Medieval Climate Change at Caerlaverock – Fieldwork Week One

Project lead, Dr Richard Tipping, looks back at the first of the two weeks of fieldwork he and Dr Eileen Tisdall are undertaking to try establish whether climate change did impact the re-positioning of Caerlaverock Castle.

Well, somewhat later than planned or advertised, the first week of fieldwork at the old castle at Caerlaverock began on 3rd July. It has been dated by oak tree rings from the drawbridge across the moat to c. AD1229. The new castle, a couple of hundred metres away and upslope, began construction only 50 years later. Why the rebuild? This is the mystery.

Subscribe to our quarterly newsletter

An image from a LiDAR survey, with the vegetation removed, showing the landscape and old coastline around the ‘old’ and ‘new’ castles at Caerlaverock

The natural environment played a role. Some archaeologists think the old castle started to fall apart because it was built on soft sediment, a deep clay-rich silt locally called ‘carse’. And sure enough, the new castle was built on the nearest outcrop of bedrock. But twenty years ago, a team of environmental archaeologists from the Universities of Stirling and Coventry found an additional reason for abandonment of the old castle. They found, in a wide ditch next to its moat, a thick layer of grey estuarine-marine mud penetrating the brown freshwater peat of the ditch. This suggested that very large storm surges impacted the environs of the old castle. Further work found very large gravel beach ridges, thrown up in these storms, stacked against the old shoreline.

That team included Dr Richard Tipping and Dr Eileen Tisdall from the University of Stirling. Now we are back to find more evidence and improve the dating of the sediments funded by the Castle Studies Trust, as Historic Environment Scotland, the custodians of both castles, seek to update what we know of Caerlaverock and improve the visitor experience.

Dr Tipping using the ‘simple’ corer to extract sediment cores

For a week, 3rd-10th July, Richard Tipping stuck tiny (2.5cm diameter) holes in sediments in the ditches and moat with a simple corer, a metre-long metal gutter with a handle on the end. Push it in, turn it around to cut out a sample and pull it out, together with a metre of sediment. Record the sediment and then repeat, usually in a straight line called a transect. He did this over 80 times in the week. Isn’t science exciting? It wasn’t too hard: the sediments in the ditches and moats are less than a metre thick before the ‘natural’, as archaeologists call it, the sterile silt of the ‘carse’. But because the sediments are so thin, and because we knew that they recorded a lot of environmental ‘events’ in a short period of time, logging the cores in detail took time. A challenge, too: there are only so many shades of grey or brown.

Dr Tipping logging the details of a sediment core

Coring started in the artificial ditch that runs parallel to the moat. This was where we first began, twenty years ago, to realise that something funny was going on; in fact, several funny things. Why, in the first place, dig a 60 m long, 3 m wide ditch alongside one side, and only one side, of the moat? Twenty years ago, radiocarbon dates on freshwater peat at the base of the ditch hinted that this ditch was actually earlier than the old castle, maybe 200 years earlier. We aim to clarify this. Coring found the same sediments we found last time, which is always reassuring. Peat at the base of the ditch was replaced by grey silt. This was sealed by more peat. There is more grey silt at the end of the ditch nearest the coast, although the upper layer penetrated the full 60 m length.

We could not understand, twenty years ago, the filling of the moat with sediment: too few boreholes, too far apart. This time, cores were spaced 1.5 to 3 m apart along a 40 m long line. Archaeologists generally excavate and record continuous sections. We can’t do this because the site is a Scheduled Monument, and besides, the sediment is under water (though this week mercifully dry underfoot), so we construct our section from transects of cores, ‘joining up the dots’ by correlation. For the first time we can show that the moat was probably affected by the same environmental impacts that hit the parallel ditch. This is only ‘probable’ because radiocarbon dating has yet to show the events are the same. But grey silt entered the moat from the coast, pushing up the moat, probably eroding earlier-formed peat but not penetrating the full length. At the end farthest from the coast, near the bridge, peat continued to form.

A line of flags showing the transect of boreholes in one arm of the moat, approaching the footbridge to the old Castle, itself precisely where the original drawbridge was.

Our cores also revealed two new aspects of moat construction. The first is that the constructors took a shortcut in making the moat. In two arms of the moat, they had incorporated natural stream channels in the moat, twice as deep as the rest of the moat. This was a way, of course, to ensure the moat was under water. But away from these channels, the ‘natural’ lay only 20-40 cm below the present sediment surface. If the water surface then was as it is now, the depth of water would have been barely above the knee. You might have waded to the castle rather than use the drawbridge.

Coring in parts of the moat also hit stone at shallow depths. The carse itself is stoneless, and so the stone was emplaced by people. Stabbing around with the corer revealed these stones to be large blocks. They are concentrated at one corner of the moat, nearest the coast. Maybe they were buttresses put in to strengthen the foundations of the castle, either as it started to subside, or maybe undermined by storm surges.

Now I’m thinking and planning the next part of the campaign, straightening my back, and hoping the insect bites subside: the joys of fieldwork. From 17th July we start again, for a week, away from the old castle and out onto the old beach ridges and the basins formed by them. In the early 2000s, tree-felling of Sitka spruce allowed us to see what we were doing. Now it’s a dense tangle of semi-natural oak woodland and not easy to move around in. Wish me luck!

Subscribe to our quarterly newsletter

Haverfordwest’s town wall revealed?

Neil Ludlow and Phil Poucher of DAT look at the results of the investigation at Haverfordwest Castle by Dyfed Archaeological Trust (DAT), as part of a major infrastructure scheme embracing the castle and its setting, has revealed what may be part of the medieval town wall, long thought to have been entirely destroyed.

The remains of the castle still dominate views of the town, particularly from the main eastern approach, crowning a steep bluff overlooking the Western Cleddau river. Founded around 1110 by one Tancard, a Flemish colonist, the castle appears to have begun as a partial ringwork and bailey, perhaps adapted from an Iron Age hillfort. Fortification in stone began under Tancard’s grandson Robert FitzRichard, a decade or so either side of 1200, with the erection of a subrectangular donjon; a curtain wall with at least one round mural tower was later added, possibly by the younger Marshal earls of Pembroke between 1219 and 1245. The castle was transformed into a palatial residence with the addition of an integrated suite of apartments of the highest quality, including hall and chamber-block ranges, and a terraced garden enclosure; they are traditionally attributed to Edward I’s queen Eleanor of Castile who received the castle and lordship two years before her death in 1290. The outer ward was also walled in stone, probably during the early fourteenth century. Although it played no part in the second Civil War of 1648, the castle was partially slighted on Cromwell’s orders and was subsequently used as a gaol, which closed in 1878.

Open to the public since 1970, and housing the town’s museum and County Record Office – but still perhaps an under-valued asset – the castle is now the subject of an enhancement programme to improve access, carry out essential repairs and redevelop the museum. The scheme extends to the castle’s setting, with improved landscaping and restoration of the surrounding burgage-plot boundaries. Preliminary archaeological work includes geophysical survey and test-pit recording.

Figure 1: The walling and archway from northeast. The castle donjon is at far left.

Investigating the castle exterior in early 2021, at the summit of the steep bluff, Andy Shobbrook of DAT came upon a stretch of walling that appears to have evaded previous investigations. Now of no great height, but probably truncated, it is pierced by a wide segmental arch of convincingly medieval form (Fig. 1). Although absent from published plans and descriptions of the castle, it is shown on the large-scale 1:500 map of the town produced by the Ordnance Survey in 1889, on which it is labelled ‘Arch’ in the Gothic script reserved for antiquities (Fig. 3). It lies just within the scheduled area of the castle, corresponding with its boundary, and appears to be in a stable condition.

Figure 2: Plan showing the conjectured layout of Haverfordwest in c.1300

The walling may be part of the medieval town wall rather than the castle defences. The town of Haverfordwest, which is notable for its three medieval parish churches – unique in Wales – was founded soon after the castle and by the close of the Middle Ages had become the de facto county town of Pembrokeshire. Defended by an earthen bank and ditch from an early period, probably before 1200, it was walled in stone after the issue of a murage grant in 1264. The defended area was relatively small, immediately next to the castle and always known as the ‘Castleton’ – while the extensive suburb around the extra-mural marketplace to the south received fortified gateways, they were never connected by any solid barrier (Fig. 2). The town wall had largely disappeared by 1700 and, while the gatehouses survived rather longer, the last were removed at the end of the eighteenth century.

Figure 3: An extract of the Ordnance Survey 1:500 map of Haverfordwest, of 1889,
showing the castle and walling (labelled ‘Arch’).

Vestiges of the wall were apparently still detectable in 1900 but all traces were thought to have been lost soon afterwards. Stretches of its former line are marked by property boundaries but its entire course is not precisely known, nor the points at which it connected to the castle defences. The walling discovered in 2021 butts against the donjon at the northeast corner of the castle inner ward, and runs northwest for 5 metres before petering out. The remains of a return at its northwest end correspond with a 90° turn shown on the 1889 map, on which it is shown to then run north-eastwards before turning west to continue along the outer edge of the castle’s northern ditch. But the medieval wall must have deviated from this line at some point, to run northwards to the eastern town gate. The arch is 3 metres wide but was probably always too low – and perhaps too wide – to represent an entry. Its function may simply have been to drain the area immediately to the west, which slopes steeply downhill towards the east and seems to have been a continuation of the castle ditch where it ran out at the crest of the bluff (Fig. 3). Two phases of work within the arch are possible, suggesting it was modified and perhaps narrowed at some point.

Subscribe to our quarterly newsletter