Castell Dinas Brân

Castell Dinas Brân is a ruined medieval castle on a steep isolated hill above Llangollen in Denbighshire, where the Dee valley meets the approach from the Horseshoe Pass. The county council owns and maintains it, with help from the Welsh heritage service. It is a scheduled monument. The hillfort and the later castle are recorded together: a later prehistoric, probably Iron Age, univallate enclosure, and a stone castle built in its north-west part during the 1260s by the lords of Powys Fadog.

Grid figures given for the site are SJ 2224 4306, SJ 22240 43060, and an easting of 322330 with a northing of 343037. The summit stands at 321.4 metres, or 1,054 feet. The ruins are open all year. The routes up are steep and exposed, and visitors are advised to wear stout walking shoes and warm, waterproof clothing.

The ground is not a level entrance. The gatehouse has been made sound enough to be opened at certain times when access is appropriate. There are no facilities on the hill itself.

The name Dinas Brân has been argued since at least the seventeenth century. In modern Welsh, dinas can mean a city, and a translation as the City of Crows is treated as a mistake. In Middle Welsh, dinas means a defended enclosure. Brân is the Welsh for a crow, so one reading is the crow’s fortress.

Humphrey Llwyd and William Camden both suggested that Brân was a personal name taken from a Gaulish chieftain, Brennus. Humphrey Llwyd, also spelled Lhuyd, was born in 1527 at Foxhall, his family’s estate in Denbigh, and died in 1568. He was a cartographer, author, antiquary and Member of Parliament, and his motto was Hwy pery klod na golyd, fame lasts longer than wealth. A legend makes Brân a Cornish prince, the son of the Duke of Cornwall.

Another makes him King Brân Fendigaid, the Blessed, also called Bendigeidfran, a British king whose enlarged story is told in the Mabinogion. Camden also suggested that the name was simply bryn, a hill. Thomas Pennant, among others, thought Brân was the mountain stream of that name which rose in the Eglwyseg rocks and ran at the northern foot of the hill. As with other Welsh streams, Brân was applied because the water was black.

Edward Lhuyd, in Adversaria, confirmed that to his knowledge the name came from the brook of this name by Llangollen. Edward Lhuyd, also written Lhwyd and in other spellings, was born in 1660 at Loppington in Shropshire and died on 30 June 1709. He was the second Keeper of the Ashmolean Museum at Oxford and published the first catalogue of fossils, the Lithophylacii Britannici Ichnographia. In English the castle has been called Crow Castle since at least the eighteenth century, in Gough’s edition of Camden’s Britannia.

By the middle of the nineteenth century that was the form said to be used by most people in Llangollen, where an inn carried the same name.

The hill is made of thinly bedded, uncleaved, late Silurian deep-water marine silty mudstones of the Dinas Brân Formation, formerly called the Dinas Brân Group or the Dinas Brân Beds. The formation is up to 225 metres thick and is thought to range upwards into the late Ludfordian Stage, the upper part of the Ludlow epoch, whose span is given as 427.4 million years ago, plus or minus 0.5 million, to 423.0 million years ago, plus or minus 2.3 million. The mudstones and siltstones run from the exposures around the castle to the Pontcysyllte aqueduct, about 4.1 miles, or 6.6 kilometres, to the east. They occupy the core of the Llangollen syncline.

The upper junction is not exposed. Near the Wern road at the foot of the Eglwyseg escarpment there is a very small outcrop of dark olive-grey mudstone with abundant remains of the brachiopod Dayia navicula. An 1895 geological map shows that exposure faulted against fossiliferous Lower Carboniferous limestone of the Clwyd Limestone Group, laid down between 363 and 325 million years ago, though the actual junction is hidden by scree, regolith and gorse. Silurian fossils also turn up in scree and rubble on the steep northern slope and in the rock-cut ditches, which served both as defence and as a quarry for the building stone.

Straight-shelled orthocone nautiloids, various brachiopods and rare trilobite remains may be found. A visit in May 2017, guided by Fiona Gale, then county archaeologist, described the castle as mostly built from slate, with weathered walls that had needed modern intervention to stop further collapse. The ditches’ mudstone and that account of slate are left side by side.

During the British Iron Age, around 600 BCE, a large hillfort is said to have been built on the summit by the Ordovices, a Celtic people of this part of north Wales. An earthen rampart, probably with a wooden palisade, enclosed roundhouses, and an extra-deep ditch was cut on the gentler southern slope. The scheduled description does not give the year 600 or name the Ordovices. It calls the enclosure later prehistoric, probably Iron Age, univallate, with a probable entrance on the south-west and a number of possible hut sites in an undulating interior.

A historic-environment summary says the fort encloses 1.5 hectares and that the later castle was built over its north-west half, with possible hut sites noted in 1993. Other Ordovician strongholds named in the same regional account are Craig Rhiwarth in the Berwyn range and Dinas Emrys near Beddgelert to the west, and Caer Drewyn, Caer Euni and Moel y Gaer near the Horseshoe Pass to the east. The people of Old Oswestry are left as either Ordovices or Cornovii. Hillforts in the Clwydian range to the north, including Foel Fenlli and Moel Arthur, are given to the neighbouring Deceangli.

The Ordovices are also placed with the Gangani to the north-west, the Cornovii to the east, the Silures to the south and the Demetae to the south-west.

In 1879 the geologist Charles Lapworth named the Ordovician period after the Ordovices, because the rocks he had in mind lay in their former north-Welsh lands. Charles Lapworth was born on 20 September 1842 at Faringdon in Berkshire, now Oxfordshire, and died on 13 March 1920. He trained as a teacher, was headmaster of the school in Galashiels from 1864 to 1875, and in 1869 married Janet, daughter of the Galashiels schoolmaster Walter Sanderson. Mapping and index fossils at Dob’s Linn showed him that a sequence thought to be a thick pile of Silurian rocks was a thinner series repeated by faulting and folding.

He finished that work while a schoolmaster, then in 1875 became an assistant at Madras College in St Andrews, and in 1881 the first professor of geology in the post he then took. He wrote that north Wales, at all events the whole of the great Bala district where Adam Sedgwick first worked out the physical succession among the rocks of the intermediate system, and in all probability much of the Shelve and the Caradoc area, lay within the territory of the Ordovices, and that the central system of the Lower Palaeozoic should be called the Ordovician after that tribe. Adam Sedgwick, born on 22 March 1785 at Dent in Yorkshire and died on 27 January 1873, was a geologist and an Anglican priest. On Welsh strata he proposed the Cambrian period in 1835, in a joint publication in which Roderick Murchison proposed the Silurian.

In 1840, during the argument later called the Great Devonian Controversy, he and Murchison proposed the Devonian. The castle hill itself is described as late Silurian, not as the Ordovician system Lapworth named. The tribal name and the rock under the walls are separate facts.

The earliest structure that might have stood at Dinas Brân is believed to have belonged to Elisedd ap Gwylog in the eighth century. Elisedd, also called Elise, son of Gwylog ap Beli and a descendant of Brochwel Ysgithrog, is given as king of Powys from about 725 to about 755, dying about 755, and as a man who took the inheritance of Powys back from the English with his sword and with fire. No archaeological evidence for a structure of his time has been found on this hill. His great-grandson Cyngen ap Cadell, who died in 855, raised the Pillar of Eliseg, also Elise’s Pillar or Croes Elisedd, near the later Valle Crucis Abbey, in the lordship of Yale.

The spelling Eliseg on the stone is taken to be a carver’s mistake. The Latin inscription, once about thirty-one lines, was transcribed by Edward Lhuyd in 1696, when it was clearer. It names Concenn son of Cattell, Cattell son of Brochmail, Brochmail son of Eliseg, Eliseg son of Guoillauc, and says Concenn set the stone up for his great-grandfather, who joined the inheritance of Powys out of the power of the Angles. The pillar was thrown down in the English Civil War and a grave under it was opened.

The lower half disappeared. The upper half was re-erected in 1779. The inscription is now illegible. In 1773 the landowner Trevor Lloyd is said to have found a stone cist with a skeleton and objects, which he removed.

Excavation of the mound in 2010, 2011 and 2012, called Project Eliseg, found a kerbed platform cairn of a type dated around 2000 BCE, burnt human bone in a small cist, a larger Early Bronze Age cist without human remains, and another cist with 7 kilograms of cremated bone from adults, juveniles and infants, plus a flint knife and a bone pin. That dig was at the pillar, not in the castle wards. The castle account and the pillar account should not be folded into one set of finds.

Powys split when Madog ap Maredudd died in 1160. He was of the house of Mathrafal. The realm was divided under Welsh law among three sons, Gruffydd, Owain Brogyntyn and Owain Fychan, a nephew, Owain Cyfeiliog, and a half-brother, Iorwerth Goch. Gruffydd Maelor, who died in 1191, is also called Gruffudd Maelor and Gruffydd Maelor I, to distinguish him from his grandson.

He was a son of Madog ap Maredudd by Susanna, daughter of Gruffudd ap Cynan of the house of Aberffraw, and a brother of Owain Brogyntyn. He married Angharad, his cousin, a daughter of Owain Gwynedd. On the 1160 division he received the lordship of Maelor, also called Bromfield, and the lordship of Iâl, also called Yale. After Owain Fychan’s death in 1187 he added Nanheudwy, Cynllaith Owain and Mochnant Is Rhaeadr.

One account adds Glyndyfrdwy to that later gain. Those lands in effect reunited most of northern Powys, the portion that became known as Powys Fadog after it passed to his son. A castle narrative says that when the last prince, Gruffydd Maelor, died in 1191, the kingdom was divided into Powys Fadog in the north and Powys Wenwynwyn in the south. The life of his son says the territory Madog ruled became known as Powys Fadog in Madog’s honour, Fadog being a lenited form of Madog, and that the remainder of the old kingdom formed Powys Wenwynwyn.

The principality’s own summary dates Powys Fadog from 1160 to 1284, names Castell Dinas Brân as its capital and Valle Crucis as its religious centre, and lists lordships that included Maelor, Mochnant, Glyndyfrdwy, Yale, and Bromfield and Yale. The princes of Powys Wenwynwyn had their seat at Powis Castle. Whether a fortification already stood on this hill in the twelfth century is stated more cautiously in one place than in another. The castle narrative says Dinas Brân may have been a fortification of the kingdom, and that Madog might have ruled from it.

Whatever stood here then would have been timber, probably a palisade around a hall and other buildings, and early records attest that this early castle was destroyed by fire. Another account says documentary evidence suggests a timber castle in the late twelfth century which burnt down without trace, so that the stone castle was the third and last on the site. A funded survey summary says the first castle was probably timber, built in the twelfth century, destroyed by fire, and that the fire was most likely the work of Gruffudd ap Madog, prince of Powys, who re-established the castle in stone in about 1270.

Madog ap Gruffydd Maelor, also Madog ap Gruffudd, was prince of Powys Fadog from 1191 to 1236 and lord of Powys. He succeeded jointly with his brother Owen in 1191 and, on Owen’s death in 1197, became sole ruler of Powys north of the River Rhaeadr and the Afon Tanat. His lands still comprised Welsh and English Maelor, Iâl or Yale, Cynllaith, Nanheudwy and Mochnant Is Rhaeadr, and after his death in 1236 they were divided among five sons while still being called Powys Fadog. He was at first close to his cousin Llywelyn ab Iorwerth, Llywelyn the Great, then drew away.

From 1212, when Llywelyn had reformed a Welsh confederacy, Madog looked instead to King John of England. Another summary of the same reign says he began neutral between Gwynedd and England and by 1215 had settled on an alliance with Llywelyn ab Iorwerth. Both sequences are left as they are given. In 1201 he founded Valle Crucis Abbey at Llantysilio, in the commote of Iâl, the last Cistercian monastery founded in Wales.

Its Welsh names are Abaty Glyn Egwestl and Abaty Glyn y Groes, the valley of the cross, taken from the Pillar of Eliseg, raised four centuries earlier. A colony of twelve monks came from Strata Marcella, near Welshpool. A wooden start was replaced in roughly faced rubble. The house is thought to have held up to about sixty brethren, twenty choir monks and forty lay members.

Madog was buried in the then completed abbey when he died in 1236. Not long after, a serious fire is believed to have damaged the church and the south range, on archaeological evidence. The abbey was the spiritual centre of the region and Dinas Brân the political stronghold. The abbey was later said to have been caught up in the Welsh wars of the thirteenth century and damaged in the rising of Owain Glyndŵr.

It was dissolved in 1537, leased to Sir William Puckering or Pickering, and in time passed through the Wotton and Noel families before the estate was sequestered in 1651. Those later centuries belong to the abbey, not to the standing walls on the hill. They show what the same dynasty endowed beside the castle, and where its founder was buried.

The prince who rebuilt in stone is Gruffydd II ap Madog, Lord of Dinas Bran. He is also called Gruffydd Maelor, Gruffudd Maelor and Gruffydd Maelor II, and his death is given as 1269 and as 1269 or 1270. He was the eldest son of Madog ap Gruffydd Maelor and held the lands in partible succession with four brothers, Gruffudd Iâl, Maredudd, Hywel and Madog Fychan. His reign is counted as thirty-three years, from 1236 to 1269.

His father’s alliance with Gwynedd changed under pressure from Gwynedd and because of his own marriage to the daughter of an English landowner. He sought support from the English king, Henry III in one account. The help did not come. In 1258 he was forced to submit to Llywelyn ap Gruffudd, who had proclaimed himself prince of Wales that year after campaigns across Wales in 1257.

Around 1260 Gruffydd started Castell Dinas Brân in the Dee valley. The castle narrative says the visible stone castle was probably built in the 1260s. The monument’s scheduling places the stone castle in the same decade, built by the lords of Powys Fadog in the north-west corner of the hillfort. A monuments summary says about 1270, if not rather earlier, and names Gruffudd ap Madog, son of the founder of Valle Crucis.

The environment summary says the castle was built in 1270. The survey summary says about 1270. If he died in 1269, a date of about 1270 sits at the edge of his life or just after it. The dates are not collapsed into one year.

He married Emma, given as born in 1224 and dying about 1278, daughter of Henry de Audley and Bertrade Mainwaring, of the house of Stanley. His brother-in-law Sir James Audley was chief governor of Ireland from Dublin Castle and a companion of Richard of Cornwall, king of the Romans. A sister-in-law, Ela Longespée, was the daughter of the crusader William Longespée the younger, grandson of Henry II. At the time of the stone castle, Gruffydd was an ally of Llywelyn ap Gruffudd, prince of Wales, and Powys stood as a buffer between Gwynedd and England.

Dinas Brân was one of several castles built after the Treaty of Montgomery had secured Wales for Llywelyn, free from English interference. Dolforwyn, near Newtown, ordered by Llywelyn around the same time, has similarities and may have been the work of the same master mason. The treaty was drafted on 25 September 1267 and signed on 29 September 1267 at Rhyd Chwima in Montgomeryshire. Henry III, whose reign is given as 1216 to 1272, acknowledged Llywelyn as prince of Wales.

It was the only time an English ruler recognised a ruler of Gwynedd’s right over Wales. Llywelyn the Great had used the title prince of Aberffraw and lord of Snowdon in the 1230s. Llywelyn’s uncle Dafydd ap Llywelyn had claimed prince of Wales from 1240 to 1246. The 1267 text still bound the prince and his successors to fealty, homage and service.

Many of its terms had been anticipated at Pipton in 1265 with Simon de Montfort. It ceded Builth, Brecon and Gwerthrynion, and the four cantrefs of Perfeddwlad. Cefnllys, taken from Roger Mortimer in 1262, went back to Roger, to be held tentatively from Llywelyn if Llywelyn could affirm his right. Llywelyn received Whittington in Shropshire and an assurance that Robert of Mold would build no castle at Hawarden for thirty years.

His brother Dafydd, who had gone to England in the early 1260s, was to be reinstated. Maredudd ap Rhys was allowed to remain a direct vassal of the English king. Signatories included Henry III, Prince Edward, the papal legate Ottobuono de’ Fieschi, Llywelyn, Goronwy ab Ednyfed Fychan and Tudur ab Ednyfed Fychan. After that settlement, one account says Gruffydd was confined to his castle for the rest of his life.

Another says his influence waned and he confined himself to building Castell Dinas Brân. The start of the work around 1260 comes before the treaty. The confinement is placed after it.

He died in 1269, or in 1269 or 1270, leaving Madog II, who succeeded and was killed in battle with the English in 1277; Llywelyn; Owain, whose daughter Gweirca ferch Owain has the oldest dated grave slab in Wales, though that slab is not placed at this castle; and Gruffudd Fychan I, who died in 1289. Daughters were Angharad, who died in 1308 and married, after 1261, William le Boteler of Wem, who died in 1283, and Margery, born in 1261, who married Sir John de Arderne. Powys Fadog was divided by Welsh custom. Madog received Maelor.

Gruffudd had Iâl, Yale, and Edeirnion, including Glyndyfrdwy. Swydd y Waun, the commotes of Cynllaith and Nanheudwy, was divided between Llywelyn and Owain. The castle narrative says the castle was inherited by the four sons, that Madog the eldest was senior, and that each son may have had apartments at Dinas Brân. A list of rulers gives Madog II the years 1269 to 1277, then Gruffudd ap Madog from 1277 to 1282, and Llywelyn Fychan ap Gruffudd in 1282.

When Gruffydd died, the small share given to the younger brothers caused rebellion, and England became engaged. By 1276 northern Powys was in disorder, brother against brother.

The peace with England did not last. Edward succeeded as king in 1272. He viewed Llywelyn as a threat, disliked Llywelyn’s plan to marry Eleanor de Montfort, daughter of Simon de Montfort and Edward’s cousin, and summoned him more than once. Llywelyn refused on the ground that he was not safe at Edward’s court.

Eleanor and Llywelyn were married by proxy in 1275. When she sailed from France, Edward hired pirates to seize the ship, and she was imprisoned at Windsor. In 1276 Edward declared Llywelyn a rebel and gathered an army. One account says he declared war in 1276.

The conquest of Wales by Edward I, also called the Conquest of Wales, is placed in two campaigns, 1277 and 1282 to 1283. The castle narrative says that in 1277 Edward launched the conquest of north Wales from Chester. In early 1277 an army led by William de Beauchamp, ninth Earl of Warwick, with support from Llywelyn’s brother Dafydd ap Gruffydd, marched from Chester into northern Powys. By the summer of 1277 Edward’s forces had reached the heart of Gwynedd.

The harvest on Anglesey was confiscated, and the loss of food forced Llywelyn toward surrender.

What happened at Dinas Brân in that year is told in more than one way. Two of Gruffudd’s sons, Llywelyn and Madog, quickly made peace with Edward, but their surrender documents state the need to recapture Dinas Brân, so the fortification was not under Madog’s control. Henry de Lacy, Earl of Lincoln, was sent with forces from Oswestry to capture it. As soon as he arrived he was told that the defenders, probably the younger brothers Owain and Gruffudd, still allies of Llywelyn, had abandoned the castle and set it alight.

The reason is not clear. They may have had no confidence they could hold it, and may not have wanted it to fall intact to the English or to their elder brother. Despite the fire, the castle was not badly damaged. The earl recommended that Edward repair and garrison it.

The English occupied it at least until the Treaty of Aberconwy, and some repair was ordered. A second account says Madog II was compelled to submit, and that under the surrender the realm would be divided between him and his younger brother Llywelyn. Castle Dinas Brân, called in that account the strongest native castle in Wales, was to be had by neither, and was to be dismantled. Men loyal to Madog, or Madog himself, remained for some time, because Henry de Lacy was commanded to take the castle on 10 May 1277.

Before the English could finish an encirclement, the small garrison abandoned the cause and burnt the castle. Madog II fled to the protection of Gwynedd and was killed later that same year while campaigning alongside Llywelyn. The castle would be slighted. A third account says the garrison burnt the castle to prevent its use by the invaders, and also that it was attacked and burnt by the Welsh in 1277 while they held it and was subsequently attacked by the English in the same year.

A fourth says the Welsh defenders set fire to it so that it would be of no use, that the damage was not excessive, and that Edward briefly garrisoned the place. In May 2017, patches of scorched stone were noticed, and the ruin was said to owe its present condition partly to slighting, meaning deliberate partial demolition, with ramps later added against erosion. The fire, the slighting, the repairs and the claim that the damage was small are all kept.

Henry de Lacy, Earl of Lincoln, was born about 1251, and a biographical note also gives 1249. He died in February 1311 at Lincoln’s Inn in London and was buried in St Paul’s, a monument later lost in the Great Fire of 1666. He was baron of Pontefract, lord of Bowland, baron of Halton and hereditary constable of Chester, son of Edmund de Lacy and Alice of Saluzzo. His father died in 1258, when Henry was about seven.

In 1266 he was heir to his grandmother Margaret de Quincy. He came of age in 1272, was knighted, and became earl of Lincoln. He was a chief councillor of Edward I and served in Wales, France and Scotland. His own life says he took part in the Conquest of Wales in 1282, was granted the lordship of Denbigh and built Denbigh Castle.

It does not, in that summary, narrate the May 1277 approach to Dinas Brân. The castle accounts do. In 1296 he went to Gascony, commanded in Aquitaine after Edmund Crouchback’s death, and was routed at Bonnegarde in 1297. He was at Falkirk in 1298 and at Caerlaverock in 1300.

He helped arrange the betrothal of the future Edward II to Isabella of France. He died an Ordainer, having turned from Edward II and Piers Gaveston toward Thomas of Lancaster. His daughter Alice de Lacy, born on 25 December 1281, married Thomas. The Denbigh grant belongs to the later war, not to a rebuilding of Dinas Brân.

The Treaty of Aberconwy was signed on 10 November 1277 between Edward I and Llywelyn ap Gruffudd. It re-established peace and is described as completing the first stage of the conquest, and as guaranteeing that Welsh self-rule would end at Llywelyn’s death. Llywelyn kept only the western part of Gwynedd, though he kept the title prince of Wales. Eastern Gwynedd was split between Edward and Dafydd.

Deheubarth, Powys and mid-Wales became a mixture of royal land and protectorates. Edward then allowed the marriage. In 1278 Llywelyn and Eleanor de Montfort were married in Worcester Cathedral, with Edward present. Dafydd was discontented with his reward and rebelled in 1282.

The rebellion took on the character of a national struggle. Llywelyn and other Welsh leaders joined. Edward came to treat it as a war of conquest.

When that war restarted, the history of this castle is not recorded in one of the accounts. It may have been retaken by the Welsh in the early months, like many other castles, but it fell to the English in the end. After Madog died, the three surviving brothers all fought for Llywelyn. Another account says that in 1282 all the rulers of northern Powys turned against England.

Llywelyn ap Gruffudd died on 11 December 1282 in an ambush at Cilmeri. He is also said to have been killed at the Battle of Orewin Bridge in December 1282. A further account says that in December 1282 the army of Montgomery marched from Montgomery Castle to Builth Wells to surprise and kill him. Llywelyn Fychan and the two sons of Madog II died in the same final conflict.

Dafydd was captured in June 1283. One castle narrative dates the end of the war to October 1282 and to the death of Llywelyn, which the other dates place in December, with the conquest running into 1283. Under the Statute of Rhuddlan in 1284 the remaining princely titles and territories were abolished. Gruffudd Fychan, brother of Madog II and last heir to the throne of northern Powys, was pardoned and reduced to a minor noble, an uchelwr.

His direct descendant Owain Glyndŵr led a rebellion in 1400. The cantrefi of Maelor, Nanheudwy, Iâl, Cynllaith and Mochnant Is Rhaeadr went into Denbighshire, and Maelor Saesneg formed the Wrexham exclave of Flintshire, an arrangement said to have lasted until the local-government reorganisation of 1974.

After the war, most of Powys Fadog, including the castle, was granted to John de Warenne, Earl of Surrey. Rather than rebuild Dinas Brân, he built a new castle by the River Dee at Holt, on the Flintshire and Cheshire border. The distance is also given as 16 miles. The hilltop castle was left to fall into ruin.

One summary says it was abandoned after 1282, when Holt became the centre of the new lordship. The site summary says abandoned in the fourteenth century. Both are kept. John de Warenne, sixth Earl of Surrey, was born in 1231 and died on 27 September 1304 at Kennington in Surrey.

Holt Castle is given as a residence. He served in Edward’s Welsh campaigns of 1277, 1282 and 1283. In 1282 Edward awarded him the lordship of Bromfield and Yale, taken from Gruffudd Fychan I and his son Madog Crypl. He started Holt, and his son William continued it.

Those lordships are the same Maelor and Yale that Gruffydd Maelor had received in 1160, later merged. Warenne was defeated by William Wallace at the Battle of Stirling Bridge and fled to York. He had switched sides twice in the Second Barons’ War, been at Lewes in 1264, and in 1270 had attacked Alan la Zouche at Westminster, of which wounds Zouche died on 10 August 1270. His choice of Holt, not the repaired hilltop, is the change that ended Dinas Brân as a princely seat.

The walls that remain are described in more than one plan. The castle narrative gives rectangular stone walls, longer east to west. The north side uses the natural slope, falling sharply for several hundred feet. On the gentler south and east an additional rock-cut ditch, 20 feet or 6.1 metres deep, and a counterscarp bank strengthen the wall.

A monuments account gives a rectangular court about 82 metres east to west by 35 metres, with a massive rock-cut ditch and counterscarp except on the north, where the slopes are headlong. The keep is placed at the south-eastern corner, where the ditch is deepest, looking toward a relatively easy approach from the Dee. It was two storeys, the strongest part, with its own narrow defended approach. On the scheduled monument, the south and west sides of the keep are well preserved.

Beside the keep, at the north-eastern corner, the castle narrative puts a gatehouse, once reached by a wooden bridge across the ditch. Almost nothing of the bridge or its supports remains, so the exact form is unclear. The keep overlooked the bridge, so archers there could cover the entrance. The gatehouse had two towers and a decorated covered passage into the courtyard.

Excavation accounts instead speak of a southern gatehouse, of its eastern end, of an external entrance corridor, and of a southern turret guardroom. The monuments account puts a great rectangular tower on the short east side, isolated from the court by a rock-cut ditch, and an ornate twin-towered gate tower. Those descriptions of a north-eastern gatehouse, a southern gatehouse and an east-side gate tower are left side by side.

On the south side stood the great hall, for dining and for receiving visitors. Enlarged windows still look south, and an arched gateway at the west end of the room led to kitchens in the basement of the adjacent apsidal, or D-shaped, tower. The scheduled description calls the arched windows those of the great hall or chapel, to the east of the apsidal tower. The monuments account puts a hall between the D-shaped tower and the tower ditch.

The tower, called the Welsh tower, is a form typical of Welsh castles of the period. It projected into the ditch and gave archers a view along the southern wall. It had perhaps three storeys, with living quarters above. In the south-west corner was a postern, a way out in a siege so the garrison could sally.

Fragments of the arch remain, and the slot for the door’s drawbar. A further stone-built range stood on the west side of the court. Traces of buildings in the west are read as workshops, accommodation and stabling. Stables, workshops, storage and maybe a chapel inside the enclosure are also described as timber, so that nothing of them remains above ground.

A slighter outer circuit of earthworks is of unknown date and purpose. In the nineteenth century Walter Hawken Tregellas recorded a local tradition that at Tower Farm, about a mile from the castle, a tower had formerly stood as an outwork. That tradition is not a surveyed wall.

The historical record is sparse beside the legends. The castle’s first literary appearance is in a thirteenth-century romance, Fouke le Fitz Waryn. There the castle, named Chastiel Bran, is a ruin during the early years of the Norman Conquest, long before the stone work of the 1260s. Payn Peveril, an arrogant Norman knight, hears that no one has dared to stay the night for fear of evil spirits.

He and fifteen followers stay. A storm rises and a mace-wielding giant, Gogmagog, appears. Payn holds him off with shield and cross and stabs him. Dying, the giant says that King Brân built the castle to defeat him, failed, and fled, and that the giant had terrorised the land ever since.

He also tells of a buried treasury of idols, swans, peacocks, horses and a huge golden ox, and dies without saying where it lies. That is the romance, not a find from the hill.

A fourteenth-century love story of Myfanwy Fychan of Castell Dinas Brân and the poet Hywel ab Einion later fed several works. By the time the poem was written, one account says, the castle had lain in ruin for quite some time, and Myfanwy had married Goronwy ap Tudur Hen, lord of Penmynydd. John Ceiriog Hughes, who lived from 1832 to 1887, wrote the poem Myfanwy Fychan in 1858. Felicia Hemans wrote Howel’s Song, set to music by John Parry in his Welsh Melodies of 1822.

Felicia Dorothea Hemans, born Felicia Dorothea Browne on 25 September 1793 at 118 Duke Street, Liverpool, and died on 16 May 1835, was an English poet who identified as Welsh by adoption. Her family came to Denbighshire, where she spent her youth. The song Myfanwy was composed by Joseph Parry, first published in 1875, to lyrics by Richard Davies, called Mynyddog, who lived from 1833 to 1877. Accounts differ on whether Parry set an existing poem or Davies wrote the words for Parry’s melody.

The lyrics were probably inspired by the Dinas Brân story. Other accounts say the song was written with Parry’s childhood sweetheart, Myfanwy Llywellyn, in mind, and the words may also draw on Thomas Walter Price’s English poem Arabella. The song was first given on 28 May 1875 at the first concert of the Aberystwyth and University Musical Society, with Parry conducting. He sold all rights to the publisher for £12.

Painters used the ruin. Richard Wilson’s Dinas Bran from Llangollen is dated 1771. J. M.

W. Turner also painted the castle and its surroundings. In the 1800s the paths were used by donkeys: a visitor hired a donkey and a donkey boy for the excursion. Between 1869 and the Second World War a camera obscura stood on the summit, and a tearoom was put up.

The turret was used as storage for that Victorian tearoom. Debris of the eighteenth and nineteenth centuries was deliberately placed inside the turret so that a vaulted room could be made into an accessible room. A portable camera obscura was later made under the Our Picturesque Landscape project, and schoolchildren made a performance about the old one. The national landscape that manages the hill with the county council gives opening as all year round, and facilities as none.

The first archaeological investigation named on the castle narrative is a 2017 geophysical survey. The Castle Studies Trust awarded £1,825, and the Welsh heritage service also funded the work, carried out for the county council. Resistivity was planned over 1.5 hectares and magnetometry over 1 hectare, to locate buildings inside the castle and to see how the medieval work had affected the prehistoric earthworks. A note of 9 July 2017 said the report was nearly ready, that the results were tantalising and told more about how the castle was used, and that some questions would need excavation.

A later private summary of that survey says nothing conclusive was discovered, while magnetic readings suggested a fire had scoured the ramparts, which that writer tied to the sons of Gruffudd burning the place rather than surrendering it. That reading is one person’s, set beside the several 1277 accounts already given. In 2020 the Clwyd-Powys Archaeological Trust carried out a condition survey of the castle and the hillfort, to inform conservation under a Dee valley project funded by the national lottery heritage fund. Earthworks beyond the hillfort were noted and were not included in that survey.

The first recorded excavation was in August 2021, four hand-dug trial trenches, inside the eastern end of the southern gatehouse and outside it at the western end and the northern side. The aim was to inform consolidation of the exposed gatehouse walls by the county council’s Our Picturesque Landscape team, by testing the buried lower parts. Inside the gatehouse at least a metre of debris had built up in the eighteenth to twentieth centuries, and earlier deposits were left unexcavated. Along the external entrance corridor, stone fragments and debris were taken down about 600 to 800 millimetres, to a basal deposit read as a medieval trampled surface.

One sherd of medieval pottery was recovered. A ledge or kerb projected from the gatehouse wall, with a portcullis slot near the east end and a carved masonry pillar base at the west end. The written account was then waiting on a second phase of work in 2022. No separate 2022 excavation has been set out in the project notice.

The same works are also described as finding surviving floor levels and buried architectural features, and as letting masons conserve the base of the walls and read the sequence of defences along the entrance passage. More was thought to lie under the eighteenth- and nineteenth-century debris put into the turret for the tearoom. The gatehouse was left structurally sound, and able to be opened at certain times.

In the autumn before an account dated 11 January 2023, rangers, volunteers and contractors, guided by the trust’s archaeologists, excavated close to 35 tonnes from the gatehouse. The material was bagged and lifted by helicopter onto an eroded path, which had widened and was threatening nationally scarce plants, upright chickweed and shepherd’s cress. Rhun Jones, senior ranger in the Dee valley, said the cleared material would repair the trail and protect the grassland. That restoration sat inside the same lottery-funded landscape project, with further support from the Welsh heritage service.

A September 2025 account of a new dig says it built on what had been uncovered in 2023, so the autumn clearance and a 2023 season are not treated as the same fortnight.

In September 2025 a three-week excavation was run in partnership by Heneb, the Trust for Welsh Archaeology, the Welsh heritage service, and the Clwydian Range and Dee Valley team, with archaeologists and heritage workers beside students and local volunteers. Sue Evans and Sam Thomson of Llangollen Museum described the first days to a local news note of 17 September 2025. The focus was the medieval gatehouse. For the first time a trench was also put on the Iron Age ramparts.

Day one was very windy, and turf was stripped on both trenches. By day two, later Victorian pottery was coming up in the gatehouse trench, probably linked to the old café. On the rampart slope, bedrock appeared quickly. Colleagues in the ditch found medieval pottery, more than at the castle itself.

The hope was dating evidence, and a display in the museum. Previous excavation, described as the latest before this season, had confirmed the portcullis, the main gate door jamb, the draw-bar holes and the base of an elaborately carved wall pillar. High-status masonry still buried might show whether the princes of Powys used Welsh or Marcher masons. The doorway into the southern turret guardroom was located.

Medieval floor levels survived in the gate passage. The original floor of the southern turret was not reached, because the backfill was deep, and it was thought almost certainly still intact. Artefacts were few: two decorated shards of thirteenth-century pottery at the eastern entrance. Extensive medieval deposits still lie under rubble and landscaping debris.

In June 2026 the trust’s field team returned with local volunteers and wardens from the national landscape. After the September trench in the gatehouse, the work moved to the rampart and counterscarp on the north-eastern side. Two trenches were opened, one on the crest of the rampart and one in the counterscarp ditch. Stony layers were taken down to medieval soils.

They produced a substantial collection of thirteenth-century material: high-status pottery, slag, animal bone and large fragments of limestone mortar. The team believed they had reached the base of the Iron Age counterscarp ditch, cut directly into the bedrock. In the smaller trench on the counterscarp bank, a large posthole packed with stones suggested that the rampart may originally have been timber-built, which fits the earlier idea of a wooden palisade and does not yet prove its date. Soil samples were still to be examined before anyone could say the Iron Age foundations had been reached.

The archaeologist Karl Marcow described the features in the trench on top of the hillfort rampart. Soil dates from that season had not yet been published.

What the place was for changes with the period, and the sources do not make those uses into one continuous garrison. The hillfort used the summit, the view down the Dee, and a ditch on the easier southern side. Possible hut sites are the only hint of houses, and no list of Iron Age finds from the castle trenches has been published. The timber castle, if it stood in the late twelfth century, would have been a lordly enclosure of palisade, hall and outbuildings for the northern princes, and it ended in fire.

The stone castle was a princely seat of Powys Fadog in the decade of the Treaty of Montgomery, when Llywelyn ap Gruffudd’s title and Gruffydd’s buffer lordship were both at stake. It had a keep, a twin-towered gate, a hall, a D-shaped tower and a postern. It was a residence as well as a fortification, and one monuments description calls the ruins stark but the lost castle magnificent and sumptuous, as a prince’s principal house. In 1277 the question was whether it would be held for Llywelyn, handed to a brother who had made peace, repaired for an English garrison, or denied to everyone by fire.

The accounts give all of those outcomes in some form. After 1282 the lordship’s centre moved to Holt. The hilltop was not rebuilt. Its later life was as a ruin in stories and pictures, a donkey path, a camera obscura and a tearoom, and then as a scheduled monument whose gatehouse and rampart were dug only from 2021, with the dates of the timber fort and of the stone castle still sitting in more than one century.