CHAPTER IX

MOTTE-CASTLES IN SOUTH WALES

It is not possible to fix certain dates for all the Norman conquests of the several provinces of South Wales. These conquests proceeded from various points, under different leaders. We might have expected that the earliest advances would have been on the Herefordshire border, the earldom of Hereford having been given by William I. to William FitzOsbern, one of his most trusted and energetic servants. Ordericus tells us that FitzOsbern and Walter de Lacy first invaded the district of Brecknock, and defeated three kings of the Welsh.[896] This looks as though the conquest of Brecknock was then begun. But it was not completed till the reign of Rufus; in 1093 Bernard of Neufmarché defeated and slew Rhys ap Tudor, King of South Wales, in a battle which the Welsh chronicler speaks of as the fall of the kingdom of the Britons.[897] William FitzOsbern died in 1071, and he had scarcely time to accomplish more than the building of the border castles of Wigmore, Clifford, Ewias, and Monmouth, and the incastellation of Gwent, that is the country between the Wye and the Usk, which had already been conquered by Harold.

It seems probable that Pembrokeshire was one of the earliest Norman conquests in South Wales, as in 1073 and 1074 the _Brut_ tells of two expeditions of "the French" into Dyfed, a region which included not only what we now call Pembrokeshire, but also Strath Towy, which comprised an extensive district on both sides of the valley of the Towy.[898] The _Annales Cambriæ_ name Hugh de Montgomeri, Earl Roger's eldest son, in connection with the second of these expeditions, seven years before the expedition of King William into Wales in 1081.[899] The House of Montgomeri certainly took the most conspicuous part in the conquest of Dyfed and Cardigan, which was completed, according to the _Brut_, in 1093.[900] Arnulf of Montgomeri, fifth son of Earl Roger, was the leader of this conquest. But his father must at the same time have been operating in Cardigan, as the building of the castle of Cilgerran, which is on the very borders of Pembroke and Cardigan, is attributed to him.

How far Earl Roger made himself master of Ceredigion it is impossible to say. Later writers say that he built the castle of Cardigan, but we have not been able to find any early authority for this statement, which in itself is not improbable. Powell's _History_ makes him do homage to William Rufus for the lordship of Cardigan, but here again the authority is doubtful.[901] The fact that a castle in or near Aberystwyth was not built until 1109 may indicate that the conquest of Northern Cardigan was not completed till it became the portion of the De Clares. This took place in 1109, when Henry I. deposed Cadwgan, a Welsh prince whom he had made Lord of Cardigan, and gave the lordship to Gilbert de Clare, who immediately proceeded to build the above-mentioned castle, and to restore Earl Roger's castle at Cilgerran (Dingeraint).[902] From this time the castle and district of Cardigan continued to be an appanage of the House of Clare (of course with frequent interruptions from Welsh invasions), and of the family of William Marshall, to whom the Clare lands came by marriage. The authority of these earls was suspended during the reign of Henry II., when he made Rhys ap Griffith, who had possessed himself of Ceredigion by conquest, Justiciar of South Wales, but in the reigns of John and Henry III., the _Close Rolls_ show that Cardigan Castle and county were generally in the hands of the Marshalls.

The conquest of Pembrokeshire must have been closely followed by that of what is now Carmarthenshire, which was then reckoned as part of Dyfed.[903] We first hear of the castle of Rhyd y Gors in 1094,[904] but it evidently existed earlier. This castle we believe to have been the important castle of Carmarthen (see _post_). It was founded by William, son of Baldwin, sheriff of Devon, and cousin of the Gilbert de Clare who at a later period was made Lord of Cardigan by Henry I. We thus see at what an early date this important family made its appearance in Welsh history.

The conquest of Brecknock (Brecheiniog) we have already briefly referred to. It must have begun as early as 1088, for in that year Bernard de Neufmarché gave to St Peter's Abbey at Gloucester the church and manor of Glasbury. The inheritance of Bernard passed by marriage to the De Braoses, and from them to the Mortimers. It is convenient to mention in this connection the Norman conquest of Radnor, of which the De Braoses and Mortimers were the heroes. A charter of Philip de Braose, not later than 1096, is dated at "Raddenoam."[905] Even during the anarchy of Stephen's reign, the Mortimers were able to maintain their hold on this district, for the _Brut_ relates that in 1145, Hugh, son of Ralph Mortimer, conquered Malienydd and Elvael the second time.[906] These two districts properly belong to Powys, though geographically in South Wales.

We leave to the last the conquest of Glamorgan, which may possibly have been one of the earliest, but whose date is still a matter of dispute, owing to the legendary nature of the Aberpergwm version of the _Brut_, the only one which even alludes to this conquest. We have, however, an initial date given us in the year 1082, when the _Brut y Tywysogion_ tell us of the building of Cardiff Castle.[907] The conquest of "Morgannwg," that is the country between the Usk and the Neath, was the most permanent of any of those accomplished by the Normans in Wales, but its details are the most obscure of any. The earlier version of the _Brut_ takes no notice of the conquest of Glamorgan; the later version which goes by the name of the _Gwentian Chronicle_[908] tells us that the Norman Robert Fitz Hamon, being called in to the help of one Welsh prince against another, conquered Glamorgan for himself, and divided it amongst his followers, who built castles in all parts of the country. The date given is 1088. It seems to be agreed by historians that while the facts of Robert Fitz Hamon's existence and of his conquest of Glamorgan are certain, the details and the list of followers given in this chronicle are quite untrustworthy.[909]

The district called Gower did not then form part of Glamorgan, as it does now, though it is still ecclesiastically separate. If we are to believe the Aberpergwm _Brut_, it must have been conquered in 1094, when William de Londres, one of the "knights" of Robert Fitz Hamon, built a strong castle in Cydweli (Kidwelly).[910]

We will now briefly notice such of the castles of these various districts as are mentioned in the sources to which we have already referred in our last chapter, taking them in the order of the modern counties in which they are found.

CASTLES OF PEMBROKESHIRE.

PEMBROKE.--Giraldus says that Arnulf de Montgomeri first built this castle of sods and wattles, a scanty and slender construction, in the reign of Henry I.[911] This date, however, must certainly be wrong, for the castle sustained a siege from the Welsh in 1094, and in 1098 Arnulf gave the chapel of St Nicholas in his castle of Pembroke to the abbey of St Martin at Sées.[912] There is no motte at Pembroke Castle; the magnificent keep (clearly of the 13th century or later) stands in a small ward at the edge of a cliff,[913] separated by a former ditch from the immense encircling bailey whose walls and towers are clearly of Edwardian date. The words of Giraldus "a castle of wattles and turf" might lead us to think that the first castle was a motte of the usual type, but the use which he makes of the same expression in his work on Ireland leads one to think that he means a less defensible fort, a mere bank and fence.[914] There is some reason, moreover, to doubt whether the present castle of Pembroke stands on the same site as Arnulf's, as after the banishment of the latter, Gerald, the royal Seneschal of Pembroke "built the castle anew in the place called Little Cengarth."[915]

But however this may be, the castle of Pembroke was certainly strong enough in 1094 to resist a great insurrection of the Welsh, when all the castles of south-west Wales were destroyed, except Pembroke and Rhyd y Gors. And it continued to be one of the chief strongholds of English power in South Wales until Edward I. completed the conquest of the country. Its splendid situation on a high cliff at the mouth of an excellent harbour, to which supplies could be brought by sea, was one of the secrets of its strength. A passage cut in the rock led from the castle to a cave below opening on to the water.

*NEWPORT, or Trefdaeth, was the head of the Barony of Keymes, an independent lordship founded at the time of the first Norman advance, by Martin of Tours.[916] There is no mention of it before 1215. The present ruined castle of Newport is not earlier than the 13th century, but about 1-1/2 miles higher up the river, at Llanhyfer, is a fine motte and bailey, which probably mark the site of the first castle of Martin of Tours.[917]

WISTON, _alias_ Gwys or Wiz.--First mentioned in 1148, when it was taken by the Welsh.[918] At a later period we find it one of the castles of the Earl of Pembroke. There is a motte still remaining, with a shell wall on top, 6 feet thick, having a plain round arched entrance. This masonry is probably the work of William Marshall, Earl of Pembroke, as he restored the castle in 1220 after it had been razed to the ground by Llywelyn ap Jorwerth.[919] The bailey is large and bean-shaped.

LAWHADEN, or Llanyhadein, or Lauwadein.--First mention in 1192.[920] It afterwards became a palace of the bishops of St David's. There is no motte, though the circular outline of the platform on which the fine ruins of the castle stand, very much suggests a lowered motte.

HAVERFORDWEST.--First mentioned in the _Pipe Roll_ of 1214-1215, when it was in the custody of the Earl of Pembroke. Although this castle is now a gaol, and the whole site masked with gaol buildings, the motte can still be seen distinctly from one side, though the keep which stands upon it is blocked by buildings. The ditch which went round the motte can also be traced. [H. W.]

NARBERTH.--This castle is first mentioned in 1115, when it was burnt by the Welsh. Said to have been the castle of Stephen Perrot.[921] The present ruins are entirely of the 13th century, and there is no motte; but Lewis states that the first castle was in another site, between the present town and Templeton; about which we have no information.

TENBY.--First mention in 1152. An important coast station. The small and curious round keep is placed on the highest point of a small island; it is a miniature copy of the keep of Pembroke, and was probably built by one of the earls Marshall, not earlier than the 13th century. There is no motte, nor was one needed in such a situation.

CASTLES OF CARDIGAN.

CARDIGAN Castle, or Aberteifi, has been so much transformed by the incorporation of the keep into a modern house that nothing decisive can be said about its original plan, but there is nothing to foreclose the idea of a previous motte, and Speed's plan of 1611 seems to show that the keep and the small ward attached to it were on a higher elevation than the bailey. That the first castle was a wooden one is rendered almost certain by the fact that Rhys ap Griffith, after having demolished the previous castle, rebuilt it _with stone and mortar_, in the reign of Henry II.[922] The Welsh chronicler speaks of this castle as the key of all Wales, an exaggeration certainly, but it was undoubtedly the most important stronghold of South Ceredigion. [H. W.]

CILGERRAN, or Dingeraint (Fig. 41).--This castle was certainly built by Earl Roger;[923] a castle of great importance, in a magnificent situation. Like nearly all the castles in our Welsh list, it was repeatedly taken by the Welsh and retaken from them. The present masonry is of the 13th century, but the original motte-and-bailey plan is quite discernible. [H. W.] It was a connecting link between the castles of Pembrokeshire and those of Cardigan, and stands near a road leading directly from Tenby and Narberth to Cardigan.

ABERYSTWYTH, also Lampadarn Vaur, also Aberrheiddiol.[924] In 1109 Henry I. deposed Cadwgan, a Welsh prince who had purchased from the king the government of Cardigan, and gave that country to Gilbert, son of Richard, Earl of Clare, who took possession, and built a castle "opposite to Llanbadarn, near the mouth of the river Ystwyth."[925] This was undoubtedly the precursor of the modern castle of Aberystwyth, but it is doubtful whether it was on the same site; the present ruins are not opposite Llanbadarn. The castle was as important for the defence of N. Cardigan as Cardigan Castle for the south. It was taken at least seven times by the Welsh, and burnt at least five times. The present ruins are not earlier than the time of Edward I., and there is no motte or keep. [H. W.]

*BLAENPORTH, or Castell Gwythan (Fig. 41).--Also built by Gilbert de Clare, and evidently placed to defend the main road from Cardigan to Aberystwyth. The motte and bailey are still remarkably perfect, as shown by the 25-inch Ordnance Map.

YSTRAD PEITHYLL.--Another of Gilbert de Clare's castles, as it was inhabited by his steward. It was burnt by the Welsh in 1115,[926] and is never mentioned again, but its motte and ditch still survive, with some signs of a bailey, close to the little stream of the Peithyll, near Aberystwyth. [H. W.]

CHASTELL GWALTER, or Llanfihangel, in Pengwern (Fig. 41).--Castle of Walter de Bec, probably one of the barons of Gilbert de Clare. First mentioned in 1137, when it was burned by the Welsh.[927] There is a small but well-made motte and part of an adjoining bailey standing in a most commanding position on a high plateau. The ditch of the motte is excavated in the rock. [D. H. M.]

[Illustration: FIG. 41.--MOTTE-CASTLES OF SOUTH WALES. Cilgerran. Blaenporth. Chastell Gwalter.]

*DINERTH.--Also burnt in 1137; restored by Roger, Earl of Clare, in 1159, after which it underwent many vicissitudes.[928] Probably originally a castle of the Clares. "In the grounds of Mynachty, in the parish of Llanbadarn Tref Eglwys, is a small hill called Hero Castell, probably the site of the keep of Dinerth Castle."[929] The O.M. shows a small motte and bailey placed between two streams.

*CAERWEDROS, or Castell Llwyndafydd, also burned by the Welsh in 1137,[930] after which it is not mentioned again. "A very large moated tumulus, with foundations of walls on the top."[931] Probably a Clare castle.

*HUMPHREY'S CASTLE, now Castle Howel, from one of its Welsh conquerors. The original name shows that it was built by a Norman, and it was restored by Roger, Earl of Clare, in 1159.[932] A moated tumulus near the river Clettwr marks the site of Humphrey's Castle.[933]

YSTRAD MEURUG, or Meyric, at the head of the valley of the Teifi, and commanding the pass leading over into Radnorshire.--Built by Gilbert de Clare when he reconquered Cardigan, and one of his most important castles.[934] Its importance is shown by the fact that it had a small stone keep, the date of which cannot now be determined, as only the foundations remain, buried under sods. There is no motte, and the bailey can only be guessed at by a portion of the ditch which still remains on the N. side, and by two platforms which appear to be artificially levelled. The castle is about three miles from the Sarn Helen or Roman road through Cardigan.

*PONT Y STUFFAN, or Stephen's Bridge, near Lampeter.--Burnt by the Welsh in 1138, and not again mentioned.[935] In the outskirts of the town of Lampeter is--or was--a lofty moated tumulus (not shown on O.M.), and traces of a quadrangular court.[936] As it is also called Castell Ystuffan, it was probably built by Stephen, the Norman constable of Cardigan. There appears to be another castle mound at Lampeter itself, near the church. Lampeter was an important post on the Roman road up the valley of the Teifi.

*NANT YR ARIAN.--This castle is only mentioned once, in the partition of Cardigan and Pembroke which took place in 1216, during the most disastrous part of John's reign.[937] There are two "castellau" marked at Nant yr Arian in the N. of Cardiganshire in the O.M.; neither of them look like mottes. This castle, as well as that of Ystrad Peithyll, seems to have been placed to defend the road from Aberystwyth to Llanidloes, which would be the chief highway between Shropshire and Ceredigion.

CASTLES OF CARMARTHENSHIRE.

RHYD Y GORS, or Rhyd Cors.--We have no hesitation in adopting the opinion of the late Mr Floyd, that this is another name for the castle of Carmarthen.[938] As it and Pembroke were the only castles which held out during the great Welsh revolt of 1096,[939] it is evident that they were the two strongest and best defended places, therefore the most important. Carmarthen also was a Roman city, and its walls were still standing in Giraldus' time;[940] it was therefore the place where one would expect to find a Norman castle. Now Carmarthen, along with Cardiff and Pembroke, continued up till the final conquest of all Wales to be the most important seat of English power in South Wales. Moreover, Rhyd y Gors was a royal castle; we are expressly told that it was built by William Fitz Baldwin, by the command of the king of England.[941] Carmarthen also was a royal castle, and the only one in South Wales at that date which belonged directly to the king. It was temporarily abandoned after William Fitz Baldwin's death in 1096, and afterwards Henry I. gave it into the custody of a Welshman, who also had charge of Strath Towy; a passage which proves that Rhyd y Gors was in that district. It was restored by Richard Fitz Baldwin in 1104,[942] and is mentioned for the last time in 1105. After that the castle of Carmarthen, which has not been mentioned before, begins to appear, and its importance is clear from the continual references to it. Placed as it is on a navigable river, at the entrance of the narrower part of the vale of Towy, and on the Roman road from Brecon to St David's, its natural position must have marked it as a fit site for a royal castle. The castle is now converted into a gaol, and disfigured in the usual way; yet the ancient motte of William Fitz Baldwin still remains,

## partly inside and partly outside the walls. It is crowned with a stone

revetment which Colonel Morgan believes to have been erected at the time of the Civil War, to form a platform for guns.[943] The bailey is rectangular and covers about 2 acres. The motte is placed at one corner of it, on the line of the walls. On the outside it is now built over with poor cottages; but the site of the ditch can still be traced.

*LLANDOVERY, or Llanymdyfri, or the castle of Cantrebohhan.--It is referred to in the _Pipe Rolls_ of 1159-1160 by the latter name, which is only a Norman way of spelling Cantref Bychan, the little cantref or hundred, of which this castle was the head.[944] It was then in royal custody, and Henry II. spent nearly £60 on its works. But it had originally belonged to Richard Fitz Pons, one of the barons of Bernard de Neufmarché, and the fact that he held the key of this cantref goes to prove that it was from Brecknock that the Normans advanced into northern Carmarthenshire. The castle is first mentioned in the _Brut_ in 1115, when Griffith ap Rhys burnt the bailey, but could not take the keep on the motte.[945] It does not appear to have been long in English hands after 1159, but its alternations were many. The 25-inch O.M. shows an oval motte, carrying some fragments of masonry, to which is attached a roughly quadrangular bailey. This was one of the many castles by which the Normans held Strath Towy.

LLANSTEPHAN.[946]--This castle stands in a splendid situation at the mouth of the Towy, and was doubtless built to secure a maritime base for Carmarthen. The motte is of unusual size, semicircular in shape, one side being on the edge of the cliff; it measures 300 feet by 200 in the centre of the arc.[947] Such a size allowed all the important parts of the castle to be built on the motte; but there was a rectangular bailey attached, which is only imperfectly shown on the O.M.; the scarp is in reality well marked on all sides, and the ditch separating it from the motte is a very deep one. [H. W.] The towers that now crown the motte are not earlier than the year 1256, when the castle was destroyed by Llywelyn.[948]

DINEVOR, or Dinweiler.--Most Welsh writers associate Dinevor with the ancient residence of the kings of South Wales, but there appears to be some doubt about this, as the place is not mentioned before the 12th century.[949] Anyhow the castle was certainly the work of Earl Gilbert, as the _Brut_ itself tells us so.[950] In 1162 it was taken by Rhys ap Griffith, the able prince who attempted the consolidation of South Wales, and who was made Justiciar of that province by Henry II. It continued in Welsh hands, sometimes hostile, sometimes allied, till it was finally taken by the English in 1277. The existing ruins are entirely of the 13th century, but the plan certainly suggests a previous motte and bailey, the motte having probably been lowered to form the present smaller ward, whose walls and towers appear to be of Edward I.'s reign. The small bailey attached to this ward is separated from it by a ditch cut through the headland on which the castle stands.

KIDWELLY (Cydweli).--This castle, though in Carmarthen, was not founded by the conquerors from Brecknock, but by Normans from Glamorgan or Gower. Kidwelly was first built by William de Londres, in 1094.[951] The present castle shows no trace of this early origin, but is a fine specimen of the keepless pattern introduced into England in the 13th century.[952] There is no motte.

LAUGHARNE, or Talycharne.--Also called Abercorran, being at the point where the little river Corran flows into the estuary of the Taff. In 1113 this castle belonged to a Norman named Robert Courtmain.[953] The ancient features of the plan have been obliterated by transformation first into an Edwardian castle, then into a modern house. There is of course no motte. [H. W.]

*YSTRAD CYNGEN.--This must, we think, be the same as ST CLEARS, which stands in the Cynen valley, near its junction with the Taff. Welsh writers identify St Clears with the castle of Mabudrud, the name of the _commot_ in which it stands. First mentioned in 1154.[954] There is no notice of its origin, but the fact that a Cluniac priory existed in the village, which was a cell of St Martin des Champs at Paris, points to a Norman founder, and renders an 11th century date probable. It was a motte-and-bailey castle, of which the earthworks remain.[955]

*NEWCASTLE EMLYN.--This castle does not appear to have received the name of "the new castle of Emlyn" till after Edward I.'s conquest.[956] The new castle, which is quite Edwardian, was probably built on a different site to the old, as "on the other side of the bridge is a considerable mount, of a military character, which must have commanded the river. It may have been the original strong post occupied by the Normans."[957] In the 12th century _Pipe Rolls_ compensation is paid to William FitzGerald for many years "as long as Rhys ap Griffith holds the castle of Emlyn," which points to Gerald, the Seneschal of Pembroke, or his family, as its founders. It is on the very border of Carmarthenshire and Cardiganshire, defending the main road from Carmarthen to Cardigan.

LLANEGWAD.--This castle is only once mentioned, in the _Brut_, under the year 1203, when it was taken by the Welsh. A small motte, called locally Pen y Knap, with an earthen breastwork round the top, is still standing about a mile from the church of Llanegwad, and is all that is left of this castle. The position commands a fine view over the Towy valley, and it is noteworthy that it stands very near the supposed Roman road from Brecon to Carmarthen. [H. W.]

*LLANGADOG.--This castle also does not appear till 1203; it was razed or burnt at least thrice in five years.[958] A mound of earth on the banks of the Sawddwy River, near where the Roman road from Brecon is supposed to have reached the Towy valley, is all that remains of it.[959] Lewis says that it stands in a large oval entrenchment, and that the motte is of natural rock, scarped conically, and deeply moated.

CASTLES IN BRECKNOCKSHIRE.

BRECON, or Aberhonddu, the seat of Bernard de Neufmarché himself.--A charter of Bernard's mentions the castle.[960] It seems to have been a

## particularly strong place, as we do not hear of its having been burnt

more than once. The newer castle of Brecon is evidently of the time of Edward I., but across the road the old motte of Bernard is still standing, and carries the ruins of a shell wall, with a gatehouse tower.[961] A portion of the bank and ditch of the bailey remains; the whole is now in a private garden. The situation is a strong one, between the Usk and the Honddu. Brecon of course was a burgus, and part of the bank which fortified it remains.

[Illustration: FIG. 42.--MOTTE-CASTLES OF SOUTH WALES. Builth. Gemaron. Payn's Castle.]

BUILTH, on the upper Wye, _alias_ Buallt (Fig. 42).--A remarkably fine motte and bailey, presenting some peculiarities of plan. It is not mentioned till 1210,[962] but it has been conjectured with great probability that it was one of the castles built by Bernard de Neufmarché when he conquered Brecknock.[963] It was refortified by John Mortimer in 1242,[964] probably in stone, as in the account of its destruction by Llywelyn in 1260 it is said that "not one stone was left on another."[965] Nevertheless when Edward I. rebuilt it the towers on the outer wall appear to have been of wood.[966] Mr Clark states that there are traces of masonry foundations and small portions of a wing wall. The bailey of this castle consists of a rather narrow platform, divided into two unequal portions by a cross ditch which connects the ditch of the motte with that of the bailey. The ditch round the motte is of unusual breadth, being 120 feet broad in the widest part. The whole work is encircled by an outer ditch of varying breadth, being 100 feet wide on the weakest side of the work, and by a counterscarp bank which appears to be still perfect. The entrance is defended by four small mounds which probably cover the remains of towers.[967] The area of the two baileys together is only 1 acre. [D. H. M.]

*HAY, or Tregelli.--The earliest mention of this castle is in a charter of Henry I.[968] The present castle of Hay is of late date, but Leland tells us that "not far from the Paroche Chirch is a great round Hille of Yerth cast up by Men's Hondes."[969] It is shown on the 25-inch O.M., and so is the line of the borough walls.

*TALGARTH.--Mentioned in a charter of Roger, Earl of Hereford, not later than 1156.[970] A 13th-century tower on a small motte is still standing, and can be seen from the railway between Brecon and Hereford.

CASTLES OF RADNORSHIRE.

*RADNOR, or Maes Hyvaidd.--Though this castle is not mentioned in the _Brut_ till 1196, when it was burnt by Rhys ap Griffith, it must have been built by the Normans at a very early period. The English had penetrated into the Radnor district even before the Norman Conquest,[971] and the Normans were not slow to follow them. A charter of Philip de Braose is granted at "Raddenoam" not later than 1096.[972] There are mottes both at Old and New Radnor, towns three miles distant from each other, so that it is impossible to say which was the Maes Hyvaidd of the _Brut_. Both may have been originally De Braose castles, but New Radnor evidently became the more important place, and has massive remains in masonry. The town was a _burgus_.

*GEMARON, or Cwm Aron (Fig. 42).--Near Llandewi-Ystrad-denny. The _Brut_ mentions its repair by Hugh Mortimer in 1145.[973] The 6-inch O.M. shows a square central bailey of 1 acre, containing some remains of masonry, lying between an oblong motte in the S. and an outer enclosure on the N., the whole being further defended by a high counterscarp bank on the W. It commands a ford over the river Aran. There is no village attached to it.

*MAUD'S CASTLE, otherwise Colwyn or Clun.[974]--A ditched motte with square bailey on the left bank of the river Edwy, near the village of Forest Colwyn. The statement that this castle was _repaired_ in 1145 shows that it must have been older than the time of Maude de Braose, from whom it is generally supposed to have taken its name. It was rebuilt by Henry III. in 1231.[975]

*PAYN'S CASTLE, otherwise "the castle of Elvael."--First mentioned in 1196, when it was taken by Rhys ap Griffith. This is also a motte-castle (and an exceptionally fine one), placed on a road leading from Kington in Hereford to Builth. Rebuilt _in stone_ by Henry III. in 1231.[976] (Fig. 42.)

*KNIGHTON, in Welsh Trefclawdd.--First mentioned in the _Pipe Roll_ of 1181. The motte still remains, near the church. There is another motte just outside the village, called Bryn y Castell. It may be a siege castle.

*NORTON.--First mentioned in the _Pipe Roll_ of 1191. A motte remains close to the church, and two sides of a bailey which ran down to the Norton brook.

*BLEDDFA, the Bledewach of the _Pipe Roll_ of 1195-1196, when £5 was given to Hugh de Saye _ad firmandum castellum_, an expression which may mean either building or repairing. An oval motte, and traces of a bailey, are marked in the 6-inch O.M.

TYNBOETH, _alias_ Dyneneboth, Tinbech,[977] and Llananno.--First mentioned in _Pipe Roll_ of 1196-1197. There is a fine large motte in a commanding situation, and a crescent-shaped bailey, now marked only by a scarp. There are some remains of masonry, and the castle was evidently an important one. It is first mentioned in the _Pipe Roll_ of 1196, and it occurs in lists of the Mortimer castles in the 14th century.[978] It is not far from two fords of the river Ithon. [H. W.]

These four castles are not mentioned in the _Brut y Tywysogion_, though the _Annales Cambriæ_ mentions the capture of Bleddfa, Knighton, and Norton by the Welsh in 1262. They all command important roads. Knighton and Norton were boroughs.

[Illustration: FIG. 43.--MOTTE-CASTLES OF SOUTH WALES. Cardiff. Loughor.]

CASTLES OF GLAMORGANSHIRE.

_Cardiff_ (Fig. 43).--The first castle of Cardiff was certainly a wooden one; its lofty mound still remains. It is placed inside a Roman station, and the south and west walls of the castle bailey rest on Roman foundations, "but do not entirely coincide with those foundations."[979] The Roman fort was probably ruinous when Robert Fitz Hamon placed his first castle there, as on the N. and E. sides the bailey is defended by an earthbank, in which the remains of a Roman wall have been found buried. The area of the Roman Castrum was about 8-1/4 acres, and evidently the Normans found this too large, as they divided it by a cross wall, which reduces the inner fort to about 2 acres. The motte has its own ditch. The position of Cardiff was a very important base, not only as a port near Bristol, but as a point on the probably Roman road which connected Gloucester with Carmarthen and beyond.[980]

The lands of Robert Fitz Hamon, in the next generation, passed into the hands of Robert, the great Earl of Gloucester, Henry I.'s illegitimate son. He was a great castle-builder, and it is probable that the first masonry of Cardiff Castle was his work.[981]

NEWCASTLE BRIDGEND.--This castle and the three which follow are all situated on or near the "Roman" road from Cardiff to St David's, of which we have already spoken. There were two castles at Bridgend, the Old Castle and the New Castle, from which the town takes its name. The site of the former is now too much cut up for any definite conclusions about it; the site of the latter has been converted into market gardens, but a motte is still standing in one corner with the ruins of a tower upon it. [H. W.] This castle is not noticed either by the _Brut_ or the Aberpergwm version; the earliest mention known to us is in the _Pipe Roll_ of 1184, at a time when the castles of the Earl of Gloucester were in royal custody, and this appears to have been one of them.

KENFIG.--This castle is close to the "Roman" road. The Aberpergwm _Brut_ says that it was one of the castles of Robert Fitz Hamon, and states that in 1092 it was rebuilt "stronger than ever before, for castles prior to that were built of wood." This is a good specimen of the mixture of truth and error to be found in this 16th century MS. There is little doubt that all the first castles of the Normans in Wales were built of wood; but it is extremely unlikely that any wooden keep was replaced by a stone one as early as 1092. The town and castle of Kenfig are now almost entirely buried in sand-drifts, but the top of the motte, with some fragments of masonry upon it, is still visible. [H. W.][982] The note in the _Pipe Rolls_ of the repair of the _palicium_ of this castle shows that the bailey wall at any rate was still of wood in 1183. Even as late as 1232 the keep was only defended by a ditch and hedge; yet it withstood an assault from Llywelyn ap Jorwerth.[983] The bailey is said to contain 11 acres, a most unusual size. Kenfig was a borough in Norman times, and it is possible that this large bailey was the original borough, afterwards enlarged in mediæval times. There is evidence that there were burgage tenements within the bailey.[984]

ABERAVON.--The Aberpergwm MS. says that Fitz Hamon gave Aberavon to the son of the Welsh traitor who had called him into Glamorgan. At a later period, however, we find it in Norman hands. The site of the castle has been entirely cleared away, but it had a motte, which is still remembered by the older inhabitants. [H. W.][985] It is not mentioned in the _Brut_ before 1152, when it was attacked and burnt by Rhys ap Griffith.

*NEATH.--The site of the first castle of Neath was given by Richard de Granville, its owner, to the abbey of Neath, which he had founded.[986] About the year 1111, according to the Aberpergwm _Brut_, Richard returned from the Holy Land, bringing with him a Syrian architect, well skilled in the building of monasteries, churches, and castles, and by him we may presume, a new castle was built on the other side of the river, though the present castle on that site is clearly of much later date. The monks of course destroyed all vestiges of the first (probably wooden) castle.

*REMMI, or Remni.--Of this castle there is only one solitary mention, in the _Pipe Roll_ of 1184. The name seems to indicate the river Rhymney, which is the boundary between Glamorgan and Monmouth. We are unable to find any castle site so near the Rhymney as Ruperra, where Clark mentions a fine motte.[987] But we do not venture on this identification without further information.[988]

CASTLES OF GOWER.

*SWANSEA, or Abertawy.--This was the castle of Henry Beaumont, the conqueror of Gower. The present castle is comparatively modern. It is inside the town; but there used to be a moated mound outside the town, which was only removed in 1804. It seems probable to us that this was the original castle of Beaumont.[989] That this first castle had a motte is suggested by the narrative in the _Brut_ which tells how Griffith ap Rhys burnt the outworks in 1115, but was unable to get at the tower.[990]

*LOUGHOR, or Aberllychor (Fig. 43).--Also built by Henry Beaumont. The mound of the castle still remains, with a small square keep on top. There was formerly a shell wall also. The place of a bailey was supplied by a terrace 15 feet wide.[991] The four castles last mentioned are all at the mouths of rivers, as well as on an ancient (if not Roman) coast road.

*LLANDEILO TALYBONT, or Castell Hu.--Only mentioned once in the _Brut_, under 1215, as the castle of Hugh de Miles. A moated mound with a square bailey and no masonry still remains.[992] It commands the river Loughor, which is still navigable up to that point at high tides.[993] On the opposite side of the river is another motte and bailey, called Ystum Enlle. Possibly there was a ford or ferry at this point, which these castles were placed to defend.[994]

OYSTERMOUTH, a corruption of Ystum Llwynarth.--First mentioned in the older _Brut_ in 1215, when it was burnt by Rhys Grug. The later version says it was built by Beaumont in 1099. The castle stands on a natural height, fortified artificially by a motte, which is of great size. There is a small bailey below to the N.E., and a curious small oval embankment thrown out in the rear of the castle towards the N.W. The architecture of this magnificent castle is all of the Edwardian style, and as the castle was burnt down by Rhys ap Meredith in 1287, it is probable that only wooden structures stood on this site until after that date. The castle is in a fine situation overlooking the Bay of Swansea. [H. W.]

* * * * *

We have now completed our list of the Norman castles built in Wales which are known to history. It must not be supposed, however, that we imagine this to be a complete list of all the Norman castles which were ever erected in Wales. The fact that several in our catalogue are only once mentioned in the records makes it probable that there were many others which have never been mentioned at all. In this way we may account for the many mottes which remain in Wales about which history is entirely silent. As there was scarcely a corner in Wales into which the Normans did not penetrate at some time or other, it is not surprising if we find them in districts which are generally reckoned to be entirely Welsh. But there is another way of accounting for them; some of them may have been built by the Welsh themselves, in imitation of the Normans. As the feudal system and feudal ideas penetrated more and more into Wales, and the Welsh princes themselves became feudal homagers of the kings of England, it was natural that the feudal castle should also become a Welsh institution, especially as it was soon found to be a great addition to the chieftain's personal strength. The following castles are stated in the _Brut_ to have been built by the Welsh.[995]

1113. *CYMMER, in Merioneth.--Built by Uchtred ap Edwin, whose name, as we have already remarked, suggests an English descent. Near Cymmer Abbey the motte or _tomen_ remains.

*CYNFAEL, in Merioneth, near Towyn.--Built by Cadwalader, son of Griffith ap Cynan, on whose behalf Henry II. undertook his first expedition into Wales, and who was at that time a protégé of the Anglo-Normans. Clark gives a plan of this motte-castle in _Arch. Camb._, 4th ser., vi., 66.

1148. *YALE, in Denbigh = Llanarmon.--Said to have been built by Owen Gwynedd, but here, as we have said, an earlier Norman foundation seems probable (see p. 272).

1148. LLANRHYSTYD, in Cardigan.--Also built by Cadwalader, who was then establishing himself in Cardigan. Probably the motte and bailey called Penrhos, or Castell Rhos, to the east of Llanrhystyd village. [H. W.]

1155. ABERDOVEY.--Built by Rhys ap Griffith to defend Cardigan against Owen, Prince of Gwynedd. It must therefore have been on the Cardigan shore of the Dovey, and not at the present town of Aberdovey, which is on the Merioneth shore. And in fact, on the Cardigan shore of the estuary, about two miles west of Glandovey Castle, there is a tumulus called Domenlas (the green tump), which was very likely the site of this castle of Rhys.[996]

1155. CAEREINION.--Built by Madoc of Powys, who was then a homager of Henry II. Remains of a motte near the church; the churchyard itself appears to be the former bailey. About a mile off is a British camp called Pen y Voel, which _may_ have been the seat of the son of Cunedda, who is said to have settled here. [H. W.]

*WALWERN, or Tafolwern, near Llanbrynmair, in Montgomery, may have been a Welsh castle. It is first mentioned in 1163, when Howel ap Jeuav took it from Owen Gwynedd, who may have been its builder. The motte is marked in the O.M. on a narrow peninsula at the junction of two streams.

1169. *ABEREINON, in Cardigan.--Built by Rhys ap Griffith, Henry II.'s Justiciar of South Wales. "A circular moated tumulus, now called Cil y Craig."[997] (It is marked on the 25-inch O.M.)

1177. *RHAIDR GWY.--Also built by Rhys ap Griffith, no doubt as a menace to Powys, as this castle was afterwards sorely contested. It is a motte-and-bailey castle, the motte being known as Tower Mount.[998]

* * * * *

All these castles are of the motte-and-bailey type, and prove the adoption by the Welsh of Norman customs.[999] It will be noticed that in the first instances they were built by men who were specially under Norman influences. But probably the fashion was soon more widely followed, although these are the only recorded cases.

The contribution made by the castles of Wales to the general theory of the origin of mottes in these islands is very important. Leaving out the seven castles attributed to the Welsh, we find that out of seventy-one castles built by the Normans, fifty-three, or very nearly three-fourths, still have mottes; while in the remaining eighteen, either the sites have been so altered as to destroy the original plan, or there is a probability that a motte has formerly existed.

##