A Reinterpretation of the Gatehouse at Tonbridge Castle

A REINTERPRETATION OF THE GATEHOUSE AT TONBRIDGE CASTLE

david and barbara martin

Between November 1996 and January 1997 Archaeology South-East carried out a detailed programme of archaeological recording and interpretation work on the gatehouse at Tonbridge Castle on behalf of Tonbridge and Malling Borough Council. At the time the council were considering reinstating the floors and roof of the gatehouse. The archaeological recording was required as a prelude to planning these works, the objective being to recover evidence for reconstructing the ‘lost’ elements of the gatehouse (see ASE Project Ref. 546). The subsequent scheme of restoration was carried out over two phases. The missing ground and first floors were reinstated in 1999-2000, whilst the new roof, associated repairs to the upper parts of the structure and reinstatement of the missing elements of the tracery within the two second-floor two-light windows were undertaken in 2002-2003. An archaeological watching brief was maintained by Archaeology South-East throughout both phases (ASE Project Refs 1113 and 1499). This present article combines and summarizes the findings of all three ASE reports.

There have been a number of previous published articles on the castle in KAS publications. A paper by Sydney Simmons in Archaeologia Cantiana concentrated principally on the castle’s setting and repairs carried out to the fabric between 1954-66 (Simmons 1996). More relevant to the gatehouse itself is Derek Renn’s paper (Renn 1981) which points out the exceptional similarity, both in design and dimensions, between the gatehouse at Tonbridge and that at another of the de Clare castles, Caerphilly. A second article by Sydney Simmons in Archaeologia Cantiana outlined the ownership of the castle under the de Clare family (the builders of the gatehouse), and subsequently, and attempts to provenance the identities of the high-quality lifelike heads carved in stone which still adorn the main chamber within the gatehouse (Simmons 1998). This present article builds upon these earlier works by giving a more detailed account of the gatehouse resulting from the detailed archaeological survey and subsequent watching briefs.

The de Clare and Stafford families both feature in this paper. Gilbert, the last of the male line of the de Clares, fell at Bannockburn in 1314. In 1348 the Staffords obtained the Tonbridge estate by marriage of Ralph, Lord Stafford, to Gilbert’s niece, the only child of Gilbert’s sister and her husband, Sir Hugh de Audley [Wadmore 1886, 33-36].

the context of the gatehouse

Tonbridge Castle, built on the north bank of the Medway immediately to the west of an ancient road crossing, is essentially a Norman motte-and-bailey castle, subsequently reconstructed in stone. As Fig. 1 shows, the motte, with the remains of its shell keep, occupies the western end of an inner bailey which is roughly oval in plan. To the north of the complex was an outer bailey, and it was from this that the castle was entered via a gatehouse located on the northern side of the inner bailey. In its present form the gatehouse represents a rebuild of c.1270 (see below; no evidence of the earlier structure was discovered during the recent investigations). Apart from a sally port against the river, there appears to have been no other means of access to the inner bailey.

A survey of the castle made in 1520/1 immediately following the execution of Edward Stafford, Duke of Buckingham, gives a good impression of the complex as it then existed:

In the Lordship of Tonbridge, in Kent, is a castle which hath been and yet is a strong fortress, for the three parts thereof [i.e. west, north and east sides]; and the fourth part on the south side being fortified with a deep running water [the Medway], was intended to have been made for lodgings, and so resteth on 26 feet height, builded with ashlar, and no more done thereunto. The other three parts of the castle being continued with a great gatehouse or the first entry [the present gatehouse], a dungeon [donjon – the shell keep on the motte] and two towers [the Stafford Tower and the Water Tower to the south-east of the gatehouse] are substantially builded, with the walls and embatelling with good stone, having substantial roofs of timber, and lately well covered with lead, except the one half of the dungeon [keep] was uncovered.

And as unto the said gatehouse, it is as strong a fortress as few be in England, standing on the north side, having a conveyance, and well embatelled on both sides [i.e. a wall walk with parapets both sides], to the said dungeon [keep] on the west side; and on the south-east side there is a like conveyance [wall walk] to a fair square tower, called Stafford Tower [on the site of the present Tourist Information Office], and from thence to another fair tower standing upon the water nigh to the Town Bridge, being builded eight square and called the Water Tower. This castle was the strongest fortress and most like unto a castle of any other that the Duke had in England or in Wales (Wadmore 1886, 48-49).

In the past this description has been somewhat misinterpreted – the reference to a dungeon in particular has caused problems. It is for this reason that the quote given above is annotated. The description is wholly consistent with the evidence as it survives, though whether the assumption is correct that the lodgings on the south side were never built is open to question. It is equally possible that they had been demolished in antiquity and their former existence forgotten. Another interpretation might be that the earlier 13th-century lodgings had then relatively recently been demolished to the level of the inner bailey in order to make way for a reconstruction which was subsequently abandoned (Simmons 1996). Whichever is the case, the substantial garderobe chutes and basement stairs of high quality still survive on the southern side of the castle, leaving no doubt that this is where the Great Hall and its associated lodgings were intended. Foundations still existed in this area in 1782 (King 1782, fig. XXIX). If never built, then it can be assumed that an earlier suite of domestic buildings within the inner bailey remained in use throughout the medieval period.

Some commentators have identified the Great Chamber on the upper storey of the gatehouse as being the Great Hall of the castle. However, this can be discounted. An impressive room this chamber certainly was, but it has none of the hallmarks of a Great Hall, particularly one built to serve a man of de Clare’s status. The approach to the Great Chamber, by either one of two very steep, narrow staircases, without the use of quality display entrance doorways, is entirely inconsistent with the entrance to the hall of a great magnate. Indeed, both staircases are themselves accessed through very plain side rooms on the ground floor of the gatehouse. Further, the chamber is not designed to have a high and low end – it is entirely symmetrical about a central axis. In addition, the gatehouse does not incorporate either a kitchen or a set of necessary service rooms; the mechanism for the main portcullis occupied part of the floor space of the room in question and, with its near-flat roof and relatively low walls, it would not have given the feeling of grandeur so necessary in a Great Hall.

It is beyond dispute that the gatehouse doubled as a second strong point of the castle, complementing the pre-existing keep, but it was not the social centre of the complex. Its every-day role was almost certainly that of apartments and administrative centre for the castle’s Constable.

The castle was last brought into active service as a defensive stronghold in 1643, during the Civil War, at which date the floors, roof and fittings of the gatehouse were repaired and various items of furniture placed within it. The repairs are fully listed in a survey of the works compiled in 1646 (Camden Miscellany, iii, 44-47). By order dated 4 June 1646 Parliament’s County Committee for Kent ordered that the repairs and alterations made to the castle be taken down ‘thereby to slight and dismantle the fortifications’ (Wadmore 1886, 55), though as regards the gatehouse all that appears to have been undertaken was to dismantle the upper parts of the spiral staircases so as to prevent access to the roofs. Certainly the building was still roofed and in use when Edward King made his remarkable survey of the castle in the 1770s or early 1780s (King 1782, 231-375). In particular, regarding the Great Chamber, he states that this was

... of great dimensions, including the whole area of the three rooms beneath. It is now indeed divided into three such apartments as those are, but the walls forming the divisions are mere modern erections, of very late years, raised (as the proprietor informed me) on the top of the original ones on the lower floor, with a view to fit up a small room as a library; which design was afterwards laid aside (Ibid, 284).

It was about the time King was writing that the section of curtain wall to the east of the gatehouse was demolished (presumably to recover the stone for resale) – it is shown extant in an engraving by Buck made in 1735 and an undated engraving by Godfrey (Simmons 1996, figs 3 and 4) but had been dismantled by the time C.T. Dodd made his watercolour in 1770 (original owned by Tonbridge and Malling Borough Council). By that date the embankment which ran along the northern side of the curtain wall had also been dug away – presumably to infill the adjacent section of deep ditch – an act which left the base of the eastern garderobe chute high above the ground.

This deep ditch originally passed immediately in front of the gatehouse, being separated from it by a steep earthen bank, and was crossed by means of a drawbridge – perhaps the ‘drawbridge 34 foot by 9 foot valued at £12:0:0’ referred to in the castle yard (i.e. outer bailey) in 1646 (Camden Miscellany, iii, 47). King remembered the ditch being filled in. He writes,

‘I well remember the whole open: and have authority to add, that on digging at the bottom were found remaining the foundations of two piers, which supported the bridge; and which were constructed in a very remarkable manner, the stones being laid in pitch, mixed with hair, instead of mortar’ (King 1782, 273).

In fact, these foundations almost certainly represented the northern wall of a massive forework built upon the sloping bank in front of the gatehouse and which projected out from it and stood a full two storeys high (Fig. 2, see north front elevation). This work would have housed at its northern end the pivot for the drawbridge with, immediately behind it, a pit which would have accommodated the counterbalanced southern end of the bridge when raised. How this forework escaped previous detection is a mystery, for the toothings for its facings are clearly recognizable in the front face of the gatehouse, whilst the foundations for its western wall were uncovered during consolidation work carried out 1958-63. Yet, although noted by several commentators, the toothings are usually explained as scars caused by the drawbridge hitting against the wall. The facework on the towers and the sporadic, rough nature of the toothings are sufficient to prove that the forework represented an addition, yet almost certainly it was an early addition. Similar foreworks are known at Beaumaris, Anglesey (1295) and Goodrich in Herefordshire (14th- century) (Hemp 1936; Radford 1958).

The bank rising from the moat up to the western wall of the gatehouse is still shown in position in an undated watercolour by the artist John Inigo Richards (original owned by Tonbridge and Malling Borough Council) but this too has since been cut away leaving the foundations on the western side of the gate exposed and the base of the western garderobe chute elevated high above the ground. As a result of this latter work, the base of the gatehouse had subsequently to be refaced and buttresses added.

Two massive cracks now rise vertically up the gatehouse, completely isolating the north­-western corner from the remainder of the structure. It is tempting to suggest that the bank at this point was deliberately removed in order to undermine the gatehouse and cause its collapse, the motive being to recover the stone as a saleable product. (It is worth remembering that this is precisely what happened to the eastern curtain wall at about this date.) If this is so, the sappers only success was in creating the two cracks which exist today.

In 1792 a mansion (now the Tonbridge and Malling Borough Council offices) was added to the east of the gatehouse, and it was at this time that the basement rooms beneath the gatehouse were linked by a passage and brought back into commission to serve the mansion’s basement kitchens, to which they were linked by an intruded doorway. In its present form, the well within the eastern basement almost certainly dates from this period, though it may be located on the site of a predecessor. A second doorway was intruded higher up the gatehouse in order to form a link between the two structures, and both the eastern garderobe chute and portcullis groove were utilized as flues serving fireplaces within the mansion. By 1886 the gatehouse was roofless and in ruins (Wadmore 1886, plate facing p. 30).

the layout of the gatehouse

The gatehouse at Tonbridge is one of the earliest of a new design introduced during the middle years of the 13th century and developed over the next two hundred years. Their purpose was twofold – to strengthen a hitherto weak point in castle defences and to give adequate accommodation for the Constable. The solution was to build an immensely strong keep-like structure, effectively forming (like the Norman keeps) a last line of defence. Being an early example grafted onto an existing complex, Tonbridge did not replace the keep, but merely augmented it, the two being linked by a strong, protected walkway running along the top of the curtain wall. This walkway was protected by parapets on both sides, thereby giving adequate cover should the inner bailey fall. It is known from the 1520/1 description that a similarly protected walkway extended south-eastwards from the gatehouse to link it to Stafford’s Tower and the Water Tower beyond. Both entrances leading off the walkways to the gatehouse – that on the west at first-floor level; that on the east on the second floor – were protected by portcullises.

The gatehouse itself comprised a large rectangular block rising through three storeys with an unlit basement beneath (Figs 3 and 4). From the front (north) of the block, on either side of the entrance, project curved tower-like bays, whilst at the rear, at each corner, are built semi-circular turrets, each housing a steep spiral staircase. The curtain walls, which are set at a slight angle to the gatehouse, abut against the structure immediately to the north of the turrets, thereby leaving the gate projecting proud in front of the walls. This feature, together with its height, adds to the building’s impressively strong appearance. In the external angles against the curtain wall are projections – effectively wall thickenings – which each accommodate a pair of garderobes, as well as the flues to the fireplaces serving the lower rooms. It is worth noting that the two garderobes on the western side are located to serve the heated western ground-floor room and the heated western first-floor chamber, as well as the western wall walk. In contrast, the two garderobes on the eastern side are located at first- and second-floor level and thus serve the heated eastern first-floor chamber and the second-floor Great Chamber, as well as the eastern wall walk which, unlike that on the west, enters at second-floor level. Thus, both wall walks and all four heated rooms and chambers had direct access to garderobes, whereas the un-heated eastern ground-floor room, and the un-heated first-floor winding chamber did not.

On the ground floor a 3.25m (10ft 8ins) vaulted entrance passage leads through the building in order to give access from the outer northern yard of the castle complex to the more private and heavily defended inner bailey. This entrance was first protected by a long-since demolished barbican (mentioned in the 1640s, Wadmore 1886, 54), beyond which was a now filled-in deep ditch crossed by a drawbridge located at the end of a demolished forework mentioned earlier. The defences incorporated into the gatehouse itself consisted of an exceptionally large portcullis worked from the second-floor Great Chamber, and beyond that a stout pair of doors. Through the soffit of the arches immediately in front of the portcullis rise three murder holes accessed from the roof, whilst, should the portcullis fail or any enemy become trapped between it and the door, a further three murder holes rise from the vault to the floor of the Great Chamber. This area is also covered from the ground floor rooms by arrow-slit windows.

Should the inner bailey fall, a separate sequence of portcullis, doorway and murder holes protect the rear entrance to the gatehouse, though in this instance the space between the portcullis and doors is not protected by either murder holes or arrow slits. Both this portcullis and the central murder hole protecting it were operated from the first-floor chamber above the passage, but the outer pair of murder holes rose to the roof.

If the passage area itself fell there was still one further line of defence for those holding the gatehouse. Each of the two side doors which gave access from the entrance passage to the rooms on either side were fitted with their own portcullis operated from the chamber above, whilst within the floor of that chamber were no less than three sets of murder holes through which the assailants below could be attacked.

In plan, both of the ground-floor rooms flanking the gatehouse passage are near identical. The only major variation is that the room on the west was heated, whereas that on the east was not. There is, however, one other subtle variation. Beside the doorway leading into the eastern un-heated room is a small recess. This feature is not mentioned by either King or Wadmore (King 1782; Wadmore 1886) though it undoubted existed by this time. Renn notes its resemblance to a piscina, then adds ‘but was probably a lamp holder’ (Renn 1981, 96). What he fails to comment on is the feature’s even closer resemblance to a holy water stoup, often found near the doorway to a church or chapel and, to the authors’ knowledge, nowhere else. This correlation could be relevant, for an inventory of 1322 refers to ‘ornaments of the chapel’, which chapel is said to have been located on the east side of the port (Wadmore, 1886, 35). According to Hasted, the wall walk leading south-eastwards from the gatehouse to the Water Tower by the town bridge led ‘over the chapel’, thus placing it to the east of the gatehouse (Fig. 1), on the site of the Stafford Tower, near the present Tourist Information office (Hasted, 2nd edition 1798, 5, 199). He gives no authority for this statement – it could have been based upon tradition, or upon his interpretation of documentary references. Whatever the source, this location for the chapel has usually been accepted without question (Oliphant 1992, plan on page 33; Wadmore 1886). Wadmore, in common with some other writers, wrongly places the Stafford Tower near the town bridge. The description of 1520 makes it clear that the tower adjacent to the town bridge was called the Water Tower, with Stafford’s Tower sited between it and the gatehouse (see Wadmore 1886, 49). Based upon the known documentary references, an equally valid location for the chapel would be the room immediately to the east of the entrance passage – i.e. the un-heated eastern ground-floor room within the gatehouse itself. It may be relevant that, with the exception of the small central first-floor winding room (see below), this is the only room or chamber within the gatehouse lacking either a fireplace or garderobe. The main argument against this room being a chapel is that it is aligned with its long axis north-south rather than east-west, and thus it would have been very squat.

The western ground-floor room, with its canopied fireplace and adjacent garderobe, is every bit as well appointed as the two heated rooms on the first floor and has the appearance of a domestic chamber rather than a guard room. Most likely, the control point to the castle was located at the barbican at the far end of the drawbridge, and it is in this location that a guard room should be sought (Fig. 1).

Beneath both ground-floor rooms are un-lit basement areas; that beneath the western room having once been divided by a cross partition. It is relevant that neither of the two turret staircases descend to this level, and therefore access must have been by ladder via trap doors in the floors of the ground-floor rooms. Both basement areas were vented by small shafts rising diagonally through the south wall. Except for a small area of the eastern wall, the facing stones within both basements had been cut back in the 18th century in order to enlarge the floor areas of the basements. However, the depth to which the sockets for the ground-floor joists over the eastern basement penetrate the wall give an indication of the original likely alignment of the wall faces. Based upon this data, it seems clear that this eastern basement (and almost certainly the western basement) was markedly smaller than the rooms on the level immediately above them.

Access to the chambers on the first floor was by way of either of the two turret staircases, which were in turn reached by doorways in the corners of the ground-floor rooms. Apart from the locations of the arrow-slit windows and doorways, the eastern and western chambers were almost identical in layout and size to the western ground-floor room. As with that room, both were heated by good quality canopied fireplaces and both had convenient access to garderobes leading off the turrets. They almost certainly served as apartments for the constable and his household. Between these two chambers was a much smaller central winding room, located above the entrance passage and originally accessed from the chambers by doorways at the northern end of the now lost dividing walls. A step at each doorway led up to this room, the floor of which was set slightly above those of the adjacent chambers – a feature also to be found in the sister gatehouse at Caerphilly (see below). Because of the step, the storey height of this room was reduced to about 3.35m (11ft), compared with about 3.6m (11ft 10ins) in the chambers. It was a small room measuring only 5.95 x 3.65m (19ft 6ins x 12ft 0ins). From within this chamber the portcullis protecting the rear doorway and those protecting the side doorways leading off the entrance passage were operated. The mechanisms must have taken up most of the available floor space, effectively precluding this chamber from other uses (Fig. 3). It was an ill-lit room, for although it was served by a window in the south wall, when raised in its usual position the portcullis acted as a heavy grill across the opening, blocking much of the light.

Both turret staircases continued up to give access to the second-floor Great Chamber, a large symmetrical room measuring 16.0 x 8.4m (52ft 6ins x 27ft 7ins), with shallow cant-sided bays projecting into the front towers. Apart from carved and moulded candle brackets to the fireplaces on the floors below, this is the only room which incorporated architectural embellishment, and even here it is restricted to relatively simple mould-ings and figure-head carving on the stops and fireplace corbels. Never-theless, the quality of the carving is exemplary and almost certainly rep-resents living persons – probably the earl of Gloucester and his close family (Simmons 1998). This is also the only room served by two-light windows. It was not a particularly lofty apartment, a feature which would have been further emphasized by its near-flat roof. It should also be stressed that part of the floor space half way along the northern wall was taken up by the winding mechanism for the main portcullis, a piece of mechanism which must have been quite substantial to judge from the size of the portcullis. This mechanism was supported on top of the north wall of the central winding room below, which, because of the main arches of the entrance, was located to the south of that of the Great Chamber.

The floor of the chamber was carried on massive, very closely-spaced joists, clearly designed to take great weight. It is likely that this arrange-ment indicates a floor covering of either stone slabs or encaustic tile (see below). The likely status and purpose of this room was touched upon earlier. Clearly it was a room of assembly, but not a Great Hall. It was not designed for easy access and, like the remainder of the gatehouse, was primarily (if not solely) a male domain – a military headquarters. The corner turrets continue up to give access to the roof and its associated parapets. Two short straight flights of steps positioned against the southern parapet continue up from the roof in order to reach the turret tops.

Windows and Shutters

Within the whole building only four types of window were used:

- arrow slits

- single-light openings with cusped heads

- two-light openings with simple traceried heads

- small single-light garderobe windows.

Unlike the sister gatehouse at Caerphilly, with the exception of the very small garderobe windows, all openings facing outwards from the castle are of arrow-slit type, a feature which gives the building a very forbidding and austere appearance (Fig. 2). Arrow slits were also used exclusively within the two stair turrets and, as far as can be told, around the parapets. The slits measure about 80mm wide and have small trefoil cusped heads and three-quarter-round oillets at their base, all set within roll-moulded-and-chamfered surrounds. Internally the trefoiled head is carried back some distance – an exquisite detail. In all cases the internal jambs are splayed and the internal face of the cills are set well above the oillets, with a steeply sloping section linking the two. In this way the archer could achieve a good line of fire downwards towards the base of the gatehouse. Where the walls are particularly thick, as in the front of the gatehouse, walk-in recesses are incorporated against the internal face of the slits in order to give ease of access. None of these slits within the building show any structural evidence to suggest that they were ever closed by either glass, shutters, hangings or by other means.

The windows serving the four garderobes are extremely small, measuring only 235mm wide by about 480mm high (9ins x 1ft 7ins) to the apex of their simple two-centred heads. Thus they did not invite attack. None were glazed, though all show evidence of having been fitted with wooden hinged shutters. Given the narrow width of the openings, the shutters would have consisted of a single stout board, shaped to the window and hung on strap hinges fitted over the pintles. At least two of the windows show no evidence for a locking point to secure the shutter closed – in the others the evidence is lost. One window has a socket for a simple iron grill.

All non-defensive windows were located within the south wall, and thus faced towards the security of the inner bailey. Even so, except at second-floor level they were not large and in number were kept to an absolute minimum – just one window to each room/chamber. Of the five single-light windows, four were identical (see example, Fig. 5). They have double-chamfered external surrounds with a two-centred arch to the external order and a cusped trefoil head to the inner. Internally the jambs are double splayed and each opening is fitted with a window seat. The fifth window, serving the central first-floor winding room, varies only in that it is less tall and has single-splayed internal jambs without window seats. The height variation is explained by the higher floor level of this room, whilst the more simple internal arrangement reflects the fact that this opening served a workroom and was usually covered by the raised portcullis.

All five windows show clear evidence of having been fitted with heavy iron grills and of having been closed by wooden hinged shutters. In each case the shutter was hinged on the western jamb and, for added security, could be locked closed by means of a stout timber bar which slid within a purpose-formed cavity extending back a considerable distance into the eastern jamb. The western first-floor window has had a groove added to accommodate glass, but no such groove can be seen cut into the surrounds of the other openings, and thus these remained unglazed. The evidence for adding glazing to the western chamber suggests a higher status for its occupant – could this have been the constable’s personal chamber?

Only on the second floor – serving the Great Chamber – did the builder feel it sufficiently safe to install two-light windows. There are two, flanking a central fireplace. Both existed in damaged form in 1996, but survived in sufficient detail for their original form to be recovered with complete confidence. They are, without doubt, the finest architectural display features within the gatehouse. As can be seen from the reconstruction drawing (Fig. 6), in their original form they each comprised two 445mm-wide lights fitted with cusped trefoil heads, the two openings being capped by a single circular ‘bull’s-eye’ tracery-light. Although it is possible that this circular light was once fitted with soffit cusps which have since fallen, this seems unlikely given that the cusping to the heads below is carved integral with the main arches. Similar circular bull’s-eye lights exist within the hall windows of similar date at Stokesay Castle, Shropshire, where the possibility of former soffit cusps is likewise debated, though in this instance soffit cusps were used in the heads of the lights below (Wood 1965, 350).

Internally the jambs are adorned by simple filleted roll mouldings, whilst above both the internal and external arches are moulded hoods with carved head stops. Those carvings on the interior have every appearance of being from real life and could represent the Earl and his immediate family (Simmons 1998). As with the windows lower down in the gatehouse, both openings are fitted with window seats: the overhang of those to the western window was neatly trimmed back in antiquity.

There are clear indications for the locations of the former ironmongery and fittings associated with the windows. Neither were glazed, though in each case the lower lights were fitted with a heavy iron grill and were closed by hinged wooden shutters hung from the jambs. They closed against the central mullion. Large sockets in the jambs were noted during consolidation work carried out in the 1950s and 1960s, though whether these were for sliding locking bars is unclear – the sockets are now blocked. The bull’s-eye lights in the tracery were neither glazed nor fitted with either grills or shutters, and were thus always left open.

The details regarding the form of the shutters to the single-light and two-light south-facing windows can only be guessed. Their shape is clear from the shutter rebates, whilst, given the evidence for locking bars, they would have been of stout construction. They probably followed the practice used in external doors – two counter-boarded skins clenched together with heavy nails or studs. Always the boards of the outer skin were set vertically so as not to leave water traps which would have encouraged rot. Given the width of the lights, there would probably have been two boards forming the outer skin; the number of horizontal boards for the inner skin would have depended upon the depth of the boards used. The shutters would have been hung on a pair of strap hinges fitted across the internal face of the shutters and lodged over the spindles of the pintles fixed into the jambs.

Doorways

It is surely significant that all doorways within the gatehouse are very plain, having rather angular two-centred arches and simple continuously-chamfered surrounds. This simplicity, even austerity, is extended to the arches of the main entrance passage. This is not to say that the architecture is undistinguished or clumsy: on the contrary. In particular, the stepped multiple-chamfered orders of the arch over the main entrance is a striking feature, handled to great effect (see Fig. 2). The choice is surely deliberate, intended to give an aura of strength and utility. Despite the simplicity of the doorways, the quality of the workmanship is first class. All chamfered jambs terminate in neatly executed stops. The stops become more plain as one rises through the gatehouse; barred-and-scroll types with claw decoration on the doorways flanking the passage, with most other stops similar, but without the claw decoration. However, on the second floor the stops are of simple rounded type, with bar absent.

With the exception of the doorways in the two demolished walls which separated the first-floor chambers, all doorways survive intact and, except for some damage to the arch of the doorway at the head of the south-western turret, they remain in good condition. Today, several of the openings go unnoticed by visitors, being located within mural passages leading off the two stair turrets. Both lobbies leading off the wall walks, for instance, are isolated from the stair turrets by their own doorways, and all four garderobes are, as would be expected, fitted for privacy with their own doors.

It is fortunate that almost all openings show sufficient evidence to allow the door hanging arrangement and locking mechanisms to be reconstructed, and this adds considerably to our knowledge of how the gatehouse functioned. Only where modern door frames have been intruded is some of the evidence now hidden, and this includes the two doorways leading off the entrance passage.

Fig. 7 shows the swings of the doors in the entrance passage recon-structed, which emphasises the amount of detail which went into planning the structure. In particular, note the economy with which the entrance passage has been designed, with just sufficient room left for the leaves of the main doors to swing back against the side walls of the passage without fouling the side doors. It is likely that one leaf of each of these two large doors incorporated a pedestrian wicket gate – a not uncommon feature (Hewett 1985, 155-187). Unless these doors incorporated some form of grill, when all leaves were closed the passageway would have been in complete darkness. That the leaves were doors rather than gates is indicated by the closing rebates incorporated into the reverse face of the arches. However, it is known that by the 17th century six-foot high gates were fitted to at least one of the doorways, these being fastened by a swivel locking bar. Similar swivel locking bars were probably fitted to the original doors, perhaps with drop bolts for additional security.

Given that the two side doorways leading off the entrance passage were protected by their own portcullises, it is almost certain that these doors would have been secured by heavy locking bars which would have slid into purpose-formed sockets constructed into the wall thickness. Unfortunately, the detail is masked by modern door frames. However, it is significant that locking bars of this type were incorporated to secure the doorways at the foot of the stair turrets, thereby giving additional security to the upper floors. The only other doors considered sufficiently vulnerable to warrant heavy sliding locking bars of this type were the two doorways leading off the walkways extending along the curtain wall — both were also protected by portcullises. These barred doors would have been of heavy counter-boarded construction capable of withstanding battery.

All other doorways within the gatehouse, including the garderobe doors, show shallow sockets in the jambs to accommodate either the ends of sliding iron bolts capable of being worked from both sides of the door, or for key-operated wooden-cased locks. Some doors show more than one socket – this could indicate a succession of lock/bolt positions, or in some instances could indicate a door fitted with a bolt and lock. These bolted doors would probably have been of less sturdy construction than those secured by sliding locking bars.

One further point needs to be discussed. The two doorways leading from the turrets into the first-floor chambers each show pintle sockets in both jambs. It may be that this indicates doors of different periods hung to open in different directions. It is, however, worth noting that full-width doors in such a location would have been difficult to use – they would either have fouled other doorways or partially fouled the staircase passage. An alternative interpretation is that these two openings were fitted with doors of two leaves.

The Portcullises

No less than six of the doorways within the gatehouse were protected by their own portcullis, as evidenced by portcullis grooves in the masonry. By far the largest of these was the portcullis protecting the main entrance, the next largest being that protecting the inner bailey facade of the structure. The other four portcullises were small, protecting standard-sized doorways leading off the entrance passage and giving access from the walkways on the curtain walls.

All of these portcullises needed permanent mechanisms in order to lift and lower them. Portcullis mechanisms could either be located above the portcullis (lifting it directly) or stand to one side of it, lifting it indirectly by means of ropes running up over pulleys. Another factor which must be borne in mind is that, even when in its down position, there must be access to the top of the portcullis in order to maintain and repair the rope attachments. Thus, unless a special access panel is incorporated into the wall (as, for instance, at Bodiam Castle) the top of the portcullis must always stand proud of the floor of the winding chamber/platform. Tonbridge has no access panels, and therefore this rule must apply. This is important in calculating the total height of a portcullis, and thus the height of its top when raised. The height of the main portcullis at Tonbridge calculates at almost 9m (29ft 6ins). Given it measured 3.55m (11ft 8ins) across, this gives an indication of the weight which needed to be lifted. When in its raised position it would have formed an imposing feature against the north wall of the Great Chamber and would have reached virtually to the roof. If wound from above, the mechanism would have been located on the roof. This is not impossible, but would have made weathering the holes for the mechanism very difficult. More likely, the ropes were carried over pulleys and back down to a capstan standing upon the floor of the Great Chamber. This makes a great deal of sense, for the heavy capstan would have rested on top of the northern wall of the central first-floor chamber which, because of the recessed arches of the entrance, was located to the south of that of the Great Chamber itself. An extant mechanism of this type survives within the Bloody Tower at the Tower of London, though the age of the present mechanism is not known to the authors. It is worth noting that just above the floor of the Great Chamber, in the north wall, can be seen the sockets which accommodated the bars locking the portcullis in either its raised or lowered position.

That it was a side-operated type of mechanism which was used for the other five portcullises at Tonbridge can be proven. Those protecting the rear doorway of the entrance passage and the two doorways leading off of it had to be worked from the central first-floor chamber as otherwise the mechanisms would have been located within the Great Chamber, within its fireplace and the centre of the floor respectively – an impossibility. When raised, the portcullis which protected the rear door (which measured only 4½m high) extended almost to the level of the ceiling of the central chamber and formed a grill barring the window serving the chamber, thus robbing it of much of its light. A similar sit-uation still exists in the Monks Bar at York. The two wall sockets used to lock Tonbridge’s rear portcullis in its up and down positions were still hidden in 1996: they were discovered when a modern floor laid over the original was removed in 2003. The mechanisms which operated these three portcullises would have robbed the central chamber of most of its usable floor space.

Although only narrow, the portcullis protecting the western doorway off the curtain wall must have been exceptionally tall, this being necessary due to the low-level location of this doorway (on the first-floor) and the fact that the portcullis was operated from the roof. As there were no access points below roof level, the top of this portcullis must have stood high above the roof when raised, and therefore must have been operated from the side, as must that protecting the eastern door. Weathering the portcullis socket at roof level in these locations was no problem, as the portcullis grooves were sited external to the building.

Floors

None of the original timber floors within the gatehouse survive, but good evidence remains to indicate the layout and details of these. Only on the western side at ground- and first-floor level (where modern floors and roofs hide the evidence) and over the central first-floor chamber (where the evidence is lost) is there doubt as to the original design. For clarity, each of the three floors will be dealt with separately.

Due to the presence of a 19th- or early-20th-century replacement in the western ground-floor room, all evidence regarding the constructional details of this floor are hidden from view. However, given the design of the gatehouse, it is fair to assume that the details for this area of floor were identical to those within the eastern ground-floor room, and here sufficient evidence survives to allow the original design and layout of the joisting within this area to be recovered. The joist sockets indicate that the floor was carried by eleven joists aligned east-west and set at average 715mm centres, though the spacing is not as consistent as in the upper floors (see below). It is also evident from both the surviving sockets and their spacing that there were no joists set against the north and south walls. Presumably the ends of the floor boards merely rested on the now lost offset which, prior to removal of the ashlar facing in the basement, ran along the head of the basement walls. Because most of the ashlar to the lower walls has been robbed, only one socket gives a clear impression of the size of the joists – this was about 290 x c.280mm in scantling, considerably less heavy than those used in the two floors above (see below). The spacing between the joists was also much larger. This variation suggests that the floor was designed to carry less weight than the floors above.

When first surveyed in 1996 the joist sockets were blocked. When these were unblocked in 1999 to install the new floor details were recovered which indicated the method of installing the original floor joisting. Two alternative methods were used during the medieval period: either pockets were left in the walls and the joists slotted into position upon completion of the masonry by pushing each joist fully into an over-deep socket at one end and then pulling the joist back so that it engaged into the socket in the opposite wall. Alternatively, the walls were built up to the level of the joists, the joists laid into position and then built in as the next lift of masonry was constructed. Almost certainly the latter was the more usual method and this was certainly the procedure adopted at Tonbridge. This is evident both from the impression of the joists’ waney edges and hacked ends visible in the mortar along the corners and across the ends of the sockets, and by mortar adhering to the roofs of the sockets, against the former joists. This evidence results from mortar being pressed against the joist as the corework was infilled between the joists and from the mortar having been spread over the joists before bedding the next course of walling in place. A complete lack of mortar across the base of the sockets indicates that the joists were laid on the wall dry – that is, no attempt had been made to bed them in mortar. On the eastern side of the basement this procedure had resulted in 570-740mm deep sockets being left in the wall when the joists rotted/removed. In contrast, because of the wide offset in the face of the west wall at ground-floor level, at this end the joists were merely rested upon a ledge, flushed up with corework laid between the joists. As a result, when the joists were removed their impression was left as a chase or housing in the upper face of the offset. When the western wall of the basement was subsequently hacked back by 460-580mm, only the ends of these chases remained.

As the spiral staircases in the turrets do not descend down to basement level (which would have been an easy matter to have accommodated during construction, if thought desirable) it is fair to conclude that the two basement rooms were reached via trap doors in the floor. This was certainly the case in 1782, for King states these ‘having manifestly never had any way down to them, but by traps in the floors of the rooms just described: ...’ (King 1782, 282). Unfortunately, he gives no indication as to where the traps were located in his day, and this point is not answerable from the surviving evidence.

Sockets in the walls give a very clear indication of the layout of the first-floor joisting within the eastern and western chambers. The most interesting discovery made during the 1999/2000 opening up related to the constructional detail of the joisting. As with the ground-floor joisting, the joists at this level too had been laid onto the wall dry and then built in as the walls of the next storey were constructed. However, in this instance the floors over each of the two guard rooms were fully framed: that is, rather than comprising a set of independent joists, the ends of the joists were ‘linked’ by being morticed-and-tenoned into trimming joists of similar scantling. Each floor had been framed-up lying on the uncompleted tops of the wall and then built in as the next storey was constructed. As a result, each trimmer was buried within the thickness of the wall. Despite the splay of the towers incorporated into the north elevation of the gatehouse, the floor was framed-up with right-angled corners, and thus the ends of the northern joists were buried deep into the walls (Fig. 8). Presumably the trimmers were incorporated into the floor construction so as to act as a ring beam. What is unclear is whether the builders considered the trimmers necessary for purely structural reasons or to give the gatehouse additional strength should it be subjected to bombardment. Perhaps they had both objectives in mind!

The joist sockets in the walls of the eastern chamber indicate con-clusively that the floor was carried by 13 joists 345-370 x c.320mm in scantling and aligned upon an east-west axis. Their eastern ends were built into the eastern external wall at the time of construction, and this was almost certainly the case at their western end too, though here the upper part of the wall has been lost. Despite an area of brick patching beneath the eastern first-floor fireplace, opening up proved there to be no sockets in this area. Thus at this point the ends of three joists must have been carried by a trimmer. Rather than relating to joist sockets, this area of brick patching appears to have represented a repair replacing some form of stone hearth associated with the fireplace. This hearth would have been ‘cradled’ by the flanking joists and trimmer. Despite the fact that the joists were not built into the wall at the fireplace, all the indications suggest that the trimming timber built into the thickness of the wall and linking the ends of the main joists continued through behind the fireplace, though it should be stressed that this point could not be proven.

When surveyed in 1996/7 the evidence for the joist sockets for the floor within the western chamber were masked from view by a modern roof. It was assumed that the layout would have been identical to that within the eastern chamber, although it was noted that an inventory compiled in 1646 mentioned 14 joists within this area, as opposed to 13 within the eastern chamber (Camden Miscellany, iii, 44-47). What the opening-up carried out in 1999/2000 made clear was that it was a mistake to dismiss the content of the 1646 inventory as being erroneous/misleading, for here there were sockets for fourteen, not thirteen joists. There were other variations, too, between the floor of this chamber and its companion to the east. Although the trimmer within the wall thickness of the eastern chamber was probably a continuous timber, passing behind the fireplace, that running within the western wall of the western chamber most certainly was not, for here the line of the trimmer was interrupted by the flue rising from ground-floor fireplace. Here too the joists immediately in front of the first-floor fireplace proved not to be built into the wall, but were instead trimmed around the fireplace’s hearth.

The level of the joist sockets show the top face of the joists carrying the boards of the floor within the eastern and western first-floor chambers were located lower than the level of the original masonry floor within the central first-floor ‘winding’ chamber, supported by the vault to the entrance passage. When surveyed in 1996 the assumed floor level within the central chamber was ascertained by taking a measurement through a pair of open murder holes to the underside of a concrete slab forming the 20th-century flat roof. Opening up work undertaken in 2002/3 showed that these murder holes had been extended upwards when the new roof was added above the entrance vault. Removal of the modern roof revealed that some of the purposely rebated ashlar blocks which formed the original tops of the murder holes remained in situ. The rebates were designed to accommodate the covers or trap doors which would have been fitted over the holes when not in use. Both these and the level of the locking points to secure the southern portcullis conclusively indicate that the floor of the winding chamber was lower than was thought to be the case in 1996 and was located 310mm above the top face of the joists within the adjacent chambers. As the thickness of the original floor covering supported by the joists is unknown − it could have been formed using either timber floor boards or stone slabs perhaps up to 75mm thick − the height of the step up into the winding chamber is uncertain, though it would have ranged between 235-290 mm. This range is sufficiently small to indicate that there would have been only one step located at the doorways, not three as originally envisaged and as suggested by the sister gatehouse at Caerphilly.

As on the floors below, joist sockets give a very clear indication of the constructional details of the lost second floor. The bricked-in sockets in the east and west walls indicate conclusively that the floor was carried by massive 370-420mm wide x 380mm deep joists aligned upon an east-west axis, with their ends built into the wall at the time of construction. The floor would have comprised three sets of joists, one set corresponding to the western first-floor chamber, another to the central first-floor chamber and the third to the eastern first-floor chamber. The approximate clear spans of these joists would have been 4.75m (15ft 7ins); 3.60m (11ft 10ins) and 5.00m (16ft 5ins). The ends of the joists away from the external walls would have been built into and carried by the now lost massive internal partitions which divided the three chambers and which are evidenced by wall scars. How far the joists were built into the demolished partitions is indicated by a pair of floor scars which represent joist sockets cut lengthways when the partitions were demolished. The external sockets in the east and west walls show that there were sixteen joists making up the eastern and western sets. Because the partitions have been lost, the number of joists to the central bay of the floor cannot be ascertained, though there would have been fewer here, not only because of the narrower width of the Great Chamber at this point, but also because of the thick wall which formed the northern wall of the central first-floor chamber. This wall rose only as far as the level of the second-floor, and thus here the top of the wall would have formed part of the Great Chamber’s floor.

A noticeable feature of the joist sockets in the east and west wall is that they are absent from the central section. Here, in each case the central two joists could not be built into the wall due to the close proximity of the fireplace flues to the face of the wall. The ends of these joists must have been carried by a trimmer set against the face of the wall.

One further point needs to be discussed. Even allowing for the number of persons and possible weight of furniture the floor was intended to carry, the joists are of exceptional scantling and are placed at exceptionally close centres. This is even more remarkable because the joists of the floor below which were of identical span, were notably smaller and were set much further apart (see Fig. 9). There must have been a very specific reason for increasing the strength of this floor. A hint is to be found in the location of the portcullis locking sockets in the north wall of the chamber. These sockets were intended to accommodate the ends of timbers which slid across the floor to lock the main portcullis in place – either when raised or closed. Both sockets are located 65mm above the top face of the wall scars, suggesting floors of similar thickness. Although all the evidence is circumstantial, it seems likely that the strength of the joisting was increased in order to carry some form of stone-paved floor, either in the form of stone slabs or, as in the service chamber at Stokesay Castle, Shropshire, a floor of encaustic tiles bedded on mortar.

Stairs

There were originally four flights of stairs within the building – two spiral staircases (one in each turret) and two narrow straight flights rising from the roof to the turret tops. In addition, there must have been ladders giving access to the two basement areas, but whether these were removable or fixed is not known.

There is an oft-quoted rule that in military buildings all spiral staircases turn clockwise in order to ensure that the right-handed defender has his sword arm free. Clearly this does not apply at Tonbridge, for here the two staircases are a mirror image of one another, that in the south-western turret winding anti-clockwise, that in the south-eastern clockwise. Both staircases are of standard central-newel type, with very tall steps averaging 280mm high. The flight is lit by arrow slits and gives access to landings, which in turn lead to the chambers, garderobes and the wall walks running along the top of the curtain wall. In each case the staircase has been truncated: that within the south-east turret now ceases one step above second-floor landing, whilst that in the south-west stops three steps above the first-floor landing. They formerly rose to doorways which gave access to the main roof. In each case the upper five steps were located within a straight flight rising through the thickness of the turret wall. It is likely that the upper parts of the flights were deliberately dismantled, probably as a result of the order dated 4 June 1646 which stated that the tenant was to ‘take them [i.e. timber, platform, planks etc] down, thereby to slight and dismantle the fortifications’ (Wadmore 1886, 55). Given the evidence, it would appear that his actions were restricted to preventing access to the roof, thereby leaving the gatehouse serviceable as accommodation.

The spiral staircases never gave access to the tops of the turrets. Instead these were reached by a pair of narrow, steep, straight flights rising against the southern parapet wall (Fig. 10). Although it has lost its lower and upper steps, as well as much of its northern parapet wall, the western flight still survives in reasonable condition. The eastern flight is now very fragmentary.

Fireplaces and Chimneys

Apart from the eastern ground-floor room (possibly a chapel) and the central first-floor winding chamber, all apartments within the gatehouse were heated. Thus there were four fireplaces. All are damaged, that within the Great Chamber being the best preserved and that within the western ground-floor room the worst, the latter now being evidenced by its cut-off candle brackets and the remains of its flue only. Despite this, even here sufficient survives to indicate that in its proportions it was identical to those on the first floor which, as far as can be told, were virtually identical to one another. In comparison, that serving the Great Chamber was wider and more lavishly finished. As would be expected at this period, all four fireplaces were of canopied type – an impressive design which was ubiquitous until the mid-late 14th century in buildings which could afford fireplaces.

Being the most complete, it is probably best to first consider the fireplace serving the Great Chamber on the second floor. Fig. 11 illustrates it in its original form with all conjectural elements indicated in broken outline. Its rear wall, which slopes back as it rises, is constructed of very-finely-tooled and neat-fitting blocks of white limestone – the only use of non-local stone (not identified) recognized within the building. At the base can be seen the broken-off remains of the projecting hearth which extended under the chamfered, slightly projecting jambs which flank the fireplace. These jambs helped support the heavy corbels which carried the canopy of the fireplace. Of the corbels, that on the west survives, whereas that on the east has been lost. The survivor has carved in relief a human face with damaged nose and forehead, but otherwise intact. The figure has flowing locks of hair and is probably female – if so, then the lost eastern corbel is likely to have been male. Running up the external edge of each of the two jambs is a shaft rising from a moulded base and supporting a candle bracket which, unlike those of the other fireplaces, is located below the level of the corbel. The lower surface of the bracket is carved with a somewhat boorish face. In each of the two examples the supporting shaft appears to enter the figure’s mouth. The canopy, supported by the corbels, has been entirely lost, although its shape is recoverable from the scars in the face of the wall where its side cheeks have been broken off. Thus, although the precise angle of the front face cannot be recovered, nor also the details of the lintel’s lower edge, the general arrangement is known with certainty. On the hoods of the first-floor fireplaces the top and bottom of the slope was defined by a projecting string course – surprisingly, it can be certain from the remains that there were no such string courses in the Great Chamber fireplace, at least to the side faces. The canopy of this fireplace varies from those on the level below in one other respect – here the chimney cap projects from the wall and thus the head of the canopy likewise projected.

The remaining three fireplaces are probably best considered by describing the form of the most complete example – that within the eastern first-floor chamber (Fig. 12), though as already noted, all three appear to have been near identical. Only at hood level do any original features of this fireplace survive, the projecting jambs having been destroyed and the fireplace recess blocked by brickwork, but sufficient survives at the upper level to indicate the width of the hearth area, which was relatively narrow in comparison to its height. Despite the damage, the candle brackets which flanked the canopy survive. These take the form of moulded caps, three-quarter round in plan. The bell of the southern cap is supported on carved foliage, and there are indications of carving beneath the bell of the northern capital too, but this was less ornate and appears not to have been foliated. Rising from the internal edges of the candle brackets can be seen the scars left by the cut-back sloping side walls of the former projecting hood: indeed the upper section still survives complete with its sloping front face. This allows the projection of the original hood to be calculated. Capping the hood is a short projecting roll-moulded string course with a face fillet. The scars of the broken-off hood shows indistinct, but conclusive, evidence of there having been a similar projecting string course running around the hood immediately above the candle brackets. As is common practice, this would have defined the junction between the sloping sides of the hood and the vertical walls of the lintel which supported it. Of the lintel, no details are known.

Garderobes (Fig. 13)

All four heated apartments within the gatehouse are served by garderobes, albeit two have to share the facilities with guards on duty on the wall walks. The garderobes are located within small mural chambers leading off the staircase landings, those on the eastern side being set at first- and second-floor levels, whilst those on the west were located on the ground and first floors. Both garderobes on the eastern side are damaged, that on the first floor having been converted into a passage linking to the mansion, whilst the seat of that on the second floor has had a late 18th-century flue thrust up through it. Each garderobe was equipped with its own lockable door, a small window closed by a shutter and, of course, a bench seat. None of the benches survive intact, though the scars for the former front walls and the chases which accommodated the seats are clearly visible. In all cases the height of the bench from floor to top of seat was 430-480mm. Chutes descended vertically from beneath the seats in order to take away the cess, two shafts being set within each of the chutes.

Waste disposal associated with medieval garderobes was by one of two basic methods – either the cess was expelled into a running ditch, steam, river or moat, or alternatively it was collected at the base of the shaft for manual removal. Cess pits were utilized where emptying was intended to be irregular, but otherwise there was merely an opening at ground level at the base of the chute. Despite the door-like openings at the base of the Tonbridge chutes now being elevated well above ground level, the Tonbridge examples were of this latter type, as too were those serving the sister gatehouse at Caerphilly. It is because of later changes to the external ground levels that the openings are now elevated – originally the castle banks rose to this level. Sensibly, the entrances to the chutes are sited outside the castle, adjacent to the curtain wall. At first sight this would appear to present a weak point in the defences, but in truth it was easy to fight off an attack by anybody foolhardy enough to try and enter the castle by climbing the chute and forcing their way through the hole in the garderobe seat!

The Roofs

There is sufficient evidence in extant weathering courses incorporated into the south wall to indicate the level of the roof covering. Furthermore, sufficient of the weathering courses survive to show that the main roof was pitched almost flat, at between 2½ - 7½ degrees and could therefore be classified as a ‘flat roof’ (see Fig 9). The only falls incorporated were those necessary to discharge rainwater. The level of the extant weathering course indicates that there was no fall towards the western wall, but rather away from it. Indeed, the higher of the pitches stated in the range given above is calculated taking into account the level of the weathering course over the western wall. Allowance should be made for the fact that this weathering course was deliberately set high in order to obtain an adequate fall around the south-western turret, towards the rainwater outlets in the south wall. Taking this into account, the weatherings on the side of the southern chimney cap (which are set to about 4 degrees) probably give the more accurate assessment of the original roof pitch.

Such shallow falls must have been designed to carry a lead covering. This is certainly known to have been the case when an extent of the castle was drawn up in 1520/1 (Wadmore 1886, 48-49). The rainwater from the roof (or at least the southern half of it) was channelled through openings on either side of the main chimney in the southern parapet wall, the leadwork presumably having been dressed through the openings to form spouts. At the walls, the lead would have been dressed up the facing to abut hard against the underside of the projecting weathering courses in order to form both a gutter and a watertight joint.

So, it is possible to ascertain the ‘pitch’ of the roof with a fair degree of certainty, but was there a continuous fall from north to south, or did the roof rise to a central east-west ridge, with the southern half draining southwards and the northern half northwards? Unfortunately, the northern wall is now too low and too heavily conserved for any rainwater outlets to remain visible. Because of the curve of the towers, if the northern half of the roof sloped northwards at least two outlets would have been needed – one set roughly central of each tower. Perhaps of some assistance are the two watercolours painted of the north front by C.T. Dodd and John Inigo Richards during the late 18th century. They are from different angles. The drawing by Richards appears to show an outlet (or a possible outlet) towards the top of the western tower and another possible outlet over the main entrance. Likewise, Dodd seems to show a matching outlet in the eastern tower, and he too seems to show the opening over the main entrance. In both instances the features are shown indistinctly as small black rectangles. These drawings are by no means proof, but add weight to support the hypothesis that the roof had a north and a south slope, rising to a very low central east-west ridge, with rainwater outlets in the north and south walls. This is the option illustrated in Fig. 9.

It should be remembered that the roof was repaired during the Civil War and still survived when surveyed by King in 1782. Unfortunately the survey of 2 June 1646 adds nothing. However, King’s description provides a little help – he states:

The ceiling of this room [i.e. The Great Chamber] was still more remarkable than the floor; being no less than 3 feet in thickness; designed manifestly to support not only the lead of the flat roof, but moreover the great weight of balistas, catapultas, and other engines of war, placed there occasionally (King 1782, 285).

This tells us that, at its thickest point, the roof measured 3 feet, which is, in fact, roughly consistent with a 4 degree pitch rising to a central east-west ridge and measured from the underside of the tiebeams. This description also confirms that the roof was ‘flat’ and covered in lead, though these points were not in dispute.

Because the turrets and turret staircases rise direct from the wall tops, any tiebeams in this area had to be built into the wall. The sockets for one of these tiebeams was discovered during consolidation work in the mid 20th century and was photographed. Because of the close proximity of the socket to the two-light window and its relationship with a brickwork repair, the soffit level of the tiebeam can be calculated from the photographs with a fair degree of accuracy. It measured approximately 450mm deep. Thus, the ends of the tiebeams must have rested on the top face of the north and south walls, probably carried over a wallplate. This point is largely confirmed by the top level of the two corbels incorporated into the north face of the chimney in order to trim the roof construction around the flue. These are set at an almost identical height to the known level of the top face of the north wall. Given the known location of the tiebeam socket, of a matching patch of walling at the opposite end of the same wall, and taking into account the location of two other possible tiebeam sockets, can anything be said about the spacing of the tiebeams? Unfortunately, the precise location of the tiebeam within the best preserved socket is not known. All that can be said is that the centre-line of the tiebeam was set approximately 2.3m to the east of the western wall, and that a second possible tiebeam socket is located with its centre-line approximately 4.5 metres to the east of the wall. This might suggest a roof of seven equal bays, but it would be dangerous to stake too much on this.

The south-western and south-eastern turret roofs rose high above the main roof and were accessed from the latter by steep flights of steps rising against the internal face of the southern parapet (see Fig. 10). The top level of the south-western turret has been reduced too low to leave traces of its roof. However, sufficient of the south-eastern turret remains to show the junction between wall and parapet and thus give the level of the underside of the joists to its roof.

The Parapets

With the exception of a few surviving inner blocks of ashlar facings to the parapets of the north wall and south-eastern turret, only over the southern wall does sufficient survive to say anything constructive about the design of the parapets. It is clear from these remains that the merlons were pierced by arrow-slit windows which, in every surviving detail, are identical to those in the north wall and the turrets of the gatehouse. In this respect the battlements are very similar to the known details from the sister castle of Caerphilly, and from the slightly later Edwardian Welsh castles. Another close parallel which could be cited is Marten’s Tower of 1270-1300 at Chepstow Castle, Gwent (formerly Monmouthshire) (Perks 1967). However, in one important respect Tonbridge complies with none of these, for it also incorporates either outwardly-splaying or canted openings of unknown width and height, but which extend down to the level of the roof’s weathering course. The purpose of these openings is unknown (unless associated with the use of canon), as indeed is their date, though they are certainly not recent.

The other point which can be noted is that at each end of the south wall the battlements must have stepped up towards the raised turrets in order to give sufficient protection to users of the turret steps. It should also be remembered that the portcullises protecting the doorways leading off the curtain wall walks to the gatehouse were operated from the roof. Therefore, against the northern face of the turrets the parapets must have been of sufficient height to protect the raised portcullises and their mechanisms.

Perhaps some idea of the height of a standard section of battlement and of the design of its copings can be gained by the impression of the curtain wall battlements at the point where these meet the western wall of the gatehouse. At this point the extant coping rises 2.49m above door-cill level (which was probably positioned just above the level of the wall walk) and is symmetrically battered on the internal and external faces, rising to a centrally-placed roll-moulded crest.

All the evidence cited above has been brought together in the reconstruction of the south parapet wall (Fig. 10). Despite the evidence, it should, none­theless, be regarded as a ‘best guess’ reconstruction. This general design can be replicated (on paper) around the building to give a conjectural impression of the gatehouse’s appearance during the medieval period – quite different to the arrangement shown by Wadmore. The reconstructions, with the external ground returned to its original levels, are shown in Fig. 2 and Fig. 14.

dating and significance

Opinions regarding the age of the gatehouse are somewhat divided, ranging in date from the early-13th century through to the opening years of the following century. King (1782, 272) considered it to belong to the early years of Henry III’s reign (i.e. 1216-40) whilst a century later Wadmore (1886, 26) dated it similarly to 1220-1240. More recently, the similarity of its design to those of the great gatehouses of Edward I’s Welsh castles has led some (but by no means all) scholars to conclude that it was modelled on these, and that a more realistic date would be about 1300 (Newman, 1969, 547-8; Platt, 1982, 98). The present guide book compromises by attributing it a construction date in ‘the second half of the 13th century’ though adds in the summary that ‘the great gatehouse built by 1258’ (Oliphant 1992, 14 and 32). This latter statement is, apparently, based upon the granting of a royal licence to Richard de Clare in 1259 to wall and embattle his town of Tonbridge and, more specifically, upon Wadmore’s observation that ‘it would have been clearly of little use to fortify the town without first properly providing for the exigencies of the castle’ (Wadmore 1886, 26). What this ignores is that the present gatehouse represents a rebuild of an earlier structure and need not be contemporary with the curtain wall.

Whilst King and Wadmore’s dates are now generally recognized as being too early, it has been convincingly demonstrated that – despite the similarity with the late 13th-century Edwardian gatehouses of Wales – a 1300 date is too late. Renn has pointed out the very close similarities between Tonbridge Gatehouse and that of the main gate of the mightiest of de Clare’s castles, Caerphilly in Glamorgan, much of which was built 1268-1271, though the work continued beyond that date (Johns, 1978; Renn, 1981, 93-103; Renn, 1989, 18). Regarding the date of the Caerphilly gate, Renn now considers that this may represent ‘a later work, perhaps a change of plan brought on by either the burning of 1271 or by the removal of the Welsh threat from the west in 1276’ (Renn, pers. comm. 1997). Both Johns and Renn have demonstrated that, rather than being the imitators of the innovations made by Edward in his Welsh castles, Caerphilly and Tonbridge are the role models upon which Edward’s gatehouses at Harlech (1283-89) and Beaumaris (1295-6) are based. To Tonbridge and Caerphilly Spurgeon adds another apparently very similar de Clare gatehouse at Llangynwyd Castle, Glamorgan. Llangynwyd was destroyed by Llywelyn ap Gruffudd in 1257 and rebuilding works are known to have been underway by 1262. The gatehouse is assumed to represent part of this rebuild (RCAHM WALES 1991, 258-63; Spurgeon pers. comm., 1997].

Renn cites the threat from a local rival magnate to argue that Richard de Clare, Earl of Gloucester – father of the Gilbert who built Caerphilly – had already built the Tonbridge Gatehouse by his death in 1262 (Renn, 1989, 34). He envisages its construction as Richard’s rapid response to Roger de Leybourn’s strengthening of his fortifications ‘up the road’ in 1260. Although Renn would now modify his 1989 statement, there still seems to be no historical reason to build the Tonbridge gatehouse after 1271 when Leybourn died. Indeed, Renn adds that Tonbridge Castle was the major pledge of the 1260s between Clare and Henry III, who seems to have been playing off Clare against Leybourn (Renn, pers. comm., 1997). He further argues that from 1268 Gilbert, the son, would have been too preoccupied with protecting his Glamorgan estates against the claims of Llywelyn ap Gruffudd to embark upon such a project (Rees 1974, 8; Renn 1981, 93-103). It was this dispute which led Gilbert to build a castle on land at Caerphilly, a project which he commenced in April 1268 (Rees, 1974, 22). Whilst accepting the undoubted relationship between the design of the two gatehouses and their undeniable closeness in date, the argument forwarded for placing Tonbridge as the prototype for Caerphilly is circumstantial. It should not be taken for granted that it was beyond the means of a man of Gilbert de Clare’s calibre to have had more than one building project in progress at any one time, or that he would not have been able to initiate work at Tonbridge whilst involved elsewhere. Granted, one of the two gatehouses is almost certain to be the prototype of the other (unless Llangynwyd was the prototype of both) – as indeed both must have influenced the design for Edward’s Harlech and Beaumaris – but which is the earlier has still not been proven beyond doubt. Renn concedes

that Gilbert might have been building both gatehouses concurrently, but starting at different times in response to different threats [and adds] It is historically difficult to see Tonbridge [Gatehouse] after 1271 and Caerphilly [Gatehouse] before 1272, but not impossible ... both might be slightly later: Gilbert certainly went on building such gatehouses.’ (Renn, pers comm., 1997).

Architectural details, as with much military architecture, are too vague to forward the argument further. Most such details can be dated on stylistic evidence to no closer than the 13th or 14th century. The details of both the canopied fireplaces and the two two-light windows serving the Great Chamber are more diagnostic – both are typical of the mid/late 13th century. The window tracery in particularly is of a transitional form, midway between plain plate tracery and fully developed bar tracery. The design is all but identical to that used within the hall at Stokesay Castle, Shropshire, acquired by Lawrence of Ludlow in 1281 and completed by him in 1291. But this does no more confirm the dates suggest by Caerphilly.

A more rewarding method of resolving the problem might be to compare the differences between the two gatehouses. These might indicate which of the two is of the more advanced design, and thus the more recent. Some of the differences were undoubtedly influenced by circumstances, for Caerphilly was a new building constructed, as far as is known, upon a virgin site, whereas the Tonbridge example replaces an earlier structure and had to take into account existing features such as earlier ramparts and ditches of considerable proportions. It is surely this factor which explains the battered masonry spurs to the basement walls at Tonbridge, as well as the level of the entrances to the gatehouse off the wall walks of the curtain wall. Other variations reflect the different needs to which the structures were to be put – the Great Chamber at Caerphilly is more lofty than that at Tonbridge and as a result small mural side chambers could be incorporated at Mezzanine level in order to work the portcullises which protected the doorways leading off the wall walks. This in turn had a knock-on effect regarding the design details of the doorways and access stairs at the head of the corner turrets.

Yet other apparent variations could be the result of inaccurate recon-struction/restoration, for the front of Caerphilly gatehouse fell in antiquity and was entirely reconstructed in 1931-33. Rees states that much of the gate had fallen in large lumps and included one whole window. All was apparently carefully recorded in its fallen state, broken up and re-fixed back in its correct location (Rees 1974, 134-36). However, Robinson and Spurgeon have more recently ‘become convinced that details in the 1931-33 [reconstruction] can be questioned. The turrets are definitely wrong, the roof line is questionable, and fireplaces known to have been in the two first-floor rooms are simply left out’ (Robinson, pers comm., 1997). Unlike Tonbridge, Caerphilly in its present-day form does not incorporate the high front arches over the main entrance which give such an impressive appearance, but can it be certain that it never did have them? Robinson can find no evidence to suggest that it did, but in both designs the ground- and second-floor plans are identical, only at first-floor level is there any variation.

In addition to the above, there are features which genuinely indicate design ‘improvements’. The front faces of the garderobe offsets at Caer-philly, for instance, are set square, whereas those at Tonbridge are splayed so as not to present vulnerable corners to attack. Likewise, all the windows in the front elevation at Tonbridge take the form of narrow arrow slits, whereas at least one of those at Caerphilly is known to have been a narrow lancet. Even more significantly, at Tonbridge the two doorways which lead off the main entrance passage to give access to the gatehouse are each protected by their own portcullis, whereas at Caerphilly they are not. In these three respects Tonbridge is arguably the more advanced and ought to be the more recent in date. However, there is one advanced feature which is found at Caerphilly but not at Tonbridge. Over the rear entrance at Caerphilly is a chute contrived so as to exit water and the like onto assailants from the chamber above. This was in addition to the murder holes in the soffit of the arch in front of the portcullis. It should be conceded that Renn is able to counter these ‘improvements’ at Tonbridge by pointing out that the Caerphilly arrangement might represent a ‘softening’ of the Tonbridge design. He points out that the Caerphilly gatehouse stood back from the moat and was immediately protected by a middle gatehouse, with still further defences further out. Sharp corners and lancets were therefore less vulnerable there. The portcullises to the side doors may have proved inconvenient to use, breaking up the room space above. They are not to be found in the late Edwardian gatehouses, though they do exist at St Briavels, 1292-5 (Renn, pers comm., 1997).

So, as is often the case, the available historical and architectural evidence is not conclusive as to which of the two gatehouses preceded the other. Simmons (1998) has argued that the discussion can best be progressed by considering the sculptured heads within the Great Chamber and perhaps they are indeed the strongest evidence to support Renn’s hypothesis that Tonbridge was built by Richard immediately prior to his death in 1262. If this is so, then the gatehouse already existed when the castle fell to the king in 1264 – an irony given the gatehouse’s strength and careful design. On the other hand, it is very tempting to surmise that it was the old gatehouse at Tonbridge which failed in 1264 and thus prompted the building of the improved one.

Whichever proves to be the earlier – Caerphilly or Tonbridge – what is beyond dispute is that the two must be very similar in date. Whether Tonbridge is early enough to have been admired by the future Edward I in 1274 when he stayed with Gilbert de Clare en route to his coronation must, for the moment, remain unproven, though the likelihood must be that it then existed. Regardless of date, Tonbridge is undoubtedly one of the best preserved of the new breed of ‘keep-gatehouses’ developed during the middle years of the 13th century to strengthen the hitherto weak link in castle defences, at the same time providing a fitting lodging and power base for the Constable. The new breed of gatehouse replaced the keep as the strong point of castle design, easily capable of being defended in isolation from the remainder of the castle (Platt 1982).

The second-floor Great Chamber at Tonbridge is fitting tribute to the importance placed on these buildings. The equivalent room in Caerphilly is probably identifiable with the ‘Constable’s hall’ mentioned in an account of 1373-74 (Johns, 1978) whilst that at Tonbridge in all probabilities represented the ‘Council Chamber’ in which on 27 August 1297 Prince Edward, as the king’s lieutenant in England, delivered the Great Seal of England to the King’s Chancellor in the presence of Lord Reginald de Grey, Alan Plukenet, Guy Ferre, and Guncelin de Badlesmere (Wadmore 1886, 32).

A final note. Spurgeon has tentatively suggested that Aberystwyth, Edward I’s first ‘keep gatehouse’ of the Tonbridge/Caerphilly type, was commenced in 1277 by the English master-mason, Henry of Hereford. In correspondence with the authors, he further tentatively poses the interesting though that Henry of Hereford may have been engaged earlier by Gilbert. Whoever the master-mason at Tonbridge was, he was clearly a man at the pinnacle of his profession.

acknowledgements

The authors wish to extend our warmest thanks to Peter Kendall of English Heritage, the Staff of Tonbridge and Malling Borough Council, and in particularly Julie Beilby, Sheila Kostyrka, Roger Pattemore, Peter Powley, Keith Price, and the Ladies of the Tourist Information Centre for all their kind help throughout the project. They also much appreciate the help given by the Staff of the Tonbridge Public Library; Mr Sydney Simmons (the former Borough Surveyor) and Pat Hopcroft of Tonbridge Historical Society for their advice, comments and help.

During the project a lively debate developed regarding the relative dating of the sister gatehouses at Tonbridge and Caerphilly. In this regard the authors would particularly like to acknowledge the great assistance given by Derek Renn, Jack Spurgeon, Dr David M. Robinson (CADW) and Dylan Roberts (RCHM (Wales)).

bibliography

Almack, R. (ed.), 1854, ‘Papers Relating to Proceedings in the County of Kent ad 1642 - ad 1646’, Camden Society.

Archaeology South East (ASE) Project Ref. 546: Report on Selective Archaeo-logical Recording at Tonbridge Castle Gatehouse, Tonbridge, Kent.

ASE Project Ref. 1113: Report on an archaeological watching brief undertaken during restoration works undertaken between September 1999 and March 2000 at Tonbridge Castle gatehouse, Tonbridge, Kent.

ASE Project Ref. 1499: A revised record and interpretation of the roofs and parapet walls at Tonbridge Castle gatehouse incorporating additional details recorded during building works carried out in 2002/3.

[Copies of all three reports are available for inspection at the National Monuments Record in Swindon, at Tonbridge Public Library and at the Kent History and Library Centre in Maidstone.]

Hasted, E, 1798, The History and Topographical Survey of the County of Kent, 2nd Edition, Vol. 5.

Hemp, W.J., 1936, Beaumaris Castle, Anglesea, Wales (HMSO).

Hewett, C.A., 1985, English Cathedral and Monastic Carpentry (Chichester).

Johns, C.N., 1978, Caerphilly Castle, Official Guide Book (London).

King, E., 1782, ‘Sequel to the observations on ancient Castles’, Archaeologia, vi, 231-375.

Newman, J., 1969, The Buildings of England – West Kent and the Weald.

Oliphant, J., 1992, Tonbridge Castle (Tonbridge).

Perks, J.C., 1967, Chepstow Castle, Official Guide Book (London).

Platt, C., 1982, The Castle in Medieval England and Wales.

Radford, C.A. Ralegh, 1958, Goodrich Castle, Hereford (HMSO 1958).

RCAHM WALES, 1991, Inventory of Ancient Monuments in Glamorgan, Vol. III, (1a) No. MR2, 258-263 (HMSO).

Rees, W., 1972, Caerphilly Castle and its place in the Annals of Glamorgan (Caerphilly 1974).

Renn, D.F., 1981, ‘Tonbridge and some other Gatehouses’, in A. Detsicas (ed.), Collectanea Historica: Essays in memory of Stuart Rigold (KAS, Maidstone), 93-103.

Renn, D.F., 1989, Caerphilly Castle (Cardiff).

Simmons, S., 1996, ‘Tonbridge Castle: Further Observations on an Ancient Castle’, Archaeologia Cantiana, cxvi, 101-143.

Simmons, S., 1998, ‘The Lords and Ladies of Tonbridge Castle’, Archaeologia Cantiana, cxviii, 45-61.

Stanbridge, A.W., 1966, Preservation of Tonbridge Castle 1954-1966 (un­publ.), Tonbridge Public Library, ref. TU12/Z27.

Wadmore, J.F., 1886, ‘Tonbridge Castle and its Lords’; Archaeologia Cantiana, xvi, 12-57.

West, J., 1981, ‘Acton Burnell Castle, Shropshire’, in A. Detsicas (ed.), Collectanea Historica: Essays in memory of Stuart Rigold (KAS, Maidstone), 85-92.

Wood, M., 1965, The English Medieval House (London).

Fig. 1 Outline Plan of the Castle in 1866.

Fig. 2 Reconstructed north and south elevations. Conjectural details shown in chain-dotted outline.

Fig. 3 Reconstruction Plans of the Gatehouse – Ground and First Floors.

Fig. 4 Reconstruction Plan of the Gatehouse – Second Floor.

Fig. 5 Design of typical single-light window.

Fig. 6 Design of one of the two-light windows serving the Great Chamber.

Fig. 7 Entrance passage showing arrangement of doors.

Fig. 8 Reconstructed layout of the first-floor joists.

Fig. 9 Reconstructed sections through the gatehouse. Conjectural details shown in chain-dotted outline.

Fig. 10 Reconstructed section through the gatehouse. Conjectural details shown in chain-dotted outline.

Fig. 11 The fireplace serving the Great Chamber, as built.

Fig.12 Reconstruction of fireplace in eastern first-floor chamber.

Fig. 13 Reconstructed plans of the garderobes.

Fig. 14 Reconstructed west and east elevations. Conjectural details shown in chain-dotted outline.