is reasonable to believe, therefore, that the Ewyas chain of border mottes was constructed in the fast-changing early years of the Conquest. Ponthendre was not the only motte to have been abandoned incomplete. The earthworks at Rowlestone and Pen-y-Clawdd are 4m and 2.5m high respectively, and construction might well have terminated before they reached their full heights. The same might well be true of any number of other mottes in the Marches: constructed quickly in the face of local resistance, they were abandoned once the need had passed, while those that were located in the best positions to guard routes in and out of Wales and Hereford were subsequently rebuilt in stone on a much larger scale.
In the case of Longtown, that resulted in today’s surviving round keep, standing 16m high on its 11m-high motte. Much robbed of its stone by those who built Longtown’s later houses and farms, the keep is buttressed by three semi-cylindrical turrets, a unique feature. More usual would be a single turret, containing a spiral staircase. Here there are two additional projections that serve no practical function, for they are largely built of solid stone.
Secrets in stone?
But when was Longtown’s keep built? The members of the Longtown Castles Project team set themselves one final challenge: to work out who built this structure and when. Previously historians had attributed it to Walter (1189-1241), the last of the de Lacy dynasty, and to the final years of his lordship. Walter was, however, heavily in debt, distracted by events in Ireland, where his vast estates brought him much trouble and expense; it seems unlikely that he could have afforded to build a new castle at Longtown or that this would have been a priority.
Much of the discussion about the date has focused on the decorated stones over the north window, typical of the Herefordshire School of Romanesque Sculpture, active in the second quarter of the 12th century. These stones are clearly not carved for this location, because they are all but invisible and their use in a relieving arch would have concealed their decorated undersides. One of the stones is incomplete and another is unfinished, which suggests that the stones were intended for use elsewhere but rejected and subsequently reused here.
If that reuse occurred soon after the stones were carved, the keep was built in the mid-12th century, rather than the first half of the 13th. The lord of Ewyas at that time was Gilbert de Lacy (d. 1163). Gilbert was a Knight Templar, known for their round churches that mirror the plan of the Church of the Holy Sepulchre constructed over the site of the Tomb of Christ in Jerusalem. Before he went on Crusade in 1158, Gilbert built the round chapel at Ludlow Castle, bearing a remarkable resemblance to the keep at Longtown.
Could evidence be found to confirm that Gilbert also built the keep at Longtown? Scouring Britain and France for references to round keeps, the closest parallel the team could find was that of Houdan, in the Île-de-France, built between 1120 and 1137 by Amaury de Montfort (1192-1241), brother of Simon de Montfort, 6th Earl of Leicester, who led baronial opposition to the rule of Henry III. Gilbert named his youngest son Amaury as a sign of the closeness of his relationship with de Montfort, a fellow Crusader. There is little doubt that Gilbert knew the castle at Houdan, and the main difference between them is that Longtown has three turrets to Houdan’s four. But then, the tower has three rooms one above the other, the main room is lit by three equal windows, and there are three waterspouts between each turret.
The Romanesque decorated stones and the parallel with the Ludlow point to Gilbert de Lacy having begun rebuilding Longtown Castle in stone around 1150. If it was not finished by the time he went on Crusade, it may have been completed by his sons Robert and Hugh de Lacy, but the conception was his, along with the possible symbolic references to the Trinity appropriate for a member of the military and religious order founded to protect pilgrims travelling to the Holy Land and to defend its holy places.
Martin Cook and Neil Kidd, The March of Ewyas: the story of Longtown Castle and the de Lacy dynasty (Logaston Press, ISBN 978-1910839478, £12.95, www.logastonpress.co.uk).