Berkeley Castle, Gloucestershire

Feature Sets (3)


Berkeley is situated on the English side of the Severn estuary in the Vale of Berkeley, some 10 miles SW of Stroud.  A stretch of the Little Avon River runs to the S of the village to enter the estuary just over a mile to the west.  The castle is on the SE edge of the village and consists of an approximately square courtyard with a roughly circular shell keep at its NW corner.  The Great Hall occupies most of the E range with the service quarters to the N of it and the State Lodgings to the S.  The Inner Gateway is at the S end of the W range, and the Outer Gateway is some way to the W of this.  The land falls away to the S and E, so that the castle dominates the view from Berkeley Heath.  Most of the castle is 14thc, but the shell keep and the Great Hall are 12thc in origin and retain some original features.

A castle was erected here in 1067 by William Fitz Osbern, Earl of Hereford (d.1071), comprising a motte and bailey with a timber keep on the motte.  In the mid-12thc the motte was surrounded by the present shell keep, and was levelled within the shell wall, so that the ground is some 20 ft higher inside than it is outside the shell keep.  This work was carried out by Robert Fitz Harding, immediately after he received the castle from Henry of Anjou (later Henry II) in 1153.  The keep is approximately circular and originally had round bastions to the NE, NW, SE and SW.  Only the NE bastion (with an elaborate 12thc window) and the SE bastion remain; the NW bastion was replaced by Thorpe’s Tower in the 14thc, and the remainder of the W stretch of the shell wall was destroyed by a Civil War battery from an emplacement on the roof of the nearby parish church.  It was consolidated but the SW bastion was not rebuilt, and the shell wall was never rebuilt to its original height on this side.  The keep is of reddish coursed sandstone rubble construction with pilaster buttresses, but this structure survives only on the S and SW sides, the remainder having been rebuilt.  On the E face of the keep, between the two surviving bastions, is a forebuilding in the form of a rectangular tower built against the shell wall.  It was not built with the wall, but added slightly later in the 12thc.  This contains the staircase leading to the elaborate main E entrance doorway to the shell keep.  The forebuilding has been considerably altered in later centuries.

Further 12thc fabric is found in the Great Hall, built against the E curtain wall.  The Great Hall was rebuilt in the 14thc and extensively restored and remodelled by the 8th Earl, who succeeded to the Berkeley estates in 1916, when he found the castle in an advanced state of disrepair.  As a result of these restorations, fabric which appears to be 12thc may not be original.  The rectangular hall has its high end at the south, and at the north is a screens passage with a gallery above, divided from the main space by a wooden screen brought from Caefn Mably (Glamorgan).  Along the east wall are three tall, round-headed window embrasures, and these are 12thc in origin but restored.  12thc chevron voussoirs have been reused in the doorway from the Keep Garden into Thorpe’s Tower, and these are also described here.  Finally the Treasury contains four carved stones of 12thc date.  Their provenance is unknown, and they may not be local or even British, having been amassed by the 8th Earl, a voracious collector.


Henry of Anjou, later Henry II, seized Berkeley Castleand its manor from Roger de Berkeley and in 1153 gave it to Robert Fitz Harding, a wealthy citizen of Bristol, nobly descended from Eadnoth the Staller, a Domesday landowner who had died fighting for his king.  Robert Fitz Harding had financed some of Henry’s early campaigns, and Berkeleywas his reward.  The problem with the Berkeleys was resolved by a double marriage: Robert’s eldest son Maurice married Roger de Berkeley’s daughter Alice (taking the name Maurice de Berkeley), and Roger’s eldest son married one of Robert’s daughters.  The castle has remained with the descendants of Maurice de Berkeley to the present day.


Exterior Features


Great Tower E doorway

In its present state the outer order arch, including the label and the right jamb are original 12thc work, and the jambs of the interior face of the doorway also contain some 12thc carved stone.  At some time, probably in the 13thc or 14thc, the doorway was reduced in size by inserting a plain, continuously chamfered, round-headed opening and a lintel made from reused corbels above it; the tympanum area and spandrels being infilled with dark red sandstone blocks.  The plain left jamb of the doorway is of similar blocks, and appears to have been rebuilt at the same time.  It seems fair to assume that the left jamb was originally similar to the right. The door way is well above the pavement, being accessed by three steps.

Height of 2nd order arch (inner radius) 1.05m
Height of door sill above pavement 0.55m
Height of doorway (2nd order arch inner apex to door sill) 3.45m
Width of chevron arch (inner diameter) 2.30m
1st order

Plain and continuous with a chamfer, probably 13th - 14thc. This is much smaller than the outer order arch, and the pseudo-lintel above the inner order arch is a row of five moulded blocks, four at least being reused corbels, as follows:

1. (L) A plain block supported by a plain quadrant corbel.

2. A corbel in the form of pair of thin vertical quadrants

3. Possibly a reused bas with its square socle below a chamfer.

4. A worn quadrant.

5. A tall corbel with horizontal features, badly eroded.  This is shown by Marklove (1840) as a triple roll.

2nd order

The plain S (left) jamb is rebuilt of red sandstone blocks from the pavement to the arch springing. The N (right) jamb is original.  It consists of a nook-shaft on an eroded base and a tall chamfered plinth, so that the lower end of the shaft is two courses above the internal floor level and the top step of the entrance. The shaft is carved with a regular design of interlacing stems, forming a trellis of lozenges, and in each lozenge in a worn suspended form, possibly a bunch of grapes. The shaft supports a capital which is worn but clearly cushion-shaped with a plain roll necking.  Each face is carved in relief with the same design.  A flat stem with a central row of beading rises vertically from the necking to the lower edge of the shield, where it splits into a pair of hollow stems that curve up to left and right, defining the edge of the shield.  They appear to terminate in leaves at the top angles of the capital, but general erosion renders their precise form unclear.  Within each shield is a pair of trilobed, fan-shaped leaves with fluted lobes, surrounded by a stem.  The curved spandrels below the shields are also filled with similar leaves with fluted lobes.  The impost block above the capital is heavy, chamfered and much worn with no trace of any carving remaining.  Nineteenth-century views show a chamfered impost with bosses or lozenge-shaped nailhead on the chamfer and two or three horizontal grooves on the face.  The arch is carved with frontal chevron consisting of two broad rolls alternating with three narrow ones. There are major losses in this area and only seven of the units of chevron, three at each end and one at the apex, are complete.  Outside the chevron archivolt is a chamfered label with a heavy lateral chevron roll on the chamfer and a narrow chevron roll on the face. The label is also badly worn, so that the design appears to vary from one voussoir to the next.


The interior face of the doorway is also of two orders: the inner (14thc) order having a triangular arch and both jambs of the outer retaining traces of point-to-point chevron.

Thorpe's tower doorway

Thorpe’s tower was added to the NW of the Great Tower in the 14thc, obliterating the bastion that originally stood in this position.  Its inner face is visible from the Keep Garden, and in this face is a tall, narrow doorway with a round-headed arch made of 11 reused chevron voussoirs, five to the left of a central, larger uncarved keystone and six to the right (plates 14, 15).  The lateral face chevron is hyphenated towards the extrados only, carved centripetally with a half-roll on the face, with a step inside it and a double-step outside, defining the edge of a raised triangle in each outer spandrel.  On three of the chevron units in the right section of the arch there are clear signs of similar chevron on the soffit, which would have formed a point-to-point design, but all the voussoirs have been shaved on their soffit faces, and in most cases all trace of the design is lost (plate 16).

The chevron hyphens vary in length, especially in the right-hand quadrant indicating that the edges of the voussoirs have been trimmed as well as their soffits, and that the original arch that the voussoirs formed was rather wider than the present one.  It must therefore have been a doorway rather than a window, and the presence of point-to-point chevron on the jambs of the Great Tower doorway raises the possibility that this arch was made of stones from its original inner order.  Detailed measurements could not confirm this suggestion (as the voussoirs have been trimmed) although they might well refute it. 


Great Hall E windows

Along the east wall are three tall, round-headed window embrasures.  The exterior of these windows was completely remodelled in the 14thc, but the interior embrasures  retain some 12thc features, albeit heavily restored. The two southern embrasures have continuously-moulded angle rolls between quadrant hollows on face and soffit and hollow-chamfered labels over the arches only.  The angle-rolls are cylindrical with extremely narrow fillets rather than true keels.  A wooden screen divides the northernmost of the three arches, and the stair to the gallery is contained within that embrasure.  The south jamb and arch of this embrasure are treated in the same way as the other two, but the north jamb rises through the gallery floor, and in place of the angle-roll on this jamb is a heavier nook-shaft on an attic base and carrying a double trumpet-scalloped capital with recessed shields and a keeled angle-trumpet, the keel being continued by the usual fine fillet over the plain chamfered necking and down the nook-shaft.

NE bastion window

This is the largest of the 12thc windows in the keep and the only one with any decoration, and served to light the apse of the Chapel of St John and to identify its position and status from the exterior. The right jamb is made of a single tall block with a shaft carved at the inner angle, a block below that forms a plain plinth (there is no base), and a block above carved with a single or double-scalloped capital with a plain necking and a high abacus.  Above this is a heavy, chamfered impost block that is considerably eroded.  The left jamb is similar but apparently had a separate angle-shaft, now lost.  Its upper block is so badly weathered that no sign of the capital remains.  The arch is of 14 voussoirs, ten of which are plain, modern replacements.  The remaining four: numbers 4, 5, 10 and 11 counting from the left, are of a lateral chevron design consisting of an inner roll, face hollow and quirked face roll with a cogwheel inner edge.  Voussoirs 4 & 5 are oddly cut; they form 1½ units of chevron together and neither of them is a complete chevron unit, either centrifugal or centripetal.  Voussoirs 10 & 11, in contrast, are normally cut as two identical centrifugal voussoirs.

Loose Sculpture

1. Lintel or frieze section

An approximately cuboidal sandstone block mounted on brackets on the wall.  As displayed the upper surface is rough – either broken or unfinished – the back could not be examined, and the other surfaces are tooled.  The centre of the lower back edge is chamfered with plain triangular chamfer-stops at each end.  The font face is carved in bold relief with a row of three vertical almond-shaped recesses with circular recesses above them, and part of a fourth similar motif cut off by the right end of the block. All recesses are framed by plain, narrow borders.  Alternating with these motifs are palmettes carved in relief, and below each palmette is a concave-sided vertical rectangle decorated with a row of vertical beading.  The block has a vertical crack running along the left edge of the central recess.


Length of chamfer 0.36m
Max. depth of block 0.23m
Max. height of block 0.40m
Max. length of block 0.75m

2. Bust-length relief figure

The figure is carved in relief on a slab of dark grey, fine-grained sandstone, set into the wall of the Treasury.  The block was originally squared at the bottom, as is shown by the remains of a plain raised border with two steps at the interior lower left angle.  Both lower angles have suffered major losses, while the upper part of the block, including most of the head, is lost.  The parts of the figure that project most, the hands and forearms, are worn smooth, suggesting that the surface has been deliberately flattened to some extent for use as building stone.  The figure is shown frontally with both hands raised to grip the vee-shaped collar of his tunic, or the cord securing his cloak.  There are remains of a beard, but nothing above this survives of his head.  The drapery folds are shown by groups of parallel curves, with each fold consisting of a flat fillet with one curved edge defined by a groove.  These swathes of drapery overlap on the right shoulder (the left is damaged), and the drapery of the chest is marked by two sets of concentric loops, side by side.  The cuffs are decorated with beading, but this is the only ornament shown.  Raised features on the flat ground that survives above the right shoulder may be the remains of hair.


Max. height of block 0.18m
Max. width of block 0.25m

3. Bird relief

The relief is carved on a slab of dark grey, fine-grained sandstone, apparently identical to that of the bust described above, and set into the wall of the Treasury.  It is irregular in shape, like a shoe in right profile, and no edges of the original block survive.  As it is set, it depicts parts of two birds; one above and the other below a flat, straight or slightly curved fillet that runs diagonally from the heel of the “shoe” at the lower left to halfway up the instep on the right.  All of the upper bird survives except its tail.  It is in left profile with wings folded against its body.  The head has a down-turned hawk-like beak with which it pecks at a fruit like a raspberry or a pine cone, an almond-shaped eye with a drilled pupil, and a neck that curves downwards.  All this is undecorated, but the wing feathers are shown as overlapping scales, with longer flight feathers indicated by parallel grooves.  The fruit is borne on a stem that emerges from under the bird’s body.  No feet are shown.  Below the fillet, only the wing and tail of a second bird, also in left profile, remain.  The tail is like a dove’s; long and scaled at its root with a fan of tail feathers indicated by grooves.  At the toe of the “shoe”, under the bird’s tail, are the remains of two fruits similar to the first.  Above the bird’s wing is part of a curved stem.

As it is exhibited the stone might well be the right way round.  The other possibility may be seen by turning it clockwise through 90 degrees, when both birds would be turned to a vertical disposition.


Max. height of block 0.15m
Max. width of block 0.28m

4. Finial

A slab in the form of a finial with, at the top, a central bud-like projection flanked by outward-curving leaves.  The sandstone block is carved on both faces.  On one side is a central, broad multiple stem that bifurcates to run into the leaves at the upper angles, where it terminates in pods bearing rows of pea-like beading.  To either side of the central stem is a pair of side-shoots ending in spade-shaped leaves.  At the foot of the stem it is clasped by a plain band, and to either side of this are the tips of leaves rising from below the edge of the block.  The edges of this face are framed by a plain raised border, except at the bottom where the block is broken off.  The reverse of the slab is incised with a simple, frontal human head with a tapering jaw, incompletely modelled, with almond-shaped eyes, drilled or gouged for pupils; the nose and brows indicated by a single incised line, and a thin, straight mouth.  The outline of a border has been incised to either side and above the head, and part of the background above the head has been crudely cut back.  The right edge is damaged, with losses to the border area.

There are traces of red paint on the plain sides of the block.


Height of block 0.21 m
Thickness of block 0.09 m
Width of block at foot 0.16 m
Width of block at top (max. width) 0.18 m


This type of stone forebuilding may have appeared first at Norwich Castle keep around 1100; earlier castles with first-floor entrances, such as Chepstow (Monmouth) and the Tower of London had wooden stairs to provide access.   After the mid-12thc the stone forebuilding was common, although the wooden form was still used at Goodrich (Herefordshire).   The forebuilding was not built with the Great Tower: the stonework does not course, and a buttress of the SE bastion has been cut back to accommodate it, but the time between the two campaigns cannot have been more than a decade or so.  It is suggested that the doorway was originally of two orders; this would explain the deep recessing of the rebuilt inner order and door opening.  In this view the outer order was as at present, but with a nook-shaft support on the left similar to that surviving on the right, and inside this was an order decorated with point-to-point chevron on the jambs.  When the inner order was remodelled, these chevron jambs were reused on the interior.  There is not sufficient evidence to decide whether the doorway originally had a tympanum, but if it did it is hard to explain why it was not simply left in place in the remodelling.

Frontal chevron similar to that on the upper doorway arch is found locally at Kilpeck in the 1130s and Leominster Priory in the 1140s, both works of the celebrated Herefordshire School (plate 11), but remained in use into the 1160s, e.g. at Beckford church (Worcestershire).  The more sophisticated point-to-point chevron, now set on the interior jambs, points to a date later rather than earlier in this range.  In sum the diagnostic evidence of the sculpture points to a date in the 1150s, as suggested by the documentary evidence, and the surviving cushion capital, carved in shallow relief with a more-or-less symmetrical Winchester acanthus design, and the nook-shaft, with its lattice of interlace, are also acceptable at this date (although the capital would be rather old-fashioned). In its original state the doorway was a very elaborate one, as befitted its position as the main entrance to the Great Tower.

For the NE bastion window it is impossible to supply a date any more precise than c.1125-60, in view of the paucity of the surviving sculpture, but a date in the 1150s or ‘60s, contemporary with the voussoirs of the Thorpe Tower doorway seems likeliest.

For the interior hall windows, the use of fillets and the keeled trumpet-scallop capital indicates a date at the end of the 12thc, c.1190-1200.

Turning to the loose sculpture which, it will be recalled, may well not be from the castle at all, the frieze section has a chamfer suggesting that it may originally have been a lintel or a window sill.  The deep indentations are probably simply decorative, intended to produce a play of light and shade across the surface, as on the Norwich castle keep doorway, or the west doorway arch at Tavant (Indre et Loire) in southern Normandy.  It seems unlikely that they were ever filled, e.g. with lead or glass; there are certainly no obvious traces remaining.  The palmette ornament that is the main element of the carved decoration was extremely widespread in the years between 1120 and 1150 and offers no guide to the place of production of this piece.

The bust-length figure may be part of a screen.  Half-length, framed figures are among the fragments at Canterbury cathedral normally associated with the choir-screen erected by William the Englishman around 1180.    However, it is not even certain that these figures were from a screen, rather than from the decoration of the cloister in which they were discovered, recycled as building material.  No twelfth-century carved stone screens survive intact in this country, and the parallels used in support of this attribution have come from Germany, and especially from the Liebfrauenkirche in Halberstadt and St Michael’s, Hildesheim; both decorated with relief figures in stucco.   Another possibility is that the stone formed part of a tomb, such as Archbishop Hubert Walter’s (d.1205), also in Canterbury, on which the gabled lid is decorated with a row of quatrefoils containing heads carved in relief.

The Canterbury screen reliefs show kings and prophets as well as unidentifiable human figures and grotesques.  Most of the human figures are shown frontally, pointing to the left or right, as if drawing the viewer’s attention to a central feature, probably a figure of Christ.  The Halberstadt figures are also half-length, and take the form of prophets with scrolls.  I know of no screen, and indeed no twelfth-century French or English figure, depicted in the same way as the Berkeley one, with both hands clasping the neck-cord of the cloak.  The pose does not correspond to any of the usual poses of Christ, while apostles shown as portraits usually carry books or, like martyrs, their instruments of martyrdom.  Prophets often hold scrolls and kings carry sceptres.  The absence of ecclesiastical vestments rules out the possibility of a churchman.  Whoever is depicted here is shown in an unusually hieratic pose; where attributes are not shown, as here, some kind of gestural performance is normal.

If the pose is unique, the style offers too many possibilities for certainty.   The earliest appearance of this kind of fold pattern is found among the cloister sculptures of Saint-Etienne in Toulouse (Haut-Garonne) signed by Gilabertus and dated to the second quarter of the twelfth century.   The style found its way to England as well as expanding over large parts of France, and the Virgin and Child in York Minster offers many points of comparison.   In view of the 8th Earl’s wide-ranging collecting habits it seems impossible to be sure of the origins of this piece, but a date in the second quarter of the 12thc is suggested.

The relief is assumed to be from the same source as the bust described above.  The stone appears to be the same, and the carving is of the same high standard.  Comparison of styles is not possible, as the two blocks include no similar features.  However, an examination of the style of the carving opens up local possibilities for the origin of these two reliefs.  Birds are common features of European Romanesque sculpture, but locally they appear regularly in the carvings of the Herefordshire School.   On a purely formal level the birds on this stone have much in common with the Herefordshire School examples: the hawk-like heads and beaks; the almond-shaped eyes; the long, fanned tails; and the treatment of feathers with fish-scale carving.  The Herefordshire School birds do not tend to have drilled eyes, however, and their wings are usually curved rather than straight, while their overall treatment is bolder and the poses more dramatic.  Nevertheless the formal similarities suggest a link through copying with the Herefordshire School examples, and it is interesting that Gloucester cathedral is seen as a possible training-ground for some of the Herefordshire School sculptors.

Turning finally to the finial, the differences between the two faces of the block indicate two possibilities:

1. The carving of the head face is incomplete.  The forms have been incised on the block, and a start made on the modelling of the head and on cutting away the background.  When complete the head would have appeared in relief and the composition framed by a border, as on the other face of the block.

2. The carving of the head face is secondary, in the nature of graffiti.  As originally carved this face was left plain.

In either case, one possibility is that this was the upper arm of a cross.  In the first case the incompleteness of the head of Christ must indicate that the work was abandoned, perhaps because the block fractured during carving.  In the second case the cross was not originally intended to bear a corpus (not all do), and an attempt was made to add one at a later date. This is not the only possibility.  The form of the finial suggests a stone bench-end, and while these are uncommon they are not entirely unknown.  On balance, the incomplete crucifix hypothesis (1. above) is a strong one, perhaps too attractive to abandon.  The forms of the foliage suggest a mid-twelfth-century date.


  • E. Amt, The Accession of Henry II in England: Royal Government Restored, 1149-1159. London 1993, 37-42.

  • A. Emery, Greater Medieval Houses of England and Wales, 1300-1500, 3, Cambridge 2006, 58-67.

  • D. Gaborit-Chopin et al, La France Romane au temps des premiers capetiens (987 - 1152). Catalogue of an exhibition, Paris, Louvre, 10 March – 6 June 2005, 286, 369.

  • Gloucestershire Archives (for drawings by Marklove and Lysons)

  • T. A. Heslop, Norwich Castle Keep. Romanesque architecture and social context. Norwich 1994, 33-34.

  • D. Kahn, Canterbury Cathedral & its Romanesque Sculpture, London 1991, 145-71.

  • R. B. Patterson, “Robert Fitz Harding ofBristol: Profile of an Early Angevin Burgess-Baron Patrician and his Family's Urban Involvement”, Haskins Society Journal, 1 (1989), 109-22.

  • H. M. Thomas, The English and the Normans: Ethnic Hostility, Assimilation, and Identity 1066-c.1220. Oxford 2005, 196-98.

  • M. Thurlby, The Herefordshire School of Romanesque Sculpture, Logaston 1999.

  • D. Verey, The Buildings of England. Gloucestershire: the Vale and the Forest of Dean, London, 2nd ed. 1976, reprinted 1992, 102-03.

  • G. Zarnecki, J. Holt and T. Holland (ed.) English Romanesque Art 1066-1200. Catalogue of an exhibition held London, Hayward Gallery, 5 April – 8 July 1984, 188.


Site Location
Berkeley Castle
National Grid Reference
ST 687 989 
pre-1974 traditional (England and Wales): Gloucestershire
now: Gloucestershire
medieval: Worcester
now: Gloucester
medieval: not applicable
Type of building/monument
Report authors
Ron Baxter 
Visit Date
28 August 2008