Castell Henllys is a small inland Iron Age (600-100 BC) promontory fort and adjacent native farmstead occupied in the late Iron Age and throughout the Roman period (100 BC - AD 400). It has been under excavation since 1981, and almost all of the interior and the entrance of the fort has been examined, together with a significant section of the farmstead. The fort is one of the most extensively excavated ever in Britain. The on-site building reconstructions are unique in Britain and include the longest-standing reconstructed roundhouse and granary. The excavations are run so as to provide practical training, and Castell Henllys is now the longest-running and largest such excavation in Britain.


Castell Henllys is 5km (3 miles) from Newport , and is situated on a small spur overlooking the river Gwaun, which is a tributary of the river Nevern. The fort is about half a hectare (1.25 acres) in area, defined on the east, south and west by steep scarps. The edge of the site on these sides is defined in places by a small Iron Age bank, and on the south and part of the west slopes a man-made terrace also forms an element in the defensive scheme.

To the north there are major earthworks, consisting of a large inner bank, ditch, smaller outer bank, and further ditch. These curve round to meet the scarp on the east, but on the west they merge at the point where the entrance lay. From the southern end of the site there are fine views both along the valley and across it to the Presely mountains and Carn Ingli. To the north, watch could be kept along the subsidiary valleys and as far as a steep scarp on to the plateau 200m (220 yds) away.

 On the gently sloping flat ground 30 m north of the fort lies another, outer bank with a ditch beyond. Beneath this bank an earlier Iron Age defence, a narrow band of upright stones set firmly in the buried soil, has been preserved. This is called a chevaux-de-frise, and is rare in Britain.


Early history

 The site had never been forgotten from prehistoric times, with the Roman period occupation next to the fort, and a probable post-Roman brief re-occupation of the site, perhaps in the 5th century AD. It is mentioned in medieval documents, and has long been marked on Ordnance Survey maps. The fort interior lay on agriculturally marginal land, rarely ploughed even in recent times, though the area to the north of the main defences, including the outer bank and ditch, was ploughed more often.


First archaeological notice

 The site was first described as unfinished, because the northern defining banks and intervening visible ditch were recognised as substantial to the north-east, but tailed away to the west (Royal Commission for Ancient and Historic Monuments of Wales (RCAHMW), County of Pembroke, 1925). The outer bank, noted on Ordnance Survey maps, was not given particular attention.


The Aims of the Current Excavation and Reconstruction Campaign

 To excavate complete roundhouse and other building plans so that reconstruction could take place, and gradually extend to excavate the interior of the fortTo reveal the full sequence of defensive development on the northern side, and examine the defences elsewhere round the perimeter of the fortTo excavate the Roman period settlement and consider its relationship to the fort and any contemporary settlements in the regionFor all periods to examine the subsistence economy and social structure


 Open area excavations were undertaken at the southern tip of the site, and immediately behind the northern rampart to provide two contrasting areas for excavation. It was in the southern area that remains suitable for reconstruction were encountered, whilst a more complex sequence of buildings was uncovered to the north.

 Round-house 1 was located in the south-eastern part of the site, and was the first to be completely excavated, during the summer of 1981, and reconstructed over the following winter and spring. It was 10m (33ft) in diameter, and is the only double-ring house so far recognised with its inner ring of posts. The wall line could be discerned on the uphill side by several phases of a shallow slot. Two post-holes to the south marked the position of the door. Outside the wall on the north-west and west sides was a wide, shallow eaves-drip gully. Roughly in the centre of the round-house was a hollow filled with rocks, probably the base for a hearth, with pairs of posts either side to support a spit.

 Round-house 2 was a much smaller structure, only 6m (20ft) in diameter, and had suffered badly from erosion in the south-western part of the site. As with Round-house 1 only the uphill parts of the wall gullies survived, but again it was multi-phase. No internal features could be identified. The lack of door posts suggests the use of a continuous wall-plate to unite the top of the wall in a horizontal ring on which the roof could sit, and this solution has been used on the reconstruction of the building. The pitch of the roof has been made less steep than that of Round-house 1, but it works just as effectively.

 Round-house 3 lay to the north of Round-house 1. It was also about 10m (33ft) in diameter, but had no internal ring of posts and the wall foundation reflected this. The foundation was a continuous wall trench, with a depth of up to 0.4m (1ft 6in) on the downhill, southern side of the building whilst on the north this was only 0.1m (5in). The doorway was marked only by a break in the wall trench on the eastern side and this indicated an unusually wide door.

 Roundhouses 1, 2 and 3 have been reconstructed on their original sites, the only such houses in Britain.


 Behind the northern defences occupation deposits survived, sealed by the wash from the bank. Some of the deposits contained much carbonised bone, fire-cracked rock and charcoal, as well as a few iron objects, pottery and glass beads. A layer with fire-cracked rock also ran up on to the rear of the bank, providing an important stratigraphic link between the defensive and occupation sequences.

After the initial phase of excavation, gradually more areas of the interior were investigated, concentrating on the eastern and western sides, and then after removing the large spoil heaps in the centre of the site, the middle spine of the site. Work uncovered further roundhouses, one of which was reconstructed. This was a unique design not found elsewhere in Britain and specially designed for the topographic peculiarities of the site. In addition, previously unsuspected roundhouses set in scoops were found, with very well preserved floors. There was also evidence for four-post granaries, one of which has been reconstructed

 The north-western part of the interior has proved to be one of the most interesting of all, with the discovery of very deeply stratified deposits. Up to one metre depth of Iron Age levels is extremely unusual anywhere in Britain, and this has allowed a long sequence of developments in this part of the site to be excavated over several seasons; work still continues on the very lowest levels.

 One of the most important discoveries in the interior area just inside the entrance has been the long sequence of roundhouses constructed on the site. Many were of substantial trench construction, though some had flimsy wattle walls set in shallow gullies. Most were around 10 m in diameter, though some of the earlier ones were smaller.


External hearths have been found, some with a more than usually elaborate construction involving the use of shale slabs on which the clay hearths were placed. Some hearths were rebuilt several times, and were surrounded by numerous stake holes indicating the use of spits and cauldron supports for cooking.

 A four post granary has also been identified late in the sequence, as elsewhere on the site. A corn-drying oven was found and this has produced a wide range of carbonised seeds, mainly from various cereal crops grown in the area and brought to the site for storage. So far no samples indicate the preliminary stages of processing of grain, such as threshing, on the site.

 The lack of evidence for agricultural processing but only final processing stages, storage and consumption may indicate that only high status individuals lived within the fort. The foodstuffs may have been brought as tribute payments from the surrounding countryside where most of the lower status families lived. In this context, the labour force for constructing the defences and the gateways was also probably brought in from the locality, albeit under expert supervision. The provision of stone for the gateway, and timber for houses and defences, must also have been brought in with considerable effort; there clearly were large areas of managed woodland, and detailed knowledge of where various resources could be found in the region.

Behind the rampart a massive slingshot hoard has been uncovered, and a student from the University of York is present measuring and weighing a sample from it to allow assessment of the effectiveness of these weapons. Scatters of slingshots were found over the interior in this part of the site, and research is being carried out to see if these stones derived from the large hoard or arrived at the site at a different time. This hoard is the largest recovered from a Welsh site that is available for modern study; indeed, very few slingshot hoards have been preserved from anywhere in Britain.

 Finds have been few, but have included whetstones, perforated stones, a few sherds of pottery, glass beads, and several spindle whorls. Indeed, more spindle whorls have been found in the north-western part of the site than all the rest of the fort interior put together.

 One of the most intriguing features excavated on the western side of the site has been an irregularly shaped ditch which runs round from the entrance along the western side of the site. It was very late, and contained a Roman brooch and a sherd of Roman pottery, the only such finds of the Roman period from the whole of the fort.

The ditch may be of late Roman or more probably post-Roman date, and reflect the refortification of the site, perhaps in the 5th century. A piece of stone walling on the eastern side of the entrance was also shown to be very late, and may be related to the ditch. This period is of particular importance in Pembrokeshire's history, since it relates to the origins of the Irish kingdom in north Pembrokeshire. Significant high status Irish settlement in the area is well attested, for example by the famous ogham inscriptions at Nevern. Further elucidation of this critical but difficult period of history may be achieved by the continuing work at Castell Henllys, and where necessary at other sites in the area.

 The refortification could be in response to the arrival of a new elite, and the abandonment of the Roman period settlement, or it may have just reflected more unstable times with the decline in Roman influence. This decline could have occurred before the Romans formally left Britain, and indeed the date of the Irish arrival is a matter of controversy.



Artefacts from the fort have been relatively rare. A small selection of pottery fragments, a few decorated with small stamped patterns, have been recovered, and a few copper alloy objects have also been found. Very little iron was thrown away, though a small hunting spearhead was recovered, as was a sickle blade. Some evidence of iron and copper-alloy working on the site has been demonstrated, with the presence of slag and crucibles. A small selection of glass beads and a fragment of shale bracelet are also notable discoveries. The most common finds have been of stone, including numerous perforated shale objects in a wide range of sizes, spindle whorls, whetstones and some flint flakes. Only two fragments of one quernstone have been found.




A single trench 2m (6.6ft) wide was excavated through the inner bank, ditch and outer bank. At a later stage extensive excavation of the rear of the inner bank took place along most of its length. This indicated that the sequence of development was not the same all along the length of the inner bank, suggesting very complex patterns of maintenance and reconstruction.

 Careful survey of the defensive perimeter revealed the extent of work on all sides of the promontory and on the flatter area to the north. Selective excavations were carried out to provide a basic understanding of these features.

 A trench was excavated on the southern edge of the site, through the bank and down the scarp slope to the terrace below. The effect of the terracing was to create, when seen from below, the appearance of a massive bank, though most of this was the scarping of the hilltop. A palisade ran along the rear of the bank, and a more substantial one intermittently along the top. The remains of a hearth made from large shale slabs was found on the rear of the bank, and charcoal and iron smithing slag surrounded it. The residues suggest the occasional, small-scale repair of iron artefacts though not the smelting of iron ore.

Some excavation took place beyond the fort, to the west of the Roman period settlement. A substantial north-south ditch was located and excavated, and both ends found. A palisade trench was also located, though it clearly belonged to a different period. As neither feature produced any Roman material, both were assigned to the Iron Age.

 A trial trench through the most northerly visible earthwork revealed a bank belonging to the Iron Age period, with a substantial ditch beyond it. Beneath the bank, now substantially eroded by ploughing in recent centuries, was a chevaux-de-frise which would have provided an effective defence in depth on the gently sloping central spine of the spur. Only one other local example is known in the region, at Carn Alw on the northern slopes of the Presely Mountains 6km (6 miles) from Castell Henllys.



 A long and complex sequence has been identified at the entrance, including some phases involving what were elaborate architectural features for their time. Excavation has been very extensive so that the whole of what survives for each period of construction can be considered at the same time. Numerous stone-by-stone plans have been produced which provide a unique record of what was found and form the basis for analysis and any reconstruction work. This has made the excavation slow and painstaking, but the quality of the results has fully confirmed that this has been the appropriate approach.

 The entrance sequence is still being excavated, and the results of previous seasons have not been fully analysed. Therefore the sequence presented here is very much a provisional one, though for the more substantial stone phases greater confidence can be expressed in the interpretations given here.

 The earliest entrance way has yet to be fully revealed, but seems to have included a line of deeply set posts, set close together, leading to a simple timber gateway. A small bank was thrown up behind this, but quickly this was not considered sufficient and a larger bank was constructed, with the existing timber line being partially buried within the bank which was largely built in front of the fence. A substantial ditch was dug outside the bank to both provide the necessary materials and to enhance the defences. At a later date, a line of quartz boulders was placed along the passageway on the western side. There was probably a cobbled stone roadway leading into the fort at this stage.

 The first major stone phase consisted of a unique arrangement, consisting of two pairs of guard chambers flanking a roadway, with a large timber tower with wooden gates hung from the uprights. Slots in the gateway walls were constructed so that massive timber beams could be slid into position to keep the doors firmly closed. Beyond the gateway to the north, leading out of the fort, flanking walls funnelled people and animals towards the entrance. A cobbled stone surface was constructed leading up through the entrance and into the interior.

 Massive stone walling was constructed on each side of the entrance, and some of this survived to a metre (over 3ft) in height, though all has now been removed as the earlier gate constructions buried beneath have been investigated. All this investment seems to have been carried out in about the 5th century BC.

 After a while, the gateway fell into disuse, and some of it was destroyed by a very intense fire that led to some of the shale turning into light, bubbly slag-like material. This may have been during an attack on the fort, or may have been a deliberate act by the inhabitants, perhaps as part of a deliberate demolition of part of the gate, or as some ritual act. The gateway certainly decayed over some time, before a second elaborate gateway in stone was constructed.

 The second major stone phase cut away much of the collapsed rubble and parts of the original walling to form flat, solid bases for the new walls. These formed a very different shape from the previous period, with a single pair of shallow guard chambers and convex walling outside which curved round the front of the massive bank on the west, and round the outer bank on the east. A four-post gate tower was constructed, and two massive post holes outside the gateway either represent an extension of the tower or are a bridge or walkway for defenders on the outer bank on the east to cross over to the massive stone structure on the west of the entrance way. A new surface was made through the gate, and once again Castell Henllys must have looked most impressive.

 The gateway again subsequently decayed and fell down, and various timber gate towers were built on the rubble.



 A large open area excavation was begun in 1981, and continued until 1986. During this time a boundary ditch, redug many times, was excavated on the western side of the site, and an entrance to the settlement located there. Within the settlement was a large hollowed out area, used for keeping animals, though this gradually filled up with rubbish.

The rubbish consisted of charcoal, small fragments of burnt clay from hearths, ovens and buildings, burnt bone, iron nails and tools including knives, a spade and a mattock, a few copper-alloy objects, notably two brooches partially surviving at the very bottom of the hollow, glass beads and vessel fragments, and pottery. The ceramic material indicates a wide range of contacts by way of the Irish Sea rather than overland to the nearest Roman town at Carmarthen. Particular significant was Samian pottery (from France), wine amphorae (from Spain), black-burnished jars and dishes (from Poole, Dorset), mixing bowls (from Oxfordshire and Warwickshire), and a range of other Malvernian and Severn valley ware jars and bowls ( from various pottery production centres in south-east Wales and the Marches). Evidence of craft activity in the form of metalworking slag was also recovered. A large number of quern fragments were found, some of the tall, beehive shape of the Iron Age tradition, others much flatter and thinner, in Roman style.

 The most spectacular find was a seven-pronged iron object, subsequently cleaned and conserved at the National Museum of Wales and on display there is carefully controlled conditions due to its fragility. The object, unique in the Roman Empire, was a beautifully crafted object, perhaps used in cooking, though its exact purpose still remains obscure. However, it and all the other finds suggest a degree of sophistication rarely found on native sites of this period in Pembrokeshire, and so implies that the settlement was that of a high status family.

Two massive stone causeways led across the muddy hollow from the entrance, one to a small timber roundhouse near the outer bank, the other towards the area in front of the main Iron Age banks, though the area concerned was not investigated at this stage. Three pits, originally lined with wicker, were found in the centre of the site, and a four-post granary was also excavated. Inside one of the post holes was a complete rotary quernstone.

Near to the visible Iron Age defences, a large ditch belonging to the Iron Age period was filled in, probably rather quickly from the evidence of subsequent subsidence, and several timber roundhouses built on the site. At least one of these may belong to the late Iron Age.

 Part of the area was later marked off by a major drystone wall, probably constructed in the 2nd century AD. There may have been a timber rectangular building at this time also. The layout of this early Roan period settlement is as yet still very tentative, but is extremely important as it has some signs of greater Roman influence than appears in the subsequent developments at the site.

 At a later date, in the 3rd century AD, a slightly oval timber house was constructed within which various craft activities took place. A large amount of pottery, glass, slag, burnt bone and carbonised grain has come from well preserved floor deposits inside this building.

 Storage facilities in the form of four post and six post granaries were present on the site, some built over the collapsed remains of the 2nd century AD wall. One of the six post granaries was burnt down, and large quantities of grain were recovered from the site. The granary was almost immediately rebuilt, but with far less substantial timbers.

On the western side of the site, a ditch dated late in the Roman period, and perhaps to be associated with the late refortification of the fort itself, has been uncovered and almost completely excavated. It has produced a very wide range of finds, including pottery from the 1st to the 4th century AD, some of which must have been moved from old middens and used to fill up the ditch when it was no longer needed.

 The long and complex sequence of buildings will now be integrated with the earlier findings for the animal stockyard hollow and the stone causeways, so that the overall layout of the settlement can be indicated. It clearly changed significantly in appearance over the four centuries or so of occupation. The area to the east, yet to be excavated, is of great importance. Every part of the settlement so far excavated has revealed different types of structures. There are domestic areas, storage areas, craft area, and a stockyard area. Yet other areas may exist to the east, and these need to be excavated to provide an overall view of the settlement.



 During most of the Iron Age, Castell Henllys was a community of many families, and may have contained a population of 100-150 people. Even so, the work involved in constructing and maintaining the defences would have been considerable. There can be no doubt that there was a concerted effort to provide substantial defences for the settlement on all sides. However, there are two main alternative interpretations of this evidence. The first is that defended communal settlements such as Castell Henllys were necessary within a violent society where mutual protection was paramount. The second is that the defences were primarily for display, indicating that the site was a high status residence and the earthworks a form of conspicuous consumption. The two interpretations are not necessarily mutually exclusive; the heightening of the inner northern bank, the choice of terracing to enhance the defences on the south and west, or the chevaux-de-frise could be for either purpose.

 The settlement was densely packed with round-houses, with the addition of four-post structures late in the sequence. The scoops on the east and west were probably in use at the same time; stamped pottery of the Croft Ambrey-Bredon hill style found in eastern Wales was found by the hearth in a scoop on each side of the promontory.

 Carbonised grain has come from a wide range of contexts, indicating a mixed economy throughout the occupation of the site. However, there is accumulating evidence to suggest that cereal production increased in importance towards the end of its life. The only two fragments of a single quernstone from the whole of the settlement were found in the latest phase of Roundhouse 3, and all the four-post structures which can be phased are in the later periods. Only burnt bone survives in the acid soils which cover most of the site, though a few deposits have yielded more. The burnt bone is mostly that of sheep, but this is probably due more to Iron Age cooking practices and bone survival than directly reflecting relative importance of sheep in the Iron Age. If sheep were roasted and beef boiled, then the chances of cattle bone becoming burnt and surviving in the acid soils would not be great. In a few layers evidence for pig, cattle and horse has been recovered, together with two deer antlers used as picks. Metal-working was undertaken in the southern and south-eastern parts of the site. External contacts are indicated by the presence of a few glass beads, a small quantity of non-local pottery and some small fragments of bronze that have survived.

 The sequence at the entrance indicates periods of massive investment on the gateway, involving the use of specialist builders or architects in some cases. The scale of elaboration in the case of some of the stone phases shows links with other parts of Britain. It also highlights the desire to present a fashionable and sophisticated appearance to the fort, and the resources to produce a monumental construction comparable to that on much larger forts elsewhere. That such sophistication is not maintained continuously through the fort's history may be an indication of periodic shifts of power within the region, or of problems in maintaining such a level of consumption on display features over the long term.

 The fort was subsequently replaced by smaller, less defensively enclosed farmsteads, of which the Romano-British settlement at Castell Henllys is an example. This indicates a dispersal of the population across the landscape, instead of being concentrated in the defended settlements.. Small enclosures are known north and south of Castell Henllys, and a further possible example has been identified by aerial photography in 1995 to the east. The site to the north has been partially excavated, and that to the south subjected to geophysical and surface survey; the newly discovered site was subject to preliminary investigation during 1996 and further work will take place in 1997 if resources allow.

 The settlement shift in the late Iron Age may be associated with a decrease in small-scale warfare, and a move to greater arable production, both linked to widespread social and political change. This may be compared with various political and economic changes in southern Britain which were the result of contact with the expanding Roman Empire. It would seem that Pembrokeshire elites were not as isolated from the major trends happening in the rest of Britain and Europe as has previously been assumed.

The Roman period evidence from the settlement indicates to what extent the elite of Pembrokeshire were influenced by Roman fashions and culture; they used a range of manufactured goods not previously available, but they continued to build and live in roundhouses, and store grain in granaries. The scale of production, and perhaps local taxation and tribute, seems to have been increased during the late Iron Age and Roman period, perhaps to allow exchange of the manufactured goods, and also to pay taxation to the Roman authorities after the conquest. This stimulus to production does not seem to have had an adverse effect on Castell Henllys and its inhabitants, though perhaps those being pressed to pay more out of their produce may have experienced more hardship. Only further excavation of other contemporary sites would help to resolve this. The pattern of Ramon contact and native reaction at Castell Henllys is an important case study which contributes to a widespread debate on the role and effect of the Roman Empire on indigenous peoples not only in western Britain but elsewhere in the Empire.

 The possible post-Roman reoccupation of the fort can be parallelled elsewhere in western Britain, but rarely has much evidence been found for this. The proximity of Meline church with its early features including the saint's name, proximity to water and circular churchyard, is interesting and can be parallelled at Nevern. A now-lost but recorded early memorial stone from the parish of Meline may well come from the church, though St Fraid's well just north of Castell Henllys has also been suggested, even though this is in Nevern parish. In both cases these probable early religious centres may have been placed in proximity to an early medieval power centre, based at Castell Henllys.

 Excavations at Castell Henllys have begun to break down old assumptions about Iron Age, Roman period and post-Roman (early medieval) Pembrokeshire and its relationships with the rest of Britain. As work proceeds, further reassessment will undoubtedly be necessary, but the efforts are clearly worthwhile.



 The Castell Henllys project has been supported by the University of York, University of Newcastle-upon-Tyne, British Academy, Society of Antiquaries of London, Earthwatch, Texodus Ltd, and the Pembrokeshire Coast National Park. I would like to thank all who have helped on the excavation and reconstruction, and particularly Ken Murphy of the Dyfed Archaeological Trust for acting as Assistant Director in recent years.

 Numerous individuals have helped supervise work on particular aspects of the site, and even more have taken part in the excavations, many coming back year after year. The importance of their devotion and hard work cannot be underestimated, and the present status of the project is dependent on all their efforts.


Background Reading

 Some people will already have attended University,evening classes or other courses; others will not. No prior knowledge is assumed, and training is tailored to individuals' levels of experience and knowledge. If you would like to carry out some preliminary reading, here are some suggestions. I have not given years of publication since many of these books have gone through several editions.




Kevin Greene, Introduction to Archaeology. London, Batsford. Paperback.

 Colin Renfrew and Paul Bahn, Archaeology. Theories, Methods and Practice. London, Thames and Hudson. Paperback

 Philip Barker, Technqiues of Archaeological Excavation. London, Batsford. Paperback.



 Barry Cunliffe, Iron Age Communities in Britian. London, Routledge. Hardback and very expensive!

 Barry Cunliffe, The Iron Age. London, Batsford. Paperback.


Information supplied by HCM; last updated March 1997