The castle extends along a limestone promontory, at the top of the tall, sheer cliff that forms the Welsh bank of the river Wye. Here the river turns eastwards from its general southern course, before turning south again to flow into theSevernestuary some 3 miles to the south at Beachley Point. The castle and town are thus contained within a loop of the river, so that the castle, on its lofty ridge, commands a considerable stretch of it in both directions. Inland of this ridge the ground falls away sharply to an area called the Dell. The castle is dominated by the late-11thcGreatTower, traditionally attributed to William Fitz Osbern, Earl of Hereford (d.1071). This must have been surrounded by a curtain wall, and the topography of the site suggests that it enclosed two baileys to east and west of the tower. When the castle passed to William Marshal in 1189, he rebuilt the baileys and increased their number, so that there are now three; the Lower, Middle and Upper Baileys. The complex thus forms a line of three baileys running along the ridge, with the main gatehouse at the eastern end of the Lower Bailey, and theGreatTowerrising between the Middle and Upper Baileys. Marshal’s work began in the Lower Bailey with the construction of the twin-towered gatehouse, and continued with the building of a wall between the Lower and Middle Baileys. In this were two round towers; one at the southern junction with the lower bailey curtain, overlooking the Dell, and the other towards the northern end. He rebuilt the southern curtain walls of the Middle and Upper Baileys overlooking the Dell, both connected to the walls of the Great Tower, and at the SW angle of the Upper Bailey he built Marshal’s Tower, a large rectangular structure with luxurious residential quarters over a kitchen, that may have served as a private retreat. After William Marshal’s death he was succeeded by his five sons in turn. In this period, between the death of William Marshal in 1219 and the death of the last two sons, Walter and Anselm, in 1245, an Upper Barbican with a simple gatehouse and a round SW tower was added to the western end of the Upper Bailey. TheGreatTowerwas remodelled in two phases, adding new windows to the upper storey, and adding an extra storey its west end to provide more accommodation. In 1245, the Marshal estates were divided between his five surviving daughters, with Chepstow going to Maud, the eldest of them. She was the wife of Hugh Bigod, 3rd Earl of Norfolk, and on her death in 1248, their son Roger Bigod inherited Chepstow. The next building work we know of was carried out by Roger Bigod’s nephew and heir, another Roger and the 5th Earl of Norfolk. Incomplete building accounts exist from the years between 1271 and 1304, documenting his construction of a grand suite of apartments against the north curtain wall of the Lower Bailey, overlooking the river, by the Master Mason Ralph Gogun ofLondon. Roger Bigod and Ralph Gogun also built the great Marten’s Tower at the SE angle of the Lower Bailey, and again remodelled the 11thc Great Tower, extending the Marshal brothers’ extra storey at the west end to cover the entire tower, adding corner turrets at the eastern angles, and inserting a floor over the hall.

The castle was a Royalist stronghold, held by Henry, 1st Marquess of Worcester, during the Civil War, and was successfully besieged twice; in 1645 and 1648. In the latter siege it was occupied by Royalist troops retreating from Cromwell’s advance to Pembroke; its battlements were razed and the interior bombarded with artillery shells. The Marquess of Worcester’s estates were forfeited to Cromwell, who carried out sufficient repairs to allow its use as a barracks. After the Restoration the Marquess was reinstated, but the king retained Chepstow as a barracks, appointing Henry, Lord Herbert, the heir, as governor. Henry continued the repair work begun by Cromwell, but after he succeeded to his father’s titles in 1667 he preferred to live at Badminton House, a Palladian mansion in Gloucestershire. In the 18thc much of the Lower Bailey was given over to industrial use; including a nail-making works, a glass-blowing retort for making bottles and a maltings. The industrial works were removed in the 19thc, and the interior cleared out and landscaped. Conservation of the fabric was begun towards the end of the century by the Beaufort estate, which then held the castle, and was continued by the Lysaght family who acquired it in 1905. In 1953 Mr D. R. Lysaght put the castle in the guardianship of the state.

Romanesque interest centres on the 11thc Great Tower, with its chip-carved east doorway and blind-arcaded interior. Some of William Marshal’s work also falls into our period, but there is little or no carved decoration surviving. It will be as well to describe the 11thc features of the Great Tower in some detail here before turning to its individual features. The tower is rectangular; 36m long and 14m wide (120ft by 45ft). It stands on ground that rises from east to west so that the undercroft floor slopes upwards in that direction, and the east end is carried on a high stone plinth. This may be entered by a doorway at the east end of the north wall, and contains no windows. Above this is the Lower Hall, to which the east doorway, originally reached via a timber staircase, gave access. The Lower Hall has three plain, round-headed windows in its north wall, and in the thickness of the east wall is a narrow staircase running from alongside the main doorway up to the NE corner of the Upper Hall, which was the main room. This hall has undergone a good deal of remodelling, but its 11thc form can be reconstructed to some extent. In the north wall was a row of seven round-headed windows. Only the easternmost of them survives, but the jambs of some of the others, formed of a distinctive yellow stone, can be made out. Between the second and third window from the west is a narrow, plain round-headed doorway that must have originally been accessed from an external staircase. The west interior wall has two plain oculi in the gable, and below them it is decorated with a blind arcade of four bays. The blind arcading continues along the S wall, but this wall has been remodelled. At its west end are four complete bays of blind arcading and the springing of a fifth, now blocked; while towards the east end is one bay with the springing of two more flanking it. To the west of this group is a round-headed window, and west of that two lengths of stringcourse divided by the traces of another, wider arch in the masonry. It has been suggested (Turner 2004) that the situation has been further complicated by the installation of a fireplace where the wider arch now is. Turner’s reconstruction of this wall shows a wide niche placed approximately centrally with four niches to the east of it and five to the west. The window is discounted as a later modification. He also suggests that the blind arcading may have continued on the east wall. The east wall now contains a pair of 13thc lancets, but disturbances in the masonry suggest that they may have replaced 11thc windows. The exterior of the Great Tower is articulated with pilaster buttresses, and a decorative band of reused Roman tiles runs right around it, just below the level of the apex of the east doorway, and rising to frame it. Construction is generally of large blocks of grey coursed rubble, but the pilasters are of yellow sandstone blocks, more accurately squared, that may be reused Roman material. The east doorway and the blind arcading are described in more detail below.

Chepstow (Estrighoiel, Strigoielg) appears in the Gloucestershire folios of the Domesday Survey, where it is recorded that Earl William Fitz Osbern built the castle, and in his time it rendered only 40s from ships going into the woodland. In the time of his son, Erl Roger (de Breteuil), however, the vill itself rendered £16 and Ralph de Limesey had half. Now the King has £12 from it. Roger de Breteuil forfeited his father’s estates to the Crown in 1075, and Chepstow was held by the Crown until c.1115, when Henry I granted the lordship to Walter Fitz Richard de Clare, whose family held the castle for most of the 12thc. Earl Richard de Clare died in 1176, and his estates passed first to his son Gilbert (d.1185) and then to his daughter Isabel, a minor. Isabel was thus made a ward of King Henry II, until her marriage to William Marshal in 1189. Marshal died in 1219 and was succeeded by each of his five sons in turn. The last of these, Walter and Anselm, both died in 1245 and the Marshal estates were divided between his five surviving daughters, with Chepstow going to Maud, the eldest of them. She was the wife of Hugh Bigod, 3rd earl ofNorfolk, and on her death in 1248, their son Roger Bigod inherited Chepstow. When he died in 1270 he was succeeded by his nephew, also Roger, the fifth earl ofNorfolk. This Roger died childless in 1306 and his estates were returned to the Crown. In 1312 Edward II gave the lordship of Chepstow to his half-brother, Thomas of Brotherton. Hugh Despenser the Younger, the king’s favourite, was granted the constableship of the castle, and in 1324 the castle itself and all its lands. At the end of the 14thc the castle passed to Thomas Mowbray, earl ofNorfolk. For details of its later history, see Turner (2002).

TheGreatTowerpresents an impressive sight from the approach fromEngland, where it is seen across the river, dominating the castle and town below from its site on the cliff. Until the beginning of the 19thc it was thought to be a Roman building, and its Imperial references may have been obvious to medieval visitors too. Architectural speculation has centred on the date of theGreatTower, and in particular on the question of whether it should be attributed to William Fitz Osbern. Until recently, authors have tended to follow Clifford Perks, author of the Ministry of Works guide, who accepted that the Domesday entry attributing the building of the castle to Fitz Osbern referred to the presentGreatTower, rather than to an earlier building on the site. If this is correct, theGreatTower, with Scolland’s Hall atRichmondCastle, may well be one of the oldest surviving secular buildings inBritain. Zarnecki (1998) dated theGreatTowerbetween 1067 and 1070, following Knight (1986 & 1991); Newman (2000) dated it within the period from 1067 and 1075; and Thurlby (2006) also prefers the early date. Turner (2004), on the other hand, has suggested that William I was a more likely builder and that the tower may have been started in the period between 1081 and the king’s death in 1087. The arguments are complex and not entirely convincing. First, the Domesday Survey records that Fitz Osbern also built castles at Richard’s Castle, Wigmore, Clifford, Ewyas Harold and possibly Monmouth, and at none of these has masonry been found that dates from his time. This is only to be expected, as the castles built by theNormansimmediately after the Conquest to secure their newly-invaded lands were generally of the motte and ring-work type, built of earth and timber. From this Turner deduced that Fitz Osbern would not have built a grand stone castle at Chepstow either, especially as it is situated far from the centre of his earldom (atHereford) on a potentially hostile frontier. It is arguable, of course, that a potentially hostile frontier is exactly where Fitz Osbern would choose to site his most impressive castle. Second, even if Fitz Osbern began the work in 1067 he would probably not have had time to complete it before his death in 1071; such a large and massive building may have taken between four and eight years to build, and the earl was constantly on the move in the last four years of his life. Of course this does not mean he did not begin the work. Third, theGreatToweritself is a very odd building, explicable only as a royal one, in Turner’s view. It contains no living quarters or garderobes, and its Upper Hall seems to be designed as a reception room for ceremonial purposes, such as receiving homage. The place of honour may have been in the centre of the long south wall, with its wide central niche and if this is so the occupant of this niche would be invisible to anyone entering by the east doorway. The humbler north doorway would offer the required view of the king in state, and Turner suggested that the elaborate east door was there to provide a frame for him when he showed himself to his subjects gathered in the Middle Bailey below. Turner linked this display of kingship to the Conqueror’s visit toSouth Walesin 1081, his only recorded visit to the principality, suggesting that it was begun at that time. If this was the intention there is no evidence that it was ever carried out; there is no record of the King’s having visitedWalesagain. It is by no means certain, of course, that the high end of the hall was at the south rather than the west, and Thurlby has pointed out that Castle Hedingham (Essex) offers a very close parallel for this kind of approach to the Upper Hall. It would certainly be unusual for the dais to be on a long side rather than a short one.

Depending on one’s view of the patronage, theGreatTowerwas begun either before 1071 or after 1081. Either date would suit the meagre sculptural ornament. Chip-carving, as found on the east doorway, appears as early as 1060 in thechurchofLa TrinitéatCaen, and the first securely dateable English example is inSt John’sChapel in theTowerofLondonin the late 1070s. The only other carved decoration is the cable moulding used on the imposts of the internal blind arcade. This appears inNormandyas early as the 1040s (at Norrey-en-Auge) and appears on the neckings of capitals from Jumieges in the 1060s. The earliest dateable British examples I know are atSt John’sChapel in theTowerofLondon(after 1070) and on capitals from Thomas of Bayeux’s York Minster (1070-1100). Both motifs continued in use until the 1120s, although the appearance of chip-carving after 1100 would be old-fashioned.

Thurlby (2006) found Anglo-Saxon features here, notably in the cable moulding, seen before the Conquest at Great Paxton (Hunts) and Hough-on-the-Hill (Lincs), and the external face of the central west oculus, cut into a squared flat frame, as at St Benet’s, Cambridge rather than made from voussoirs. Nevertheless he accepted that most of the carved decoration reflects developments inNormandy. An interesting parallel for the doorway is found atSt Leonard’s, Hatfield (Hereford). It has no chip-carving but its tympanum is of opus reticulatum and its three-part lintel has a joggled central section (stepped rather than wedge-shaped). Hatfield is important as it was held in 1086 by Hugh l’Asne, a member of Fitz Osbern’s household.