William Marshal, Pembroke Castle and Angevin design

William Marshal, Pembroke Castle and Angevin design

Neil Ludlow Pembroke Castle keep 1201-1207. View from the southeast. © Neil Guy Neil © southeast. the from View 1201-1207. keep Castle Pembroke

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16213 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

William Marshal, Pembroke Castle Blois), built very early in the twelfth cen- and Angevin design tury by an ally of the Angevin counts 1 This paper argues that the castles of Wil- (Mesqui 2013, 185 et al.). The form was liam Marshal the elder owe their influences adopted by the counts of Blois during the almost entirely to the works of the English twelfth century, and in Angevin-held Tou- Crown – as might be expected given his raine, Maine and Normandy. close association with the Angevin kings. Of the 52 cylindrical great towers in The dominant narrative, suggesting influ- France that were begun before c . 1203 ence from the court of the French king (Table 1; Fig. 1),2 nearly half (23) are Philip II Augustus, fits in with neither form, thought to have been built by the Angevin function, detail and relative chronology, kings or their vassals. A number were nor with the political circumstances of the built by King Henry II. Henry set about period. Philippienne influence only be- ‘improving and repairing’ his French cas- comes apparent in the work of his sons, tles in 1161 (Howlett 1889, 209-10), when after 1219. Attention will accordingly be the donjons at Château-sur-Epte and drawn to the cylindrical donjons built in Neaufles-Saint-Martin (both Eure, near France by the Angevins, and to the distinc- Gisors in Normandy), may have been tion between the ‘donjon’ and other forms commenced (Corvisier 1998(2), 147, of dominant tower. 491); the Norman exchequer accounts suggest that both donjons were complete By 1201, when work at Pembroke Castle 3 began, Marshal had spent over fifty years in by 1180 (Stapleton 1840, 72, 110). Both and around castles on both sides of the are perfectly cylindrical, without any pro- Channel, as squire, sportsman, soldier and jections (Fig. 2), and like most Angevin courtier. He had plenty of opportunity to donjons, are unvaulted; Château-sur-Epte follow their development in the Angevin shows external offsets which, as in the donjon at Pembroke, do not always corre- dominions and beyond. His experience of 4 the French royal domain was however spond to the floor levels. largely confined to the pre-Philippienne Henry II’s cylindrical donjons at Neuf- period. He is recorded in the Île-de-France marché, Seine-Maritime in Normandy just four times after Philip’s succession in (Baudry 2002, 57; Corvisier 1998(2), 1180, and each visit was short, and circum- 492-5), and Bourg-le-Roi, Sarthe in scribed. Maine (Corvisier 1998(2), 64-71), may 5 There are several published accounts of also have been begun in the 1160s. A Pembroke Castle (e.g. King 1977; King fifth, at Châtillon-sur-Indre (Indre), in 1978; Renn 1968), and descriptions herein Touraine, is possibly from the 1180s will be brief. See Appendix 1 for a discus- (Corvisier 1998(2), 201; Deyres 1984, sion of the function of its donjon. 364, 374-6); the rest cannot be closely dated (see Table 1). The cylindrical donjons of the Angevin kings King Richard I continued the tradition at the important ducal castle of Bonneville- Excavation suggests that the origins of the sur-Touques, Calvados in Normandy free-standing cylindrical donjon lie in the (Boüard 1966, 355; Mesqui 2013, 24 fig., Loire region, at Fréteval (Loir-et-Cher, 167, 215-6; Fig. 3).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16214 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 1 – Northern France in c. 1203 showing circular great towers.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16215 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 2: Two cylindrical donjons of King Henry II: (L-R) Château-sur-Epte and Neaufles (both Eure, Normandy).

Fig. 3: Plan of Bonneville-sur-Touques, Calvados (from Mesqui 2013). Cylindrical donjon, 1190s (Richard I); towers and probably curtain, 1199-1204 (King John).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16216 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Now truncated and used as a private dwell- tower at Châteaudun (Eure-et-Loir), built in ing, the Bonneville donjon was under con- the 1170s-80s (Erlande-Brandenburg 1970, struction by 1195.6 It was again unvaulted 142). The French Crown built a small num- (Boüard 1966, 355) and, like Pembroke, it ber of circular donjons of a rather different was originally free-standing, on level nature (see below), while the form had ap- ground, towards the edge of a pre-existing peared further afield by the 1190s, in eastern ringwork. The towered enceinte is an addi- France and the Holy Roman Empire (Mes- tion of King John’s reign (Boüard 1966, qui 2013, 185). 7 371; Mesqui 1997, 456; see below), butt- Donjons and tours-mixtes ing obliquely against the base of the donjon in a manner not unlike the secondary cur- The free-standing cylindrical donjon, then, tain at Pembroke. was firmly rooted in the works of the An- gevin kings. But it was late to arrive in the Richard also built the cylindrical donjon British Isles and, when it did, does not ap- at Longchamps (Eure, Normandy), where pear to have been employed by the Crown, £480 was spent in 1198 (Corvisier 1998(2), under whom engaged, peripheral towers 393-6; Stapleton 1844, 300, 307, 310, were favoured.12 There is a fundamental 350). He is also thought to have built the distinction between these towers and the round tower on the motte at Le Muret, free-standing donjon: only in the latter, Cléry (also Eure), which accounted for which was neither defensive (thus normally £420 in 1198 (Pitte 2002, 168 and n. 19; unlooped) nor, in many cases, residential, Stapleton 1844, 310), while the similar was it possible for access to the space within tower at Bonsmoulins (Orne, Normandy) to be properly regulated (see Dixon 2002, may represent the £419 spent there in 9-11; Marshall 2002b, 28; Mesqui 1993, 8 1194-5 (Stapleton 1840, 245-6). It is sug- 252) – all would see it, but few would ever gested here that the king’s extensive cam- enter (Marshall 2002a, 142). paign at Pont-de-l’Arche (Eure) included a further free-standing cylindrical donjon The engaged tower, on the other hand, (see below and Fig. 18 ).9 In Poitou, the might be used as a residential chamber- free-standing round tower at Montreuil- tower but could also be equipped for de- Bonnin (Vienne) is also thought to be Rich- fence (the ‘tour-mixte’ of French authors, ard’s work (Baudry 2002, 57-9 Fig. 4).10 see e.g. Mesqui 2013, 152-75), becoming a more public space and allowing for multi- Cylindrical donjons were similarly built ple functions.13 For instance, the cylindrical by influential vassals of the Angevin kings, Tour du Moulin at Angevin Chinon (Indre- sometimes under their advocacy like those et-Loire), now thought to have been added at Conches-en-Ouche (Eure), probably by King John 1200-04, though possibly a built 1180-1200 (Corvisier 1998(2), 255; little earlier (Dufaÿ 2011, 88-91, 97-100; Landon 1935, 96; Mesqui 1997, 457; Pitte Mesqui 2013, 167, 216), was a tour-mixte 2002, 166), and Brosse (Indre in Poitou), of this kind. It is engaged with the curtain from 1194-1240 (Baudry 2002, 50 and fig., as a flanking corner-tower, rather than free- 11 52, 57-9, 66-7; Fig. 14). standing, and is pierced for archery, but the The donjons of Blois belong to a distinct six large windows in the upper floor also tradition, sometimes featuring vaulted interi- suggest a domestic function, possibly as a ors and well-represented by the magnificent ‘prospect chamber’.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16217 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 4: The cylindrical donjon at Montreuil-Bonnin (Vienne), thought to have been built by Richard I.

Fig. 5: Plan of Château Gaillard, Eure (from Brown 1976).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16218 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

King John also built towers of this kind in 1170-5 (Baudry 2002, 60-1; Mesqui 1993, the British Isles. The large round tower at 330); its overall visual impression may Dublin Castle (the ‘Record Tower’), is sim- reference the rectangular donjons of his ilarly engaged as a corner-tower which, as ancestors, as Henry II’s Dover is also at Chinon, appears to have been a tour- thought to have done (see Brindle 2015, 12; mixte with arrowloops (now altered; Man- Hulme 2014, 206).16 Jean Mesqui felt that ning 2003, 72, 84-7; Fig. 8). Also in Ire- the machicolation was yet another aspect of land, the circular ‘Reginald’s Tower’ at the visual ostentation that was the king’s Waterford, from c. 1207-15, is similarly paramount concern at Château Gaillard thought to have been built to flank King (Mesqui 1993, 334). As at Chinon, the John’s castle there, becoming part of the tower is well-furnished with windows town defences on the demise of the castle yielding extensive views. later in the thirteenth century (Murtagh The octagonal tower at Gisors (Eure) is, 2017, 163-5). The term ‘dominant tower’ next to Fréteval, the earliest non-rectilin- may be preferable for these structures, to ear donjon that can be dated with reason- which access appears to have been less able confidence, belonging to the reign of 14 restricted than it was to the donjon. Dom- Henry I c.1109-35, and perhaps begun in inant towers of this form, with both defen- 1123 (Corvisier 1998(2), 328; Howlett sive and domestic functions, were adopted 1889, 196). The polygonal form was not at a number of baronial castles in the British widely adopted in twelfth-century France Isles during the thirteenth century. Among (Mesqui 2013, 187; see Table 1), but the earliest may be the flanking round tow- found favour under the Angevins in the er at Barnard Castle (Co. Durham), which British Isles e.g. Henry II’s Chilham, Or- is like Chinon in many respects (Marshall ford and Tickhill, and King John’s Odi- 2002b, 27), but articulates with the cham- ham and Athlone (Goodall 2011, 127-30, ber-block to which it is attached, while also 167; Hislop 2016, 117).17 The donjon built 15 featuring a battery of arrowloops. These in the 1170s-80s at Conisbrough (Yorks.), may be contrasted with the narrow open- by Henry II’s half-brother Hamelin of ings in the body of Pembroke’s donjon, Anjou, is fundamentally cylindrical and, discussed in Appendix 1, which are lights as built, was free-standing (see Brindle rather than arrowloops. and Sadraei 2015, 6, 29, and plan);18 it has Though engaged and peripheral, it is likely a similar relationship with the enclosure that the large semi-round tower built at Rich- periphery, and secondary curtain, to that ard I’s Château Gaillard (Eure), in 1196-8, seen at Bonneville and Pembroke. Its was not intended for defence (Fig. 5). It shape – a cylinder with six, solid polygo- overlooks a steep cliff rather than any practi- nal buttresses – was also employed by cal line of assault while, unlike the mural Hamelin in Normandy, in the free-stand- towers, the body of the tower is not pierced ing donjon at his castle in Mortemer, for archery; functionally, it can be regarded Seine-Maritime (Brindle and Sadraei as a donjon (see Marshall 2002b, 32). Fac- 2015, 11; Mesqui 1997, 458; Renn 1961, ing the interior, the prominent beak shows 132); their relative dating is uncertain buttressed machicolation like that employed (Goodall 2011, 151; Guy 2013a, 203).19 by Richard’s father in the double-donjon at Whilst it will be shown below that Mar- Niort (Deux-Sèvres in Poitou), in around shal’s donjon at Pembroke has its models

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16219 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design in Angevin Normandy, it also has affinities at several levels, have no contemporary with Conisbrough. Otherwise it has no parallels in Britain (Fig. 7). Not apparently close parallels, either in form or detail, in ceremonial, they perhaps served regional any contemporary castle-work in the Brit- officials in a residential capacity, while ish Isles where it appears to be the earliest being actively defensive (Mesqui 2013, tower of its kind. 162-8).22 In this they differed fundamen- tally from the cylindrical donjon while A rather different trend is apparent in the their double entries, opening to the interior Île-de-France during the twelfth century and exterior of the castle, are also un- where, while exciting variations in form known in Britain. were being explored, the plain cylinder was not greatly favoured by the French Crown The Tour Philippienne was widely used in until the late 1180s (Mesqui 2013, 185; the Capetian royal domain after 1202 Table 1). A polygonal shape was chosen at (Mesqui 2013, 162-8; also see Knight 2008, Magny-les-Hameaux, while circular Châ- 65). Doubtless intentionally, it was symbolic teaufort, like the similar baronial tower at of King Philip’s dominion and four more Maurepas, shows large square turrets; all were built in Normandy after its capture in may belong to the 1130s-50s (all Yvelines; mid-1204 (Falaise, Lillebonne, Rouen and Corviser 1998(2), 165, 402, 418). The don- Verneuil-sur-Avre) – that is, after the Pem- jon at royal Compiègne (Oise) is an unbro- broke donjon had been commenced. Con- ken, unvaulted cylinder (Fig. 6), for which struction continued into the 1220s with a similar date has been argued (Corvisier Dourdan (Essone), the last of Philip’s reign 2002, 44), though it could be 1150s-70s (Mesqui 1997, 153; Fig. 15). And following (Mesqui 2013, 185).20 royal precedent, variants of the form became adopted in other regions of France during King Philip II introduced the cylindrical the early thirteenth century (Mesqui 2013, tour-mixte into the Capetian landscape. 168, 185). While the first of Philip’s towers The first of these ‘Tours Philippiennes’ may represent the earliest loopholed tours- appears to have been at Bourges (Cher, in mixtes, most of them post-date the similar Berry), erected in the late 1180s but now tower at Chinon and their influence on King gone (Mesqui 2011, 311 and n. 54; Mesqui John’s designs is uncertain.23 2018, 139), closely followed by the Lou- vre, Paris, which can be dated 1190-1202 Pembroke was Marshal’s only donjon. (Fleury and Kruta 2000, 2, 50-1). These His other dominant towers – at Kilkenny towers were isolated;21 subsequent Tours in Leinster and Usk, Mon. – are peripheral, Philippiennes, beginning in the mid-late engaged and defensive tours-mixtes like 1190s at Gisors and Vernon, Eure (Mesqui those at Angevin Chinon, Dublin and Wa- 2011, 311), were engaged with the curtain, terford; they are, however, not isolated by furnished with arrowloops and normally ditches like Capetian Tours Philippiennes. occupied corners, surrounded by a ditch Marshal is considered to have begun work (Mesqui 2018, 129-39). They were of a at Kilkenny Castle, his seigneurial caput in stark, functional design not generally seen Leinster, in 1207 (Bradley and Murtagh in either the donjons or tours-mixtes of the 2017, 222; Tietzsch-Tyler 2017, 185), and British Isles; large window openings, for it seems to have been fairly well-advanced instance, were absent while their rib-vaults, by 1210 when King John visited (Holden

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16220 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

ABOVE: Fig. 6 – Interior of the cylindrical donjon at Compiègne, Oise. BELOW: Fig. 7 – The Tour Philippienne at Vernon (Eure), built by Philip Augustus 1196- 1200 (from Mesqui 2011 and 2018). cf. Fig. 15.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16221 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design et al. 2004, 193).24 The large, engaged Dundrum, Rochester etc.; see Appendix 1). round tower at the southern corner (the It is therefore likely that preliminary ‘White Tower’; Fig. 8) appears primarily groundwork, recruitment and other prepa- administrative, and ‘essentially a public ration commenced immediately after his space’ (Tietzsch-Tyler 2017, 186) which, visit. The project’s importance is confirmed like Dublin’s Record Tower, is pierced for by the appointment to the stewardship of archery (Murtagh 2017, 129-31, 171 fig).25 Pembroke, in 1202, of one of Marshal’s Influence from Chinon may be implicit – leading household knights, Nicholas Ave- Marshal was at the French castle through- nel. And the post was no sinecure – Avenel out August 1201 (Hardy 1835a, 3; Hardy was recorded at Pembroke in July 1202 1844b, 19-20) – but the Kilkenny tower (Brewer 1863, 227; Davies 1946, 322, was under construction at more-or-less the 327). The donjon may have been more-or- same time as the Record Tower, begun less complete prior to Marshal’s departure some time after 1204 (Sweetman 1875, for Leinster in spring 1207: following Der- 35); their immensely thick walls are unique ek Renn’s formula, the six seasons of work to these two towers and reveal a close would account for 18-22 metres (60-72 relationship between them, if no clear indi- feet) of its total height of 24 metres (78½ cation of which was the earliest.26 The feet; see Renn 1973, 25-6), although there cylindrical ‘Garrison Tower’ at Usk, which may be some evidence of a change of plan occupies an angle rather than a corner, may at parapet level (see Appendix I), possibly be somewhat later (see below and Fig. 9); after Marshal’s return in 1213. it too is pierced for archery (Knight 2008, The inner curtain overlies the base of the 61 fig., 63 figs.). Unlike Tours Philippi- donjon, at least in its restored form which, ennes, these towers appear to have been judging from the surviving stump against neither self-contained residences nor, nec- the Dungeon Tower, follows the original essarily, chamber-towers, and none lie in line (Figs. 16 and 35 ). The curtain was association with the domestic buildings.27 therefore secondary but, like the screen- Marshal at Pembroke wall closing off the Wogan cavern beneath the inner ward, was probably commenced A start-date in spring 1201 is now general- before 1207 (see below; Day and Ludlow ly agreed for Pembroke’s donjon, follow- 2016, 66). All evidence suggests that, un- ing Marshal’s visit in late autumn 1200, der the Marshals, Pembroke Castle was with his wife, on his way to Leinster limited to the present inner ward, the outer (Crouch 2016, 10-24, 137; Goodall 2011, ward being an entirely new addition of the 162). The lordship of Pembroke had re- mid-thirteenth century (as argued in Day cently been released to him by King John, and Ludlow 2016, 67-8, 88-9, 93-4). coinciding with his first real opportunity to tour his western estates. The erection of the Marshal in France donjon was almost certainly celebratory We will see that in matters of detail, as and commemorative: he was at last able to well as overall form, Marshal work fol- mark his marriage, inheritance and eleva- lows Angevin precedent. Nevertheless, in- tion to the earldom of Pembroke – all of fluence from Philippienne designs is which were hard-won – at its titular caput generally claimed. But the relative chro- (cf. the celebratory donjons at Hedingham, nology poses insurmountable problems:

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16222 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 8 ABOVE: plan of King John’s Dublin Castle in its final phase (from Manning 2003); BELOW: Ben Murtagh’s reconstructed plan of Marshal’s Kilkenny Castle (from Murtagh 2017).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16223 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

ABOVE: Fig. 9 – Plan of Usk Castle, Mon. (from Knight 1977).

BELOW: Fig. 10 – Plans of the Angevin Tour du Diable at Gisors, Eure (from Mesqui 2013).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16224 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design the only Capetian castles that can possibly 1935, 113-42; Round 1899, 195, 379, 473, have influenced Marshal were built before 498), and was at Bonneville-sur-Touques Philippienne designs took hold.28 For in- in January 1200 (Hardy 1837, 32), at least stance, he is last recorded in Paris in 1188 – a year before work commenced at Pem- two years before the Louvre was conceived broke. England and France were again at – when he appears to have been an envoy to war 1202-4, and tensions had been mount- King Philip on behalf of Henry II (Holden ing since spring 1201 when King John et al. 2002, 423).29 began strengthening his Normandy castles (Hardy 1835b, 35). While Marshal Instead, in close company with the An- achieved a détente with King Philip after gevin kings from 1186 onwards (Crouch 1204-5, in which – exceptionally – he was 2016, 68-9), Marshal will doubtless have permitted to keep his Norman possessions, seen a number of Angevin cylindrical don- the relationship could not be called close jons. L’Histoire de Guillaume le Maréchal, and, apart from a possible, very brief visit the biographical poem composed in his in 1216, Marshal was never to see these honour in the mid-1220s, implies that Mar- lands again (Power 2003, 211-13).30 shal may have been with King Henry II during the negotiations at Châtillon-sur-In- In addition to his experience of these dre in October 1188 (Holden et al. 2002, Angevin cylindrical donjons, Marshal 399-400; Holden et al. 2006, 109; also see spent at least one night at Châteaudun Vincent 2000, 5-6); he had certainly trav- while campaigning in Blois in 1188, eled extensively in Touraine during the (Holden et al. 2002, 409; Holden et al. 1180s. Henry II’s donjons at Neaufles and 2006, 110); if the tower was not complete Château-sur-Epte were lost to King Philip by this time, it will have been well under in 1193 (Powicke 1913, 161-2; Stapleton way. Influence from Châteaudun, at Pem- 1840, cxlii; Stapleton 1844, cxxviii), but broke, has been suggested by a number of Marshal may have visited them, along with authors (e.g. Avent 2006, 89; Goodall Neufmarché, between 1186 and 1190 when 2011, 162; King 1977, 164). Like the don- he is recorded in eastern Normandy with jon at Bonneville it occupies level ground, kings Henry and Richard (Holden et al. and shows certain other resemblances to 2002, 371-2, 375-95; Landon 1935, 26-31). Pembroke – as well as important differ- He was moreover at Fréteval in July 1194 ences, discussed below. (Holden et al. 2004, 29-35), which was then Marshal was only to visit France three under Angevin control. times after 1203, primarily to meet Philip Warfare between the English and French Augustus on King John’s behalf. He crowns – In which Marshal was active with- crossed to Normandy in late April-May in the Angevin lands – dominated the later 1204 (Hardy 1835a, 40); journeying from 1190s, ruling out any peaceful contact with Rouen, still in English hands, he met Phil- the French court (or recruitment of Cape- ip firstly at Le Bec-Hellouin, Eure (Hold- tian masons), but yielding many opportuni- en et al. 2004, 145), and then at Lisieux, ties to witness the building of Richard I’s Calvados (Crouch 2015, 163-4; Stapleton circular donjons. Marshal was frequently at 1844, cxxxviii). In April 1205, Marshal Château Gaillard from 1196 onwards (Cal. met Philip at Compiègne, and then Anet Charter Rolls 1226-57, 185; Hardy 1837, (Eure-et-Loir) near the Norman/French 15-16, 22, 28-30, 33, 64, 68-70, 96; Landon border (Crouch 2015, 11, 163-4; Holden et

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16225 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design al. 2006, 148). No Philippienne castle ex- The Angevin Tour du Diable at Gisors, isted at these venues; the earlier donjon at from the 1180s-early 1190s, shows some Compiègne has been discussed above. loops with niches; others however have Marshal’s whereabouts during his third simple triangular embrasures, with full- visit, in March-April 1216, are unknown centred rear-arches, like those in Marshal and our only clue is from Philip’s itinerary, work (Mesqui 1993, 254, 265 fig.; Mesqui which locates the king at Compiègne, Pont- 2013, 263 fig.; Fig. 10 ). Loops without de-l’Arche, Melun (Seine-et-Marne) and niches were also used in the apices of the Paris during these months (Samaran and beaked flanking towers at Loches (Indre- Nortier 1979, 15-32). The first two were et-Loire), in Touraine, which are now at- peripheral to the royal domain and thus a tributed to Richard I (Héliot and Deyres more likely choice of venue than Paris or 1987, 48-53, 75), and in the eastern outer Melun; the castle at Pont-de-l’Arche is dis- gatehouse (Porte Saint-Nicholas) at Fal- cussed below. Either way, work on most of aise Castle, Calvados (though heavily re- Marshal’s castles in the British Isles was, stored), which recent investigation has by 1216, well-advanced. shown to be Angevin work of the late twelfth century (Fichet de Clairfontaine et Loops, offsets and batters al. 2016, 245). Also reflecting Angevin precedent, though Triangular loops are also seen in King often attributed to Philippienne influence, John’s southern towers at Arques-la-Ba- is the Marshals’ use of deep basal batters, taille, Seine-Maritime (Langeuin 2002, 356 external offsets, and narrow, simply- fig., 362; Fig. 13), almost certainly built splayed triangular loops without internal 1201-03 when over £1500 was spent on niches or casemates. While such loops do works there (Hardy 1835a, 15; Hardy predominate in Philip’s work, we have seen 1835b, 69; Langeuin 2002, 362-5; Packard that this belongs mainly to the period be- 1927, 67-8), as well as in the circular flank- tween 1202 and 1223. The vast majority of ing towers at Bonneville-sur-Touques Philippienne loops have high sills in the which are thought to have been added by external half, and lintelled embrasures King John before 1204 (Boüard 1966, 371; (Mesqui 1993, 263 and fig., 281); the for- Corvisier 1998(2), 62-3; Mesqui 1997, 456; mer are absent from Marshal work, while Fig. 3). The narrow slits that now survive at the latter contrast with their preference for Richard I’s Château Gaillard are mainly arches. The arrowloop was a comparatively lights, but a triangular-splayed opening, new innovation which was still evolving flanking the postern at the northeast corner between 1180 and 1210 and, while the of the inner ward, may be a loop.31 niched embrasure may have been preferred by the Angevin kings (Mesqui 1993, 253- We have seen that William Marshal had 4), it was by no means universal, particular- extensive experience of Château Gaillard, ly in Normandy; as John Gillingham notes, influence from which is strongly apparent Angevin architecture embraced a variety of in his work (see below). Arques-la-Ba- forms (Gillingham 2001, 118). Moreover, taille was in his custody during 1202 (Har- dy 1835a, 9-11, 15, 22; Packard 1927, Marshal himself used niched embrasures at 32 Kilkenny Castle, 1207-13 (see Murtagh 67); he also revisited Bonneville twice, 2017, 160, 173). with King John, in autumn 1203 (Hardy

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16226 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

1837, 32, 111; Holden et al. 2004, 141).33 ry towers at Pembroke itself, at Younger Loops without niches also appear in the Marshal Cilgerran, Pembs. (Wiles 2014, Norman gatehouses at Domfront (Orne), 191 figs.), and at Kidwelly Castle, Carms., built in the years either side of 1200 (Mes- in work from the 1270s through to the qui 1997, 152-3 and fig.), and Bricquebec early fifteenth century (Kenyon 2007, 32 (Manche), from the end of the twelfth cen- fig., plan). tury (Corvisier 1998(2), 84-5); Marshal was probably with King Richard at Dom- Cruciform loops occur in Younger Mar- front at Christmas 1198 (Cal. Charter shal work at Cilgerran and Chepstow (up- Rolls 1257-1300, 173; Landon 1935, 141; per barbican), in Pembroke’s Dungeon Stapleton 1844, xliv), when work on the Tower, and the ‘towered keep’ at Carlow gate may have already begun (see Staple- which are also probably Younger Marshal ton 1844, lix).34 (see below). Otherwise, their only Mar- shal use is in the outer gatehouse at Loops of this kind were also employed Chepstow (Avent 2003, 62-3), which by the Crown in the British Isles, in the may similarly be post-1219 (see below). northwest gatehouse at Dover Castle, They are also seen in Angevin castles, Kent (the ‘Norfolk Towers’), at some appearing in Richard I’s Bell Tower at the point between the early 1180s and 1216 Tower of London (Impey and Parnell (Brindle 2015, 23; Chapman 2007, 106; 2011, 21-3; Knight 2018, 164) and in Hulme 2014, 220, 226; Fig. 11), and at King John’s outer gate at Kenilworth.36 Limerick Castle, Ireland, where they are However, it is not altogether certain that universal at all levels in King John’s orig- they are not later modifications, of earlier inal work of c .1211-12 (Tietzsch-Tyler loops, at both sites (Neil Guy, pers. 2013, 143 fig., 150 fig.; Hulme 2014, 220; comm.), which raises questions about Fig. 12). Triangular loops without niches their dating elsewhere. At any rate, they were also employed in an ‘Angevin’ mi- first appeared in France in the Angevin lieu by William, son of Hamelin of Anjou territories, around 1200, e.g. in the inter- (lord of Conisbrough), in the shell-keep val towers at Cluis (Indre) in Touraine towers at Lewes Castle (Sussex) shortly (Corvisier 1998(2), 233; Mesqui 1997, 35 after 1202 (Goodall 2011, 185; Fig. 11). 435), at Brosse (Mesqui 1997, 82; see Similar loops appear in other baronial below) and, slightly later, under Angevin work of the period, for example in the sponsorship, at Coudray-Salbart and gatehouse and Carrickfergus Tower, from Parthenay (both Deux-Sèvres), in Poitou 1199-1214, at Warkworth Castle in Nor- (Knight 2018, 164; Mesqui 1993, 268, thumberland (Goodall 2015, 13, 26, 35) 272). Moreover, they never really caught and in the north gate at Helmsley Castle, on in the French royal domain (Baudry Yorks., built by Robert de Roos, custodi- 2002, 64; Mesqui 1993, 291-2 and map). an of royal Bonneville-sur-Touques, in 1191-1227 (Kenyon 2017, 32-3 and plan). Basal oillets make an early appearance in They persisted into the late Middle Ages in Richard I’s work at Loches (semi-circu- west Wales, where a distinctive regional lar), and in Richard or John’s work at tradition developed in Marshal-dominated Montrichard (Loir-et-Cher) and Dom- areas (Day and Ludlow 2016, 68; see be- front, where they are triangular (Mesqui low), and are seen in mid-thirteenth-centu- 1993, 272, 284-5, 299; Mesqui 1997, 152-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16227 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 11 – Loops without niches at (L-R) Angevin Dover Castle, 1180s-1216; and Lewes Castle, Sussex, 1202+ (Dover from Brindle 2015).

Fig. 12 – Plan of Limerick Castle (from Tietzsch-Tyler 2013). Nb. the current rectilinear plan was not achieved until the reign of Henry III.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16228 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

3), a shape adopted at baronial Cluis and tectural detail suggests a similar date for Warkworth (see above) and elsewhere. Chepstow’s Middle Ward (see below). At Semi-circular oillets were also used at Bro- any rate, the rectangular oillet appears to sse (Mesqui 1997, 82). The circular oillets be an indigenous British feature, and did at the Bell Tower, London, are possibly a not appear in France until the mid-thir- late- thirteenth-century modification (see teenth century (Mesqui 1993, 282-3). above), which may have implications for Dressed external surrounds, as used by those present in Marshal work at Chepstow Marshal in loops and slit-lights at Pem- and Pembroke. Oillets of any kind were broke, Caerleon, Chepstow, Kilkenny and notably absent from King Philip’s castles Usk also appear in Angevin work at (Baudry 2002, 64; Mesqui 1993, 284-5 and Kenilworth (Lunn’s Tower), Dover map), but were adopted by his subjects, (Avranches Tower and Peverell’s Gate), under Angevin influence, during the early and at London’s Bell Tower where they thirteenth century e.g. the chamber-tower may, however, belong to later modification at Laval, Mayenne (also triangular), now (see above). Plunging embrasures, also em- dated 1218-30 (Chollet and Gousset 2012, ployed by the Marshals, were in fairly wide 261-2; Mesqui 1993, 285). use by the late twelfth century, in both royal Basal oillets of rectangular form were and baronial work e.g. Gisors and Cluis. used extensively in Marshal work, elder External offsets, as in Pembroke’s don- and younger, but are restricted elsewhere. jon, Chepstow’s Middle Ward towers and They occur at Pembroke, notably in the Usk’s Garrison Tower, are seen in An- donjon parapet and the Dungeon Tower gevin work both in Normandy and the (n.b. oillets were unnecessary in the slit- British Isles appearing, inter alia, in Henry lights piercing the body of the donjon), in II’s donjon at Château-sur-Epte (Fig. 2), Chepstow’s Middle Ward and upper barbi- his Avranches Tower at Dover and flank- can, in the Garrison Tower at Usk, and at ing towers at Bamburgh, Northumberland Caerleon Castle (Mon.). The last, which (late 1160s-80s), Richard I’s towers at demonstrably post-dates Marshal’s acqui- Loches and the Tower of London (1190s), sition of the castle in autumn 1217 (Holden and under King John, in the southwest et al. 2004, 391; Jones 1952, 95-6; Knight tower at Arques (1201-4; Fig. 13) and 2008, 60),37 is the only dateable example, Lunn’s Tower at Kenilworth. which compromises its usefulness in dat- ing the others. They have all been suggest- They also appear in British baronial work ed to be late, from 1213 onwards (Knight of the period e.g. the shell-keep towers at 2008, 60), but Marshal may have employed Lewes. Absent at Compiègne, they appear the form for some time before it appeared in around 50% of Tours Philippiennes at Caerleon: a rectangular oillet can be seen (Mesqui 2018) and, while employed at low down in the spiral stair to Pembroke’s Vernon in the 1190s (fig. 7), are mainly Wogan cavern which, given the overall post-1204 as at Chinon, Falaise, Lille- ‘Transitional’ character of the surrounding bonne, Rouen and Verneuil-sur-Avre – a work, similar to that in the donjon, was pattern followed in Philip’s mural probably under construction before Mar- towers.38 Offsets are, however, present in shal left for Leinster in 1207 (Day and some other cylindrical towers in France, Ludlow 2016, 66; see Appendix 1). Archi- for instance at comital Châteaudun.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16229 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

ABOVE: Fig. 13 – King John’s southern towers and gatehouse at Arques-la-Bataille (Seine-Mari- time), built 1201-3. Left: plan (from Pitte 2007). Right: the southwest tower, from east.

BELOW: Fig. 14 – Plan of the northwestern mural towers, and donjon, at Brosse, Indre in Poitou (from Baudry 2002).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16230 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

The pronounced, plain basal batters seen at like the Marshal towers, and none occupies Pembroke and Chepstow are similarly a a corner (Hulme 2014, 217; Fig. 5); King feature of Angevin design, both in mural John’s mural towers at Arques are closely towers and gatehouses – Dover’s inner ward comparable, but show barrel-vaulted interi- (1180s), King John’s mural towers at ors (secondary?; Langeuin 2002, 356 fig., Arques and Bonneville (Figs. 3 and 13), the 361-5; Fig. 13). Similar-shaped towers Porte Saint-Nicholas at Falaise (Fig. 20), were built in similar locations at, for exam- etc. – and in several great towers e.g. Dub- ple, the north curtain at Chinon, by Richard lin, Conches-en-Ouches, and Château Gail- I or John (Mesqui 2013, 77 fig.), at King lard where it is very substantial (Fig. 26). No John’s Bonneville (south tower, see Fig. 3), batter is present at Compiègne (Fig. 6), nor and later in his castle at Limerick, in the in the Tours Philippiennes at e.g. Chinon, northeast tower which occupies a length of Montlhéry and Vernon; when used in curtain wall between two angles (Fig. 12). Philippienne work, moreover, batters are Comparable interval towers were built on often shallow as at Dourdan, Falaise, Gisors the northwest curtain at Brosse by the vis- and Péronne (see e.g. Mesqui 2018, 132-9; comte of Brosse, under Angevin authority, see Fig. 7). A prominent feature at the Lou- between c.1204 and 1242 (Baudry 2002, 50 vre is the stepped plinth that runs around the and fig., 52; Fig. 14).40 base of all walls and towers (Fleury and In Philippienne castles, a preference for Kruta 2000, 8, 12, 16, 57); nothing like it is 39 D-shaped interval towers is apparent, be- present in any surviving Marshal work. ginning at the Louvre and continuing at ‘Marshal’ towers Falaise, Péronne, Dourdan and many oth- A number of well-known and well-defined ers (see Fig. 15), while gate-towers too characteristics appear in Marshal work, were normally D-shaped – a pattern that elder and younger, from which a distinct was largely followed in baronial France style evolved that was to be highly influen- during the early-mid thirteenth-century tial in west Wales for over fifty years (Day e.g. at Diant, Seine-et-Marne and Mez-la- and Ludlow 2016, 68; Goodall 2011, 162, Maréchal, Loiret (Mesqui 2013, 60-3 and 164; also see Wiles 2014); most of them, figs.; Fig. 15). Most Philippienne interval however, also occur in Angevin work of or gate towers that were circular in plan are the period. For instance, while the interval later than 1204, including those at Rouen, towers in Chepstow’s Middle Ward show at Montargis (Loiret, Île-de-France) from the ‘standard’ D-shaped plan, the Marshals 1200-23, and in the contemporary outer also employed circular towers – not only at gate at Yèvre-le-Châtel, also Loiret (Cor- visier 1998(2), 435; Mesqui 1997, 418-19; corners, but also in shallower angles as at 41 Usk, Cilgerran, and the Dungeon Tower at Mesqui 2013, 22 fig., 43 fig.). Pembroke (see Figs. 9 and 16). But in this The pronounced ‘bulging’ outline (or the Marshals were following in royal foot- entasis) that characterises the Middle steps. As noted by Richard Hulme, these Ward towers at Chepstow, the Dungeon towers closely resemble those at King Tower at Pembroke and the towers at Richard’s Château Gaillard, particularly in Cilgerran can also be seen in Angevin its middle ward where they recurve strong- work, in the southwest tower at Arques ly to join the curtain, have circular interiors (Figs. 13 and 33); it too was adopted

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16231 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 15 – ‘Philippienne’ plans. Left: royal Dourdan (Essone). Right: baronial Diant (Seine-et- Marne). Both first quarter of the thirteenth century (from Mesqui 2013). Fig. 16 – Overall plan of Pembroke Castle (from Day and Ludlow 2016).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16232 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design elsewhere in west Wales, for instance in the qui 1993, 179-80). Like other features inner ward towers, from the 1270s, at mentioned above, they became widely Kidwelly. None of which however is to adopted in thirteenth-century west Wales, deny that certain idiosyncrasies can be seen from which they spread to other regions of in Marshal work, elder and younger; none Wales, and beyond (Ludlow 2018, 268-9). of them, however, appear in King Philip’s Geometric planning castles, instead contrasting markedly with his preferred regularity. They include the King Philip II’s castles are noted for their asymmetry of the tower footprints at Chep- geometric planning, which was subse- stow, Cilgerran and Usk, with their eccen- quently adopted by his successors at e.g. tric interiors, and the unequally-sized Saumur, Maine-et-Loire, built 1220-30 towers in the gatehouses at Chepstow, Car- (Mesqui 1997, 348), and by his barons at marthen, and probably Kilkenny (see be- e.g. Diant and Mez-le-Maréchal, Loiret low). (Mesqui 2013, 47 and fig., 60-1 and fig., 63; Fig. 15). It has also been suggested at Another Marshal attribute and, perhaps, Marshal’s Kilkenny and Usk (Knight their main departure from Angevin prac- 2018, 164; O’Keeffe 2009, 287 and n. 37; tice, is the rounded rear face seen in some Sweetman 1999, 50). of their towers e.g. the outer gatehouse at Chepstow, and in the interval towers at Usk The earliest geometric castle within the and Cilgerran, which may all be later-peri- Capetian and Angevin ambit is thought od Marshal work. But although the rounded to be Druyes-les-Belles-Fontaines rear face is a defining feature of the en- (Yonne), an important pre-Philippienne ceinte at Angers, built by Louis IX during castle dated to the 1170s (Fig. 17); as the 1230s, and was later employed by the the residence of Peter de Courtenay, French Crown at Carcassonne and else- Count of Nevers and a cousin of the where (Mesqui 2013, 74-5 and fig.; 354 French king, it undoubtedly influenced fig.), it does not seem to have been repre- Philippienne planning (Mesqui 2013, sentative of Philippienne design. It was 266-7). Quadrangular and symmetrical, again to prove persistent in west Wales, with circular corner towers, it shows where mid-thirteenth-century towers at neither donjon nor dominant tower Pembroke Castle itself (Fig. 16), at Llanst- (Mesqui 2013, 40-1 and fig., 60). Mar- effan and Laugharne (Carms.), and shal passed close by, or possibly even Llawhaden (Pembs.), all but stand astride through Druyes with kings Richard and the curtain.42 Philip, on their way from Donzy (Nièvre) to Vézelay Abbey (Yonne) for Contemporary with the elder Marshal’s the crusader conference in July 1190 towers at Pembroke and Usk are boldly (Landon 1935, 36; Round 1899, 35); projecting, rectangular latrine turrets (Figs. Peter de Courtenay accompanied the 9 and 16). Similar rectangular latrine tur- kings to Palestine, and the right of hos- rets also appear in Angevin work of c.1180- pitality at Vézelay, claimed by the 1204 at Château Gaillard, Chinon and Fal- Counts of Nevers (Huygens 1976, xxvi, aise (Mesqui 1993, 176; Mesqui 2013, 39 38, 47, 229), was doubtless significant fig., 123 and fig., 264 fig., 296), but were during the royal stay at the abbey.43 not a feature of Philippienne design (Mes-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16233 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

The two kings arrived in Palestine the spatial relationship between the latter and following summer,44 where quadrangular the curtain wall is however paralleled at castles without donjons had also been built. Bonneville, and later at Pembroke (see Philip spent less than fifteen weeks there, all Figs. 3 and 16). Also suggestive of Rich- of them at Acre apart from three days at ard’s hand are the solid, semicircular gate- Tyre (Runciman 1951, 34, 52). Richard’s towers, similar to those in the Porte des experience, though longer, was confined to Champs at Falaise, built by the Angevin the Kingdom of Jerusalem and specifically kings in the later twelfth century (see be- the coastal strip between Acre and Ashkelon low), and a circular interval tower like (now in Israel). The castles he is known to those discussed above.47 Richard visited have seen, some of which he began rebuild- Pont-de-l’Arche twice in summer 1195 ing, were concentrated around the latter and (Landon 1935, 102, 104), perhaps an indi- were mostly rectangular enclosures with cation that work was underway. four square corner towers, dating from the The Norman exchequer rolls, combined 1130s onwards (Kennedy 1994, 31-2; with recent research – much of it French – Landon 1935, 57-62; Pringle 1989, 14, 17- show Richard I to be one of the busier 18).45 The circular corner towers at Druyes castle-builders of the medieval world (if not may be the earliest in either France or Brit- so in Britain): in addition to the above-men- ain where flanking was, until the 1190s, tioned sites he spent heavily in France, for mainly an Angevin feature, and the curtains example at the eastern Normandy castles of at Henry II’s Loches, Gisors, Rouen and Le Bellencombre, Boutavant, Courteilles, Châtelier (Indre-et-Loire, Touraine) appar- Gamaches, Lyons-la-Forêt, Moulineaux, ently show the earliest rounded interval Orival, Radepont, Tillières-sur-Avre, Vau- towers (Corvisier 1998(2), 184; Héliot and dreuil and Verneuil-sur-Avre (Pitte 2002, Deyres 1987, 24, 42; Mesqui 2013, 37, 43, 166; Powicke 1913, 281-2; Stapleton 1840 69, 259-61). All predate the adoption of the and 1844, passim).48 But much of this work rounded tower into mainstream Crusader has been lost or altered beyond recognition; design (see below). we have only a partial glimpse of the full At least one castle of Richard I, begun on repertoire of his designs. his return, appears to show geometric plan- It is also still uncertain how much Rich- ning and a circular corner tower. Nearly ard, and western Europe generally, may £1000 was spent by the king at Pont-de- have owed to experience of crusader Pal- l’Arche in 1195 (Stapleton 1840, 137, 156, estine (see inter alia Goodall 2011, 162; 236-7).46 The castle, which fell to Augustus Hulme 2014; Kennedy 1994, 186-9; Mes- in mid-1204, has now gone but the sources qui 2013, 37-8, 41, 264-6; Pringle 1988, show that it was not of standard Philippi- xxi-xl; Pringle 2014, 48-56). For instance, enne design. It comprised two baileys in an while the main body of work at the Louvre arrangement not unlike that at Château had to wait for King Philip’s return from Gaillard (Fig. 18); the inner ward, howev- crusade, at the end of 1191, to get fully er, was a regular rectangle. The question of under way (Mesqui 2013, 162), it had been how much of this work might have been commissioned before he departed (Fleury Richard’s, and how much Philip’s, is still a and Kruta 2000, 2), including, presum- matter of debate (Launay 2015), including ably, its geometric layout – which, as the freestanding donjon (see above); the

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16234 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 17 – Axonometric view of the castle at Druyes-les-Belles-Fontaines (Yonne), built in the 1170s; the gate-tower is later (from Mesqui 2013). Fig. 18 – Plan of the castle at Pont-de-l’Arche, Eure (from Mesqui 2013)

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16235 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 19 – Coudray-Salbart, Deux-Sèvres. Left: in c.1202-6. Right: in 1242 (adapted from Mesqui 2013). Fig. 20 – L-R: The Porte Saint-Nicholas at Falaise, built by Henry II or Richard I, and Chepstow’s Middle Gate.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16236 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design noted above, was probably derived directly The transformation of Coudray into a from Druyes. Marshal too had spent time in geometric castle may have coincided the Middle East, having undertaken a pil- with the establishment of a rectilinear grimage between early 1184 and early 1186 plan at King John’s Dublin. We saw (Crouch 2015, 9; Holden et al. 2006, 105- above that there is some uncertainty 6). But, while this potentially means expe- regarding the constructional history of rience of eastern fortification, our only the castle but, as at Coudray, develop- source for the visit is the biographical His- ment was incremental; it only assumed toire, which is silent about Marshal’s its present form after 1213, with work movements to, from and within Palestine continuing until around 1230 (Manning (Holden et al. 2002, 371).49 2003, 72-3; Fig. 8; Tietzsch-Tyler 2018, King John had visited Paris in the early 121). Staged development has also been 1190s and 1201 but, while geometric plan- recognised at John’s Limerick Castle, which did not achieve its current geo- ning is evident in the later Angevin castles metric plan until the reign of Henry III at Dublin and Coudray-Salbart, it may not (Tietzsch-Tyler 2013; Fig. 12). result directly from Philippienne influence: Richard I may have provided models closer Ben Murtagh has pointed out that any to home (see above). The Tower of Lon- rectilinearity at Marshal’s Kilkenny – don, too, was by 1200 a markedly rectilin- which, at present, is uncertain – is proba- ear castle, even if not ‘correctly’ geometric bly incidental and dictated by the underly- (Impey and Parnell 2011, 22 fig.).50 ing, earlier ringwork (Murtagh 2017, 173; see Fig. 8). The same appears to be the Dublin and Coudray-Salbart were, more- case at Usk which is not truly rectangular, over, the result of gradual development. In showing a sub-oval plan defined by ten its final phases, before it was lost to the lengths of straight walling, similarly fol- French in 1242, Coudray-Salbart was a lowing a preexisting earthwork enclosure rectangular castle of great regularity. Work (Fig. 9);52 asymmetrical in outline, the probably began in 1202, paid for by subsi- enceinte incorporates an earlier rectangu- dies from King John (Curnow 1980, 44-5). John was in Poitou from June until October lar tower (Knight 2008, 57-8). 1206, staying at Coudray itself on 22 Sep- Marshal’s work at Usk is difficult to date; tember, and at nearby Niort in August and a date after 1213 has been argued on the October (Hardy 1833, 73-4; Hardy 1835a, basis of the arrowloops (Knight 2008, 60), 667) so further visits are possible. Howev- the uncertainties over which have howev- er, Coudray did not become geometric until er been discussed above. Nevertheless the it was enlarged in Phases 3-4, which may earl stayed at Usk in July 1217, and at his belong to the second or third decades of the castles at Chepstow and Goodrich, Herefs. thirteenth century; the castle John saw in (Hardy 1833, 314; Pat. Rolls 1216-25, 1206 was much smaller and of irregular 79), and while this visit was primarily a plan (Mesqui 2003, 79-80; Fig. 19). It is response to the growing threat from the however possible that Phase 3 work may neighbouring Welsh lord Morgan ap Hy- have started, at least, when John was in wel of Caerleon (Holden et al. 2006, 186- Poitou during his 1214 campaign and, 7), he was again at Usk in December 1217 though he is not recorded at Coudray itself, (Hardy 1833, 348). The visits may coin- a visit cannot be ruled out.51 cide with building work in at least one of

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16237 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design these castles – namely Goodrich, which Gates and gate-towers possibly does show evidence of geometric Marshal castles show a variety of arrange- planning. ments at the main entry but, by and large, It has long been thought that the founda- they reveal a preference for relatively sim- tions of an early, circular tower at the ple forms. Chepstow’s Middle Gate is a southwest corner of Goodrich, and the straightforward doorway with one pair of lower courses of its east curtain, belong gates and no portcullis (Avent and Turner to the Marshal period (Ashbee 2009, 30; 2006, 64-5). Its detail, which survives, is Radford 1958, 5, 8). According the His- in the ‘Franco-Angevin’, Transitional toire, the castle was attacked in October Gothic style employed throughout north- 1216 (Holden et al. 2004, 269; Holden et ern and central France during the late al. 2006, 166), which may have triggered twelfth century (Fig. 20). It can be com- building work; the earl is again recorded pared with that in the Porte Saint-Nicho- at Goodrich in September 1218 (Hardy las at Falaise, built by Henry II or Richard 1833, 370), perhaps providing an oppor- I (see above) – particularly in its cham- tunity to check its progress, which may fered jambs and imposts – and in the entry have continued under his sons. The foun- at Bonneville-sur-Touques. The Chep- dations show a plain, circular corner tow- stow jambs show pyramidal chamfer- er which is otherwise little-known stops in a very early context, but barred (Shoesmith 2014, 175); its relationship stops were used in King John’s work, with the east curtain however suggests a from 1201-4, in the Gloriette at Corfe compact, geometrical castle defined by Castle (Dorset), where the triangular- the present square ditch. Nevertheless, headed rear-arches were emulated by Ron Shoesmith considers that no work Marshal in Chepstow’s upper bailey tower was undertaken at Goodrich under the (see Turner 2006, 79-80). So it may be Marshals, and that the castle retained its that the Middle Gate was begun soon after earlier, uncertain form until the late thir- Marshal’s return from Normandy in late teenth century, when he suggests the 1203. present ditch was cut and the corner tow- It has been suggested that the outer ward er was built (Shoesmith 2014, 22, 106-8, and its curtain were already in existence 117-18, 129-30).53 when the Middle Ward was built, because If Goodrich does feature Marshal-period of the disposition of loops in its south geometric planning, the long passage of tower (Avent and Turner 2006, 64; the time since the earl’s experience of either rectangular oillet here is discussed above), Palestine or Druyes-les-Belles-Fontaines but this may be incidental; the fact that the must be considered. Neither is then likely Middle Ward east curtain was flanked by to have been a direct influence which, two towers, the southern of which is a instead, was again probably derived from corner tower, suggests that it was an exter- the Angevin Crown.54 nal line of defence when built. Straight curtains were not unknown by c.1200 (cf. London, Warkworth etc.), while Chep- stow Castle occupies a ridge across which a straight defensive line is easiest.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16238 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

The gate at Usk is also a simple doorway Angevin design is explicit in the southeast but here the arch is notably plainer, with a gate built in 1201-4 at Arques, which was single order, while moulded detail is ab- a single tower, of circular plan, with a sent. And, unlike Chepstow, there is a port- straight passage. Now destroyed, it project- cullis (Knight 2008, 65), possibly operated ed wholly from the curtain into which, like from the parapet as in the Younger Marshal the mural towers, it strongly recurved postern at Cilgerran (see Hilling 2000, 21). (Langeuin 2002, 361 fig., 363; Fig. 13).56 The main gate at Cilgerran, probably from A similar gate-tower was built, before the 1230s, is a simple, square tower, 1204, in Falaise’s inner ward (Fichet de though showing two portcullises (Hilling Clairfontaine et al. 2016, 239, 246) and 2000, 13-14, 18). Little remains of the later, during Phase 4, at Coudray-Salbart similar gate-tower at the Marshals’ Tenby (Fig. 19). St John’s Tower at Dover fol- Castle, Pembs., which is difficult to date lows the same pattern, though isolated on a closely. It is not unlike Cilgerran’s in gen- forework, and may be roughly contempo- eral form, but is considerably smaller and rary with the Coudray gate-tower having shows arrangements for just one portcullis. been added between 1217 and 1221 (Brin- It was later given a barbican enclosure. dle 2015, 23, 44 fig.; Fig. 11). A cylindri- cal gate-tower was also employed by The Phase 1 gatehouse at Coudray-Salbart Walter de Lacy at Trim Castle, Co. Meath, (c. 1202-06) was a semicircular tower with a in 1201-07 or 1220-23 (McNeill 1990, 329 straight passage (Fig. 19), a plan that was not fig., 331, 334; O’Keeffe 2017, 44-7); un- widely adopted in France and, where occur- like Arques and Coudray it stands astride ring, is mainly late thirteenth-century or later the curtain, as it may have done at Falaise as at Bressuire and Thouars, both Deux- where Walter was recorded in June 1200 Sèvres (Mesqui 2013, 314-5). However, it (Hardy 1837, 69; see below).57 The form was employed, probably by William Mar- was later adopted in other regions of north- shal II, in the inner barbican of the Marshal ern and western France, persisting into the castle at Dunamase (Co. Laois) in Ireland fifteenth century (Baudry 2002, 53; Mes- (Hodkinson 2003, 33 fig., 41-2, 48).55 qui 2013, 312-16). A gatehouse of similar plan was built, at A striking feature of Marshal Pembroke around the same time, at Hubert de is the large, D-shaped gate-tower or Burgh’s Grosmont Castle (Mon.), although ‘Horseshoe Gate’ (Fig. 16). Like the inner here it extends into the bailey (Knight curtain to which it is attached, it is sec- 2009, 9 fig., 20-1). It is likely that, during ondary to the donjon and was probably his long periods in Poitou 1202-4, and begun 1204-7 (Day and Ludlow 2016, 1213-1214/15 when he was seneschal of 66); now truncated, it lacks evidence for a the province, de Burgh saw the Phase 1 portcullis.58 It was a hollow tower, pro- Coudray gate-tower before it became sub- jecting wholly from the curtain and en- sumed within the later work there. So he tered through its flank, with a second entry may have been the source for Dunamase, at right-angles. Similar forms occurred as a close ally of William Marshal II until throughout the Mediterranean and Near 1226, when the latter supported a rival’s East, and the design was adopted in the claim in Ireland costing him the justi- twelfth century by the Crusaders (Kennedy ciarship there (Crouch 2015, 20). 1994, 59-61, 91, 150-2; Pringle 1988, xxx,

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16239 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design xxxiv, 32-3; Pringle 1989, 223); it has been feature: a socketed pivot-stone, within suggested that it provided a model for the which the gates were hung (King 1978, Pembroke gate, via Marshal’s pilgrimage in 106; Fig. 21). Like right-angled entries, the early 1180s (Avent 2006, 89-90; King pivot-stones had long been widespread in 1977, 164-5; King 1978, 107 and n. 90; Prin- Muslim fortifications in the Mediterranean gle 1988, xxxviii n. 56; Pringle 2014b, 48, et and Near East, and were adopted by the al.). But all these towers are square in plan. Crusaders during the twelfth century (Fig. Significant questions arise from the twenty- 21).61 While Marshal may have seen simi- year gap before the design was employed at lar pivot stones in the early 1180s, the use Pembroke, its use in a rounded tower and, of such an engineering feature at Pem- perhaps most crucially, the presence of a very broke is likely to represent a mason’s pref- similar D-shaped gate-tower at the castle of erence rather than a patron’s instruction Saranta Kolones in Paphos, Cyprus. (Malcolm Hislop, pers. comm.). Richard I Saranta Kolones was begun after the con- undertook building works at a number of quest of Cyprus by Richard I, in May 1191, castles in the Kingdom of Jerusalem (see on his way to Palestine (Pringle 2014b, 48). above), and his masons will doubtless Much of the work appears to have been have been familiar with the feature, while undertaken by the Lusignans, kings of Jeru- the presence of masons with experience salem and Cyprus, between 1192 and 1205 under the Angevin Crown, at Marshal’s 62 (Avent 2006, 90 and fig; also see Runciman Pembroke, is argued below. 1951, 44-7, 84, 103), but was the castle David King considered it possible that planned in summer 1191, under Richard I’s another ‘horseshoe’ gatehouse was built administration? Its concentric plan pre-exist- by the Marshals, after 1217, at Caerleon ed at Henry II’s Dover (as noted in Pringle Castle, on the strength of two drawings 2014b, 48). The rounded tower moreover from c.1800 (King 1977, 165). Malcolm was almost unknown in existing Crusader Hislop has recognised that a fourth exam- work (see e.g. Kennedy 1994, 18, 85-96, ple exists at Montgomery Castle, built by 115; Pringle 1988, xxxii-xxxiii) and, while Hubert de Burgh 1223-8 as a postern to his Armenian influence may lie behind its use at inner ward (Hislop 2016, 168-9; Fig. 22). Paphos (Pringle 2014a, 368-70),59 both gates While independent Angevin influence is are, in form, very like the D-shaped mural possible, the initially close relationship towers at Angevin Gisors (cf. Figs. 10 and between de Burgh and the younger Wil- 16). Might it be a western adaptation, under liam Marshal must be taken into account. King Richard, of an eastern design? It may be significant, too, that William A connection with Richard I might explain Marshal II was responsible for administer- Marshal’s adoption of the design.60 And ing scutage in Montgomery in early 1225 while Paphos may more likely be post- (Hardy 1844a, 16, 28, 30), while his clerk 1192, Richard may yet have provided the Peter of Luton disbursed payments for conduit to Pembroke: Geoffrey de Lusig- works there, on behalf of the Crown, in 1224-5 (Hardy 1833, 586, 591, 597, 617; nan returned from Palestine and was inter- 63 mittently in the king’s retinue, with Hardy 1844a, 28, 31). Marshal, 1196-99 (Landon 1935, 113, 123, The adoption of the ‘horseshoe’ form at 144; Runciman 1951, 84). Nevertheless, a fifth castle, Caldicot (Mon.), probably the Pembroke gate exhibits another eastern during the 1230s (Guy 2016a, 161, 174),

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16240 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 21 – Left: the lower pivot-stone in Marshal’s ‘horseshoe’ gate, Pembroke Castle. Right: the upper pivot-stone for the gate at Krak des Chevaliers, Syria (photograph: Jean Mesqui).

Fig. 22 – Plan of the inner ward at Montgomery Castle, 1223-8, showing the ‘horseshoe’ postern gate-tower at the north end (adapted from Knight 1993).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16241 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design may be attributed to the dominance of the disposition of the towers67 – the chambers younger Marshals at nearby Chepstow over the passage, and the rear section, ap- (King 1978, 106-7).64 Two further exam- pear to be later additions. It is even possible ples may exist in the north of England. The that the towers themselves were not com- bailey at Sandal Castle, Yorks., is occupied menced until Marshal’s son succeeded in by a thick-walled, D-shaped structure lead- 1219 (Guy 2016a, 195 and n. 18); their ing to the donjon, entered from the side and cruciform loops are normally a feature of shown as a tower in a sixteenth-century Younger Marshal work (see above), while drawing (Hislop 2016, 174 and n. 6; Ke- the spiral-stair shaft protrudes into the nyon 1990, 81). It was probably built ground-floor chamber of the southern tow- c.1265-71 by the earl of Surrey, John de er, as in the post-1219 Dungeon Tower at Warenne (Hislop 2016, 174), whose moth- Pembroke (see below) and Cilgerran’s west er was William Marshal’s eldest daughter tower. But the dating of the original entry Matilda (Crouch 2015, 39);65 the sixty relative to the Middle Gate – which was years separating it from Pembroke must suggested above to have predated the outer however be taken into account. The gate- ward – remains problematical. tower at Barnard Castle is similar, though Another twin-tower gatehouse was for- ruinous and yet to be closely dated (Fig. merly to be seen at Kilkenny (Fig. 8), but 23); dates between the twelfth and fifteenth was demolished in the eighteenth century centuries have been suggested (see Guy meaning that, although it may have been 2013b, 46, 56), though construction in the begun by Marshal c.1207,68 we cannot be late thirteenth century and a relationship sure whether its constructional history was with Sandal is favoured by Malcolm His- equally complex. Ben Murtagh and Dan lop (pers. comm.).66 Tietzsch-Tyler agree however that the Twin-towered gatehouses towers were unequally-sized, as at Chep- In contrast with the above entries, Chep- stow (Murtagh 2017, 171; Tietzsch-Tyler stow’s outer gatehouse is a complex, twin- 2018, 126); Murtagh also suggests that towered structure. Its doors, which sur- they had round backs and lacked any ad- vived in situ, yielded a felling-date of joining rear-section, as in the intermediate 1159-89; other features of the gates sug- stages at Chepstow. While Marshal’s oth- gested that joinery had taken place shortly er work is notable for its conservatism, after felling, while the timber was still rather than innovation, Kilkenny’s impor- green, and had not been re-used in the gates tance as his new caput in Leinster must be (Avent and Miles 2006, 52). So the en- taken into account, just as Chepstow was trance arch within which they hung appears his favoured castle in England and Wales to have been built fairly soon after Marshal (see Appendix 2). But an asymmetric, gained possession of Chepstow in 1189. twin-towered plan, with one tower larger But Neil Guy has shown that the gatehouse than the other, is also seen in the gate- is the product of incremental development, house at Carmarthen Castle, which was probably extending into the early four- almost certainly built by the younger Wil- teenth century (Guy 2016a, 192-5). While liam Marshal, in the mid-1220s, rather than the entry itself may be the work of the elder by his father (Ludlow 2014, 183-4, 189, Marshal – along perhaps with the basic 199-202). The gatehouse was rebuilt in

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16242 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

1409-11, but following the footprint of its Rolls 1257-1300, 173; Landon 1935, predecessor in which the towers also re- 141). Marshal probably also had experi- curve strongly into the curtain (but not the ence of the gatehouse at Domfront, as entry; Fig. 24). noted above; probably begun around There is moreover little indication in the 1198, it has two D-shaped, but seven-sid- sources that the elder Marshal showed any ed towers, with closed backs (Mesqui great interest in his British estates before he 1997, 152-3 and fig.).71 The two latter was created earl in 1199 and he preferred, gatehouses are comparable with the north- when not on Crown service, to reside in west gate at Angevin Dover, built between Normandy until its loss in 1204 (Crouch the early 1180s and 1216 (Brindle 2015, 2016, 96-8; Power 2003, 211). And it was 23), where the D-shaped towers appear to here, in his caput at Longueville-sur-Scie have open backs like the similar gate- (Seine-Maritime), that Marshal himself house at Pevensey, Sussex, which is may have built a twin-towered gatehouse. thought to have been built by Richard I The towers are rounded but are now in very (Chapman 2007, 102-9); here, however, poor condition, difficult to interpret, and the towers are elongated in a form that subject to very little published analysis was later adopted – with closed backs – at (Jean Mesqui and Jacques le Maho, pers. Kenilworth (c.1210-15) and Dublin (per- comm.; Fig. 25).69 They are solid as at haps 1220s; Fig. 8).72 Pont-del’Arche, Falaise’s Porte des King John’s twin-towered gatehouse at Champs, and at Hamelin of Anjou’s Conis- Limerick Castle can be dated to c.1211-12 brough and Castle Acre (Norfolk), which (Tietzsch Tyler 2013, 145). It is not unlike may all be from the 1190s (see above; Guy Chepstow in the sheer scale of the towers 2016a, 151), but appear to recurve behind (D-shaped here, though with circular inte- the entrance passage. They also show very riors), and its triangular loops (Fig. 12). deep offsets, as in Richard I’s Bell Tower Influence in the other direction, from Mar- at London; it is even possible that, like the shal to the king’s work, is however unlike- Bell Tower, their lower stages were polyg- ly: John’s suspicions of the earl, onal (cf. Domfront). So the gatehouse – beginning in 1205, did not fully abate which lacks a portcullis – may be from until he was recalled from Ireland in 1194-1203, when Marshal was primarily 1213,73 while we have seen that the Chep- 70 resident in Normandy. stow gatehouse may be the later of the Twin rounded gatehouse towers neverthe- two. Moreover, the Limerick gatehouse less had strong Angevin precedents. In differs from Chepstow in its overall ar- addition to the solid Porte des Champs, the rangements, with a contemporary cham- near-contemporary Porte Saint-Nicholas at ber over the gate-passage, which extends Falaise has hollow, D-shaped towers, ap- into the bailey. An early context for the parently with closed backs (Fichet de Clair- similar gatehouse at Carrickfergus Castle, fontaine et al. 2016, 232, 244-5; Fig. 20). Co. Antrim, has recently been questioned Marshal is known to have been at Falaise and a start date after 1210, under King in January 1203 (Holden et al. 2004, 137), John, now seems likely – perhaps ac- and may have stayed there with Richard I counting for the £500 spent at the castle in late December 1198 (see Cal. Charter 1217-22 (Tietzsch-Tyler 2018, 130).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16243 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

ABOVE: Fig. 23 – The ‘horseshoe’ gate-tower in the inner ward at Barnard Castle, Co. Durham.

BELOW: Fig. 24 – Plan of the gatehouse at Carmarthen Castle (first floor), showing asymmetrical towers: early fifteenth-century on a footprint from the 1220s (from Ludlow 2014).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16244 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Architectural detail eville and Neufmarché – and e.g. Fréteval – Marshal-built carved detail only survives in it lies on level ground. In this it is also similar situ in Chepstow’s Middle Gate and at to the ‘Angevin’ donjon at Hamelin of An- Pembroke, in the Transitional Gothic win- jou’s Conisbrough. Conisbrough is also the dows of the donjon and Wogan screen- only pre-existing donjon of this kind in wall. The donjon was commenced in 1201, which a large, twin-light window, directly influenced by Angevin prototypes in north- over the entry, survives – as at Pembroke ern France (see above) including the stark which it may have influenced (noted by simplicity of its overall treatment. The de- Goodall 2011, 163; Figs. 26 and 27). Both tailing and dome however show specific donjons also show a large third-storey win- Franco-Angevin attributes: it is likely that a dow; in design, however, the windows are master-mason(s) from Normandy, Anjou or rather different at the two sites. Touraine, or with direct experience under The Pembroke windows appear ultimate- the English Crown there, was personally ly to be derived from the Franco-Angevin responsible. If not present from the first, his pattern seen at Château Gaillard (Fig. 26), recruitment may have occurred after the and in Henry II’s work in, for example, the final loss of these territories in 1204-5, rectangular donjons at Portchester and Ap- when a number of craftsmen who had pleby, Westmorland (Goodall 2011, 37 served the Crown in Normandy were re- fig., 137 and fig.). The twin lights at all corded in the British Isles (Hislop 2016, four occupy plain, arched surrounds with 151; Hulme 2014, 225-6). It was observed solid tympana, and broad mullions – above, moreover, that the Pembroke chamfered at Pembroke and Château Gail- mason(s) may have had direct experience lard – take the place of cylindrical shafts; of royal works. Lindy Grant has shown that Pembroke and Château Gaillard also share the Angevin kings used native artisans in slightly pointed outer arches. The lights Normandy (Grant 1994, 78); at least one however are lintelled in the Angevin win- was from the vicinity of Château Gaillard, dows (‘Type B3’, see Mesqui 1993, 234 and was in King John’s service after 1204, fig.), unlike the voussoired heads em- the master-carpenter Nicholas de Andeli ployed at Pembroke which are pointed on (Colvin 1963, 61-2; Hislop 2016, 153 n. 15; the second floor, and semicircular at third- Hulme 2014, 225).74 In this context, it may floor level (Mesqui’s Type B2). The over- be significant that Marshal was in Pem- all disposition of Pembroke’s windows, brokeshire during late 1204, and David but not their form, is broadly paralleled in Crouch suggests that he spent some weeks the tower at Châteaudun (Mesqui’s Type in the area (David Crouch, pers. comm.), A2), visited by Marshal in 1188. where he probably issued his grant to The figurative carved heads in the Pem- Monkton Priory, Pembroke (Caley et al. broke tympana are demonstrably derived 1846, 320-1; Crouch 2015, 161-2); the from those in the windows of the rectangu- composition of its witness-list suggests lar donjon at Chambois in Normandy (Or- Marshal at least visited Pembroke Castle to ne), built by William de Mandeville, Earl view the ongoing work there.75 of Essex, before his death in 1189 (Mesqui There was no motte at Pembroke for the 1993, 217-8 and fig.; Mesqui 1997, 102; donjon to occupy and, as at Angevin Bonn- Fig. 27). Marshal purchased Chambois

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16245 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

ABOVE: Fig. 25 – Marshal’s castle at Longueville-sur-Scie (Seine-Maritime), Normandy. Left: sketch plan. The rectangular block behind the gatehouse is secondary (the large southwest flanking tower, at least, is late-medieval). Right: the southern gatehouse tower, from northeast, in 2017.

BELOW: Fig. 26 – The donjons at (L-R) Conisbrough (Yorks.) and Château Gaillard (Eure).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16246 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 27 – Left: the second-floor window in the donjon at Pembroke (photograph: Neil Guy). Right: a window in the donjon at Marshal’s Chambois (Orne). Note the figurative carved heads in the tympana at both.

Fig. 28 – Pembroke Castle from the northwest in c. 1775 (detail from a drawing by Richard Wilson; National Library of Wales).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16247 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design from King John while in France during out as today, detail is otherwise very dif- May-October 1200 (Hardy 1835b, 39), but ferent from Monkton and St Mary’s and lost it to Philip Augustus in early 1204 all surviving surrounds, apart from in the (Power 2003, 208). Here, the lights also windows, are very plain. The twin-light have slightly pointed heads, in pointed out- window in the Wogan screen-wall is very er arches, but are cut from solid lintel-slabs like those in the donjon and is clearly (Mesqui’s type B5). close in date, ie. 1204-7; it occupies a semicircular outer arch, the lights have While there are no close parallels for the two-centred heads, but dogtooth and Pembroke window design in the British carved heads are absent. Isles, aspects of its detail suggest some English influence from the West Country, Vaulting and domes namely the use of dogtooth ornament in the Although King John used vaulting in the second-floor window.76 Dogtooth was Tour du Moulin at Chinon, it was not widely used in England and Wales during widely used by the Angevins or their bar- the late twelfth and earlier thirteenth centu- ons in either donjons or tours-mixtes (Ta- ries, and can be seen nearby in the nave ble 1) and, when occurring, is normally triforium at nearby St David’s Cathedral confined to basement level e.g. Conis- (Pembs.), from 1182-1200, which belongs brough and Barnard Castle.80 It was more to the ‘West Country School’ of architec- characteristic of Blois, where it could also ture (defined by Brakspear 1931; Thurlby be used on the entry floor as at Château- 2006, 307-8, 328-35). It was, however, un- dun (Mesqui 2013, 137 fig., 214-15), but usual in Angevin France (Hourihane 2012, such expansion of vaulted space remained 300).77 The absence of capitals in the Pem- unusual in France until adopted by Philip broke window-surrounds may also betray Augustus in the late 1190s (see below).81 West Country influence (see Brakspear 82 1931, 6-7), although it can also feature in Châteaudun also has a summit-vault, the Franco-Angevin tradition (Mesqui though with a flat surface rather than 1993, 202 fig., 234 fig.). domed as at Pembroke. Nothing quite like the Pembroke dome exists in other cylin- So it is possible that, like the oolite from drical donjons in either France or the Brit- which its dressings were cut, at least some ish Isles except, perhaps, in the slightly of the donjon’s free-masons were drawn later tower at Mocollop (Co. Waterford) from Marshal’s West Country heartland.78 which may have been built in direct hom- An eighteenth-century view of the castle age (see below). Other significant differ- suggests that the main first-floor entry had ences between Châteaudun and Pembroke a semicircular-headed surround, possibly suggest that it was not a leading influence with continuous roll-moulded order(s) in on Marshal: the internal proportions are the West Country style (Fig. 28; also see very different, while Châteaudun shows Renn 1968, 38), like those surviving in the neither a basal batter nor hourding-sock- entries at nearby Monkton Priory and its ets, while containing both a well (housed dependent church of St Mary, Pembroke, in a masonry shaft) and a mural passage, which are similarly in oolite (Fig. 29).79 neither of which occur at Pembroke. It is However, the depiction may be convention- possible, moreover, that it was an integrat- alised: most surrounds appear to be robbed- ed chamber-tower: toothing in the face-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16248 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 29 – The main entrance at Monkton Priory church, Pembroke, from c. 1200. The doorway at Pembroke St Mary is very similar, as perhaps was the entry into Pembroke’s donjon. Fig. 30: The roofing/dome support arrangement in the donjons at (L-R) Conisbrough and Château Gaillard. A similar ring of corbels can be seen at both (Conisbrough from Brindle and Sadraei 2015).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16249 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design work suggests that a contemporary floor, has been interpreted as an anteroom building was formerly attached (Corvisier serving an audience chamber below (Mar- 1998(2), 148-58; Marshall 2002b, 34). shall 2002b, 32); the large window over- looking the river suggests it may also have Instead, an immediate and perhaps more been a prospect chamber, consistent with a persuasive antecedent for Pembroke’s grand timber dome above. dome may lie closer to home, in Conis- brough’s donjon. The upper works here The Conisbrough donjon was built in the have long been interpreted as an attic floor 1180s, Château Gaillard in 1196-8, and (see Brindle and Sadraei 2015, 9, 13), Pembroke in 1201-7 – around a twenty- which sits rather uneasily within the strati- year period: it is possible that successive fied social environment of the donjon,83 masons from the same Norman/Angevin and Philip Davis’s suggestion that a timber tradition worked at all three.85 Each shows dome instead crowned the tower is both a pronounced basal batter which, at Châ- compelling and consistent with the struc- teau Gaillard, rises two-thirds the height of tural evidence (Davis 2016, 254-7).84 He the tower, possibly due to its location near envisaged a high-status third-floor cham- the edge of a chalk cliff. Conisbrough and ber beneath a grand, decorative domed Pembroke are moreover among the few ceiling – precisely the kind of scheme that cylindrical donjons with four storeys. No we can picture at Pembroke which may be close personal association between Marshal the realisation, in stone, of the same model and Hamelin of Anjou is known,86 but they (see Appendix 1). The Pembroke master moved within the same court ambit while was apparently familiar with, and probably both – respectively as favourite, and nota- had experience of the ecclesiastical archi- bly loyal blood-relative – had close contact tectural tradition of Angevin Anjou and with the Angevin kings whose generous Aquitaine, and its masonry domes at, for patronage, until 1207, Marshal enjoyed. example, the royal funerary church at Fon- The three donjons share the same spatial tevraud Abbey, Maine-et-Loire (Bony relationship with the domestic buildings in 1980, 86; Hulme 2014, 220). their respective castles: it is close, but not The structural evidence suggests that a integrated. Differences in their appoint- similar arrangement to that at Conisbrough ments however suggest differing uses. existed in the donjon at Château Gaillard Conisbrough contains fireplaces, latrines, which, at summit level, shows the same basins and a private chapel; Pembroke circle of closely-spaced internal corbels (as only shows fireplaces (see Appendix 1) noted by Hislop 2016, 39; Fig. 30). A con- while Château Gaillard lacks both latrines centric inner ‘parapet’ wall rose above the and any fireplace. Pembroke and Conis- wall-walk at Conisbrough (Brindle and brough feature hourding-sockets (Brindle Sadraei 2015, 9), arguably carrying a coni- and Sadraei 2015, 13; see Appendix 1), cal roof over the timber dome (Davis 2016, but not Château Gaillard. A large cistern 251, 254); similar walling has been suggest- lies beneath basement level at Château ed at Château Gaillard (Hislop 2016, 147; Gaillard (see Renn 1968, 38), and a well at Fig. 31), and may be paralleled at Pembroke Conisbrough (Brindle and Sadraei 2015, (see Appendix 1). The upper chamber at 8), but neither is so far known at Pem- Château Gaillard, which is the entrance broke. Curving mural stairs were em-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16250 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design ployed at Conisbrough and a spiral stair at towers elsewhere in France it was, like Pembroke, while Château Gaillard appears barrel-vaulting, seldom similarly used in to have been served by internal timber the British Isles.88 87 stairs. A fine, white Magnesian limestone, Nevertheless, the Pembroke dome may ideal for closely-jointed ashlar facework represent the earliest secular use of any and similar to that used at Château Gaillard, kind of vaulting in west Wales and, like so Chinon, Loches etc., was available locally many of the castle’s attributes, influenced at Conisbrough, allowing its Angevin asso- regional building, secular and ecclesiasti- ciations to be emphasised. No such material cal, throughout the thirteenth century and was on hand at Pembroke, which is rubble- beyond (see above).89 The dome was emu- built from the same tough Carboniferous lated by summit-vaults in later work at limestone as the cliff on which the castle Pembroke Castle itself, in the engaged, stands; in this, however, it has precedents in dominant tour-mixte built in c.1230 at the the undressed facework used at Château- southeast corner of nearby Manorbier Cas- sur-Epte and Neaufles (Fig. 2). tle (King and Perks 1970, 107, 114-15) – There are however further similarities caput of a subordinate barony of Pem- between the three castles, which may be broke – and in the similar northwest tower more than incidental. The comparable rela- at Laugharne Castle, Carms., from the tionships between donjon and curtain wall 1240s-50s (Avent 1995, 9-10, 34). The have been discussed above. The curtain at southwest tower at Kidwelly, from the Conisbrough, which may be from the late 1270s, has a flatter saucer vault at the 1190s (Guy 2016a, 151), has solid semicir- summit, to which a second was added cular interval turrets as at Henry II’s Loches when the tower was heightened in the and Rouen (see above), and like those in his early fourteenth century (Kenyon 2007, donjon at Niort (Mesqui 1997, 266-7; Mes- 9-11, 37), while the slate roofs over both qui 2013, 196); these turrets may also be towers at Younger Marshal Cilgerran may reflected in the lobate inner curtain, other- have mimicked shallow domes (Hilling wise unparalleled, at Château Gaillard (His- 2000, 15 fig.).90 Similarly, the twin-light lop 2016, 147; Tietzsch-Tyler 2013, 154) donjon windows at Pembroke were emu- and in the solid semicircular turret on the lated in the engaged, dominant eastern Wogan screen wall at Pembroke. tour-mixte at Cilgerran, in the similar In contrast to the above-mentioned towers, southeast tower at Manorbier, and in later work at Pembroke itself, in both the castle it is the rib-vault that predominates in King 91 Philip’s Tours Philippiennes (Mesqui 2013, and the town walls. 215-6, Fig. 7). And it is often used at every Such emulation shows ‘the admiration in level, sometimes in combination with bar- which Pembroke was held’ (Goodall 2011, rel-vaulting, to give summit-vaults at many 162-3; also see Wiles 2014), but also rep- sites e.g. Lillebonne and Gisors, but with resented a visible means by which Marshal flat surfaces as at Châteaudun. Its absence tenant lords might express their homage to from Pembroke’s donjon is another attribute – and affinity with – the earls. The same that sits uneasily with suggested Philippi- process may also have been at work in enne influence. And though the rib-vault Ireland. Mocollop Castle, Co. Waterford, became widely adopted in cylindrical great shows a free-standing cylindrical donjon

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16251 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 33 ABOVE: Plans of the ‘towered keeps’ at (top) Nemours, Seine-et-Marne, and (bottom) Carlow Castle (from Leask 1964, and Mesqui 2013). brough from Brindle and Sadraei 2015). Fig. 32. BELOW: First-floor plan of the donjon at Le Châtelier (Angevin Touraine), showing Fig. 31: ABOVE: Viollet-le-Duc’s reconstruction the stair turret. The curtain may be closely of the donjon at Château Gaillard. contemporary (from Corvisier 1998).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16252 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design begun 1215-20 (O’Keeffe and Whelan lights), and a large cistern lay beneath forthcoming; Tietzsch-Tyler 2018, 121). It basement level as at Château Gaillard. was almost certainly built by Thomas Fitz- The Lacy brothers, Hugh of Ulster and Anthony (d. 1226/7), who had been granted Walter, lord of Meath, were close associ- the castle and lordship in 1215 (Sweetman ates of William Marshal with whom they 1875, 89-90). The window(s) may emulate were military allies in 1207-8, perhaps those at Pembroke, while the third storey belonging to his wider affinity (Crouch shows a vault which O’Keeffe and Whelan 2016, 124-30, 218; Holden et al. 2004, consider to have been built when the tower 189). It is possible therefore that Pem- was heightened in the fifteenth century. broke, which was probably more-or-less However, FitzAnthony had been part of the complete when Marshal left for Ireland in Marshal mesnie since 1207, and was their 1207, was an influence at Dundrum. But, seneschal of Leinster 1215-25 (Bradley and while Hugh de Lacy is not recorded in Murtagh 2017, 242; Crouch 2015, 89-93; France until his exile of 1210-1219/21 Holden et al. 2004, 177), and it has been (Duffy 2017, 296; Smith 2006), Walter suggested that the feature is primary, built was part of King John’s entourage in Nor- by FitzAnthony as a summit-vault in emula- mandy from September 1199 until March tion of his lord’s donjon at Pembroke (Dan 1201: he witnessed deeds at Rouen in Sep- Tietzsch-Tyler, pers. comm.). tember 1199 (Hardy 1837, 24) and at Caen The cylindrical donjon in Wales and and Falaise in June 1200 (Hardy 1837, 67, Ireland 69), while he presumably accompanied the Pembroke may be the earliest free-stand- king in Poitou and Touraine during this ing, fully-circular donjon in the British period. It is possible, then, that Dundrum Isles: the rest appear to be somewhat later, was instead, or additionally inspired by the enjoying their heyday in the 1220s-30s. Angevin circular donjons of these regions. However, the similar tower at Dundrum, Walter de Lacy also provides a link Co. Down, was begun by Hugh de Lacy between Ireland and the cylindrical don- between 1205 and 1210, to celebrate his jons of southeast Wales and the Marches, elevation to the earldom of Ulster; minor of which one of the earliest may be his works by King John in 1210-11 represent tower at Longtown Castle, Herefs., almost its completion (McDonald 2017, 270-1, certainly built between his return from 276; McNeill 2003, 98, 105). Though exile in 1213 and his departure for Ireland smaller overall, and comprising just three in 1224 (Duffy 2017, 316; Hislop 2016, storeys, like Pembroke it is a free-standing 125):92 Walter was sheriff of Hereford cylinder without projections, on level between 1215 and 1223, spending most of ground, with a basal batter and a spiral stair this period in the region (Flanagan 2004). that climbs the full height of the tower; it The Longtown donjon comprises three also shows evidence for a timber porch, as storeys, like Dundrum, while structural at Pembroke (see Appendix 1; Guy 2015, evidence indicates that both were floored 50). There is however neither dome nor using radial cross-joists (McNeill 2003, external offsets, architectural detail is ab- 104). Unlike Dundrum, however, it occu- sent, windows appear to be single lights pies a motte. It also features three project- (while the basement is pierced by slit- ing, semicircular turrets, one of them

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16253 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design housing the spiral stair in a manner that ‘chemise’ wall like those around the tow- became emblematic of cylindrical donjons ers at Château-sur-Epte, Neaufles, and in the region. Though absent from Dun- many others including Condé-sur-Noireau drum and Pembroke, a semicircular stair- and Fréteval where, as at Lyonshall, there turret had been employed in Angevin is no motte. France in the cylindrical donjon, probably In many cylindrical donjons nearby, for from 1180-1200, at Le Châtelier (Corvisier example Caldicot and Skenfrith (both 1998(2), 185-8; Fig. 32) the solid turrets Mon.), the semi-circular stair turret occurs recall those at Conisbrough, Loches and alongside a second feature that is equally Niort (see above), and, perhaps, the semi- characteristic of the region, but absent from circular pilasters in the round tower at Con- Angevin France, Ireland, and the Pembroke dé-sur-Noireau (Calvados) in Normandy, donjon: a roll-moulded basal string-course built under Angevin rule 1170-89 (Corvisi- (Guy 2009, 23, 82-107; Renn 1961, 132). er 1998(2), 256-62).93 While Lacy influ- There is also a string-course at Lyonshall ence may lie behind the wider adoption of but it is simply-chamfered (Renn 1961, the semicircular stair turret in Irish cylin- 141), perhaps confirming that the donjon is drical towers of the mid thirteenth-century, early and, along with Longtown, a primary e.g. Kiltinane, Co. Tipperary (see McNeill influence in the region.95 Nevertheless, Cal- 1997, 94-5), it may already have been em- dicot and Skenfrith were built by patrons ployed by King John in the tour-mixte at with extensive experience of Normandy, Ardfinnan (also Co. Tipperary);94 the simi- Touraine and Poitou, and possibly received lar turret against the circular northeast tow- at least some inspiration directly from An- er at Helmsley Castle, built 1191-1227 by gevin examples: Caldicot was possibly be- Robert de Roos, was probably also influ- gun shortly after Longtown by Henry de enced by royal precedents (see above; Ke- Bohun (d. 1220), who had enjoyed a cross- nyon 2017, 23). channel lifestyle until the loss of Normandy Lyonshall Castle, (Herefs.), near Long- in 1204 (Powicke 1913, 487-8) – although town, was held of Walter de Lacy’s honor it may be later – while Hubert de Burgh, of Weobley. It features the scant remains of builder at Skenfrith, was a senior Angevin a free-standing cylindrical donjon (Guy commander in France and former sene- 96 2017, 119-21; Renn 1961, 141), which is schal of Poitou. without a turret and may therefore be Influence from Pembroke’s donjon was, in among the earliest in the region. Its tenant west Wales, mainly in matters of detail, lord, and builder, was the Marshals’ house- discussed above; surprisingly, perhaps, no hold knight and kinsman Stephen other cylindrical donjon is known in its d’Evreux, who is recorded in their affinity immediate locality, although its visual pres- from 1199 until his death in 1228, and who ence is evoked in the towers at Younger represented the elder Marshal, in Leinster, Marshal Cilgerran, to which homage was during the Lacy alliance of 1207-8 (Holden paid at Manorbier and Laugharne. The two et al. 2004, 175; 2006, 153). It is difficult circular donjons of west Wales, at Dinefwr to imagine that it was conceived indepen- and Dryslwyn, Carms. – both c.1230s (Rees dently of the Pembroke donjon, but direct and Caple 2007, 10) – show basal string- influence from Angevin France appears to courses and were clearly influenced from lie behind the presence of a surrounding southeast Wales. The base of a very small

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16254 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design round tower on the motte at Nevern Castle, We have seen that solid, circular turrets Pembs., is probably too small to be a func- appear at Henry II’s Niort and, also in tional donjon (nine metres in diameter) the Angevin domain, the donjon at Châ- while its masonry was bonded with clay teau Romefort, Ciron (Indre, in Tou- (Caple 2016, 387); it may represent a purely raine), from c.1195-1200 (Mesqui 2013, symbolic tower, possibly uninhabitable and 207). Hollow circular towers were used not closely related to the Pembroke donjon, at Vic-sur-Aisne (Aisne, in Picardy), of the kind discussed by Pamela Marshall which was probably built during the (Marshall 2002b, 27, 34). While the excava- 1190s (Rolland 1984, 143), but there tor favours a date before 1195 for all mason- were only two, at one end of the donjon. ry at the castle (Caple 2016, 384), The fully-developed plan, with four compelling arguments have been made that hollow corner towers as in Ireland, the tower was an addition of 1204-15 (Wiles 97 seems first to appear at Nemours (Seine- 2014, 189; also see Turvey 1989, 64). et-Marne), built by the Capetians’ high The remaining cylindrical great towers of chamberlain around 1200 (Mesqui 2013, the British Isles appear to be from the 128 and fig., 207; Fig. 33). Given its 1220s-40s, or later, and similarly result date, and its location in the Île-de- from secondary, internal influences (see, France south of Paris, it cannot have inter alia, Goodall 2011, 181; Renn 1961, been seen by William Marshal I (or his 135). In Ireland many are engaged as masons) before – nor probably during – tours-mixtes or chamber-towers (McNeill his French visit of spring 1216, after 1997, 95), probably influenced by Dublin which work on his Irish castles is in any and Kilkenny, e.g. Nenagh and Kiltinane, case unlikely (see above).98 both Co. Tipperary (see McNeill 1997, 31, Richard Marshal however appears to 88, 94; Murtagh 2017, 162-75), while the have visited Nemours: he was retained, Lacy’s free-standing tower at Clogh Ough- from late 1217 until mid-1219, in the ter, Co. Cavan, follows their lead at Dun- household of King Philip who is known drum. But we have seen above that to have stayed at the castle during this Mocollop may have been directly influ- period (Crouch 2015, 23, 190-3; Power enced from Pembroke. 2003, 210, 213-6; Samaran and Nortier Philippienne influence: the younger 1979, 127-9).99 Richard succeeded to his Marshals father’s Norman lands in late summer The ‘towered keeps’ of Ireland, concen- 1219, spending most of his time there trated in Leinster (see Leask 1964, 143; until 1231. Yet he maintained regular McNeill 1997, 118-24; Sweetman 1999, contact with his brother William Mar- 39, 60; et al.), are of a form unique in the shal II during the thaw in Anglo-French British Isles until the fourteenth century, relations of the early 1220s, when peri- when a similar plan was adopted at e.g. odic visits by Richard, to both Britain Dudley (Worcs.) and Nunney (Somerset). and Ireland, are suggested in the sources (Holden et al. 2006, 32; Power 2003, In the absence of any indigenous anteced- 100 ents in Ireland, it is thought that their 225 n.). origins lie in similar towers built in France (see O’Keeffe 1990, 22, et al.).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16255 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

The towered keeps at Ferns (Co. Wexford) 42-3, 162); Richard is specifically record- and Lea (Co. Laois) are now thought to be ed at Montargis after his father’s death in from the mid-thirteenth century (Dempsey July 1219 (Holden et al. 2006, 194). In 2017, 247; McNeill 1997, 144; O’Keeffe addition, William Marshal II and Richard and Coughlan 2003, 144-8).101 That at Car- together visited King Philip in June 1220, low is suggested to have been begun by the spending time at Paris and Melun (Crouch elder Marshal, 1210-15 (O’Conor 1997, 15- 2015, 252-4, 290-91; Stapleton 1844, 16; Tietzsch-Tyler 2018, 121; Fig. 33), but cxxxviii), after which Richard maintained the evidence presented above, taken with its his close associations with the French cruciform loops, argues that the towered court (Crouch 2015, 23; Power 2003, 213- keep concept is more likely to have been 16). William’s continued contact with his introduced to Leinster, ultimately from brother, and with France, is manifest in his France, by his sons during the 1220s.102 It recruitment in 1224 of an author, from may be significant that William Marshal II Touraine or Anjou, to compose the bio- was quit of a years’ service to the king in graphical Histoire (Holden et al. 2006, both 1222 and 1225 ‘in aid of the fortifica- 22).104 tion of one of his castles in Ireland’ (Hardy Philippienne influence may also be ap- 103 1844a, 32; Sweetman 1875, 158, 192), parent in Pembroke’s Dungeon Tower, while the town of Carlow received its first which is clearly secondary to the elder charter from the earl in 1223 (Ballard and Marshal’s inner curtain of c.1204-7 (King Tait 1923, xxxv, 16, 53). 1978, 105, 118).105 Neil Guy has observed The younger Marshals may provide a that the exterior shows three deep sockets, context for other Philippienne attributes following a spiral incline like the helical seen in Ireland in the 1220s, such as the scaffold sockets seen to great effect in clockwise/counter-clockwise stair system Edward I’s North Wales castles (Guy in the donjon at Mocollop, which has prec- 2016b, 30 and appendix; Fig. 34). Helical edents at e.g. the Tour du Coudray at Chi- scaffolding appears to have been first used non (O’Keeffe and Whelan forthcoming); in King Philip’s castles, c.1200-1223 (in- the same French influence is implicit in the cluding Dourdan, Rouen, Villeneuve-sur- cylindrical lighthouse tower at Hook Point, Yonne and Verneuil), but afterwards was Co. Wexford, begun by Walter Marshal in chiefly confined to Savoy and North 1241-5, which is rib-vaulted at every level Wales (Hislop 2016, 60, 119). While it is like a Tour Philippienne (Murtagh 2016, possible that the Dungeon Tower (Fig. 34) 230; see Appendix 2). While he was with was added by the elder Marshal, following Philip Augustus 1217-19, Richard Marshal his visit to King Philip in spring 1216 will have seen a number of Capetian cas- (when the sources may hint at attention to tles: the king was frequently at both the west Wales: Hardy 1833, 228, 271), we Louvre and Gisors, as well as Péronne have seen that his movements do not ap- (Somme), Vernon (Eure) and Montargis pear to have taken in any Philippienne (Samaran and Nortier 1979, 126-7, 185-6, castles. The Dungeon Tower shows many 240-6), all castles with Philippienne plan- Marshal attributes including rounded arch- ning and/or Tours Philippiennes, and all es, entasis, a summit vault and narrow- thought to have been underway by 1217 splayed loops – some of which are cruci- (Mesqui 1997, 457, 464; Mesqui 2013, form – but all appear in Younger Marshal

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16256 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design

Fig. 34 – The Dungeon Tower at Pembroke Castle. The three inclined sockets can be clearly seen.

Fig. 35 – Sketch elevation of the donjon and Dungeon Tower at Pembroke Castle, from southeast, showing relative levels.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16257 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design work elsewhere while, in contrast to the (Crouch 2015, 20), perhaps significantly donjon, dressed stone is more-or-less ab- as the earliest use of helical scaffolding sent. Externally, it is confined to two basal otherwise known in Britain is in the cylin- oillets, one rectangular, the other circular, drical donjon at Llywelyn’s Dolbadarn and while both are probably medieval – the Castle, Caerns., normally assigned to the latter is shown in pre-restoration photo- 1230s (Goodall 2011, 227); it has hitherto graphs – they may be secondary insertions been regarded as an isolated instance until (see above); the suspicion may be borne out the Edwardian castle-building of the by their anomalous oolite dressings. As 1270s-80s.106 But Marshal influence at noted above, the tower’s affinities appear to Dolbadarn is in no sense a given and, lie more closely with younger Marshal while it is difficult to envisage another work at e.g. Cilgerran and Chepstow. The conduit for Llywelyn’s use of such scaf- simple, annular ‘capital’ that terminates the folding, his personal contact with the earls spiral-stair newel is also in a style more was limited (Walker 1994, 43-4); there are characteristic of the 1220s-30s and may be moreover few other notably ‘Marshal’ at- the work of a mason with experience of the tributes in the Dolbadarn donjon, apart West Country School, in which similar ter- from its circular shape.107 minal capitals appear in imposts and verti- cal roll-mouldings (Brakspear 1931, 8-9); Conclusion the school retained its influence in south- Almost all features of William Marshal I’s west England, south Wales and the west castles have precedents in works of the Midlands into the 1240s, and lies behind the English Crown. Marshal was unusual detail in younger Marshal work at Chep- among major barons in that he owed ev- stow, chiefly by Gilbert Marshal 1234-41 erything to his royal patrons (later, the (Coldstream and Morris 2006, 103, 112). same could be said of Hubert de Burgh), Of the younger Marshals, William II may and he had no inheritance of his own until be the most likely builder of the Dungeon his brother died in 1194. So it should Tower, under French influence via his come as no surprise that Marshal chose to brother Richard; the latter’s activities as emulate Angevin buildings: it can be ar- earl, 1231-4, suggest he was not personally gued that his references to them were de- responsible. William appears to have been liberate, and intended to be understood. in Pembrokeshire in 1221 (Jones 1952, 98), Marshal’s emulation may also have been and had been commanded to ‘repair’ the another expression of the loyalty that he castles and Narberth and Wiston (Pembs.) was so keen on proclaiming, and was per- in October 1220, following Welsh attacks haps facilitated by the loan of royal ma- (Pat. Rolls 1216-25, 254-5). He was in sons. His debt to the French Crown was of western Wales, on campaign, from April- lesser magnitude and, politically, some- October 1223 (Crouch 2015, 18-19, 285-9; thing to be played down. While other in- Edwards 1935, 3-4, 30), and again in Au- fluences may be present in his buildings, gust 1225, when there is some suggestion in there is no evidence that he drew any the sources that he was strengthening his direct inspiration from Philip Augustus’s castles (Crouch 2015, 195-6; Hardy 1833, works, nor of a mechanism through which 59, 79). William headed a delegation to it could be transmitted. Philippienne work meet Llywelyn ap Iorwerth later in that year has little in common with the earl’s and is

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16258 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design mainly later than 1200, a period when hos- more frequent during the later Middle Ag- tilities prevented constructive Anglo- es, show that patrons could, from an early French contact. Neither was Philip a great date, be responsible (at least in part) for builder of donjons, generally preferring the overall design, as suggested of Richard I at defensive tour-mixte. Château Gaillard (Hislop 2016, 148; Mes- The evidence discussed above suggests that qui 1993, 334, et al.; also see Hulme 2014, Marshal may not have begun any serious 230). The Pembroke team seems to have building in the British Isles until after he was included free-masons from the West Coun- created earl in 1199; it is only then that he try, stylistic evidence for which is most reveals an interest that, hitherto, had largely apparent in the lower two storeys; after been confined to his estates in Normandy 1204, at least, masons who had personally where he was resident during the later 1190s. worked in Angevin France appear also to It is possible that the Pembroke donjon, be- have been employed and their influence is gun in 1201, was his first major British con- felt in the upper storeys of the donjon, struction, marking his elevation to the including the windows and dome. They earldom at its titular caput; his main castle at were also responsible for the similar win- Chepstow already had an impressive great dow in the Wogan cavern. tower, built in the eleventh century. Marshal’s work elsewhere, which may Both architecturally and functionally, the largely belong to the post-1204 period, sim- Pembroke donjon belongs to an Angevin ilarly followed Angevin practice. It was not tradition well-established in northern until his sons succeeded in 1219, with their France. Marshal presumably engaged his close associations with the Philippienne Pembroke team in late 1200 or early 1201. court, that French influence began to be felt Full-scale war the following year may have in Marshal design. However, it remained placed a demand on experienced masons, for secondary to their established practices. Normandy, but there is no evidence for a A wider debate concerns the extent of hiatus in the lower stages of the Pembroke Capetian influence on the Angevin kings, donjon. Work similarly continued uninter- as well as their barons. The Angevin rupted at King John’s Corfe Castle where a architectural tradition was very broad- major campaign took place from 1201 to based (see Gillingham 2001, 118) and, in 1204, including the Gloriette in which we common with much of western Europe, begin to see an identifiably ‘English’ form of reveals the dominating influence of the early Gothic (Goodall 2011, 159, et al.), French Gothic aesthetic. But Capetian emulated by Marshal at Chepstow (see influence must not be taken as a given. above). The Pembroke master also refer- The two realms were subject to the same enced Angevin precedents and may even trends in military design, and discerning have been instructed, by Marshal, to follow specific patterns of influence is not always a particular template, cf. the contract from possible – while, as French authors are 1224 in which Count Robert III of Dreux quick to point out, the current could de- instructs his master mason to build a donjon monstrably flow the other way. It has been specifically like the one at Nogent-le- seen above that many aspects of Angevin Rotrou, Eure-et-Loir (reproduced in Jost design originated independently of 2002, 182-3). Such contracts, which became French precedent.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16259 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Acknowledgements Hardy, T. D. (ed.), 1835b Rotuli Norman- Many thanks to David Crouch, the late Philip niae, Johanne et Henrico Quinto, Angliae Davis, Neil Guy, Malcolm Hislop, Jacques le Regibus, 1 (London: Record Commission). Maho, Jean Mesqui, Denys Pringle, Dan Hardy, T. D. (ed.), 1837 Rotuli Chartarum Tietzsch-Tyler and the late Rick Turner for 1/1, 1199-1216 (London: Record Com- information, images and discussion. mission). References Hardy, T. D. (ed.), 1844a Rotuli Litterar- Primary sources (published) um Clausarum 2, 1224-1227 (London: Record Commission). Ballard, A. and Tait, J. (eds), 1923 British Borough Charters 1216-1307 (Cambridge Hardy, T. D. (ed.), 1844b Rotuli de Liber- University Press). ate ac de Misis et Praestitis, regnante Johanne (London: Record Commission). Brewer, J. S. (ed.), 1863 Giraldi Cambren- sis Opera, 3: ‘De Invectionibus’, ‘De Me- Holden, A. J., Gregory S. and Crouch, D. nevensi Ecclesia Dialogus’ and ‘Vita S. (eds), 2002 History of William Marshal, 1 David’ (London: Rolls Series). (London: Anglo-Norman Text Society 4). Calendar of Charter Rolls 1, Hen. III 1226- Holden, A. J., Gregory S. and Crouch, D. 1257 (London: HMSO, 1908). (eds), 2004 History of William Marshal, 2 (London: Anglo-Norman Text Society 5). Calendar of Charter Rolls 2, Hen. III 1257- 1300 (London: HMSO, 1906) Holden, A. J., Gregory S. and Crouch, D. (eds), 2006 History of William Marshal, 3 Caley, J., Ellis, H. and Bandinel, B. (eds), (London: Anglo-Norman Text Society 6). 1846 Monasticon Anglicanum, by Sir Wil- liam Dugdale, 4 (London: Record Commis- Howlett, R. (ed.), 1889 ‘The Chronicle of sion). Robert of Torigni’, in Chronicles, Ste- phen, Henry II and Richard I, 4 (London: Crouch, D. (ed.), 2015 The Acts and Letters Rolls Series). of the Marshal Family: Marshals of England and Earls of Pembroke, 1145-1248 (Cam- Huygens, R. B. C. (ed.), 1976 Monumenta bridge University Press: Camden 5/47). Vizeliacensia: textes relatifs à l’histoire de l’abbaye de Vézelay (Turnhout: Davies, J. C. (ed.), 1946 Episcopal Acts Brepols). relating to the Welsh Dioceses 1066-1272, 1 (Cardiff: Hist. Soc. Church in Wales 1). Jones, T. (ed.), 1952 Brut y Tywysogyon: Peniarth MS. 20 Version (Cardiff: Univer- Edwards, J. G. (ed.), 1935 Calendar of sity of Wales Press). Ancient Correspondence concerning Wales (Cardiff: University of Wales Press). Landon, L., 1935 The Itinerary of King Richard I, with studies on certain matters Hardy, T. D. (ed.), 1833 Rotuli Litterarum of interest connected with his reign (Lon- Clausarum 1, 1204-1224 (London: Record don: Pipe Roll Society 51). Commission). Packard, S. R. (ed.), 1927 Miscellaneous Hardy, T. D. (ed.), 1835a Rotuli Litterarum Records of the Norman Exchequer, 1199- Patentium 1/1, 1201-1216 (London: Re- 1204 (Northampton, Mass.: Smith Col- cord Commission). lege Studies in History 12, 1-4).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16260 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Patent Rolls, Hen. III 1216-1225 (London: Secondary sources (published) HMSO, 1901). Ashbee, J., 2009 Goodrich Castle (Lon- Patent Rolls, Hen. III 1225-1232 (London: don: English Heritage). HMSO, 1903). Avent, R., 1995 Laugharne Castle (Car- Round, J. H. (ed.), 1899 Calendar of Docu- diff: Cadw). ments Preserved in France 1, AD 918-1206 Avent, R., 2003 ‘William Marshal’s (London: HMSO). building works at Chepstow Castle, Mon- Samaran, C. and Nortier, M. (eds.), 1979 mouthshire, 11891219’, in J. R. Kenyon Recueil des actes de Philippe Auguste Roi and K. O’Conor (eds), 50-71. de France, 4, 1 Novembre 1215 – 14 Juillet Avent, R., 2006 ‘William Marshal’s castle 1223 (Paris: Imprimerie Nationale). at Chepstow and its place in military ar- Stapleton, T. (ed.), 1840 Magni Rotuli chitecture’, in R. Turner and A. Johnson Scaccarii Normanniae 1 (London: Soc. of (eds), 81-90. Antiquaries). Avent, R. and Miles, D., 2006 ‘The Main Stapleton, T. (ed.), 1844 Magni Rotuli Gatehouse’, in R. Turner and A. Johnson Scaccarii Normanniae 2 (London: Soc. of (eds), 51-62. Antiquaries). Avent, R. and Turner, R., 2006 ‘The Mid- Sweetman, H. S. (ed.), 1875 Calendar of dle Bailey’, in R. Turner and A. Johnson Documents relating to Ireland, 1171-1251 (eds), 63-70. (London; HMSO). Baudry, M-P., 2002 ‘La politique de forti- Secondary sources (unpublished) fication des Plantagenêts en Poitou, 1154- Corvisier, C., 1998 ‘Les Grosses Tours de 1242’, in J. Gillingham (ed.), 43-69. plan circulaire ou centré en France avant Biller, T., 1998 Die Adelsburg in 1200: étude sur les antécédents de la poli- Deutschland: Entstehung - Gestalt - Be- tique castrale de Philippe Auguste’, 3 vols deutung (Munich: Deutscher Kunstverlag). (PhD thesis: University of Paris). Bony, J. (ed.), 1980 The Art of the West in Day, A. and Ludlow, N., 2016 ‘Pembroke the Middle Ages, by Henri Focillon, 1: Castle Geophysical Survey 2016’ (DAT Romanesque Art (Oxford: Phaidon). Archaeological Services for Castle Studies Boüard, M. de, 1966 ‘Fouilles au château Trust – see: de Bonneville-sur-Touques’, Annales de http:/castlestudiestrust.org/docs/Pembroke_Cas tle_ Geophysical%20_Survey_FINAL.pdf). Normandie 16/4, 351-78. Guy, N. (ed.), 2016b ‘Summary notes for Bradley, J. and Murtagh, B., 2017 ‘Wil- CSG study day at Pembroke’ (unpublished liam Marshal’s charter to Kilkenny, 1207: handout). background, dating and witnesses’, in J. Bradley, C. Ó Drisceoil and M. Potterton Launay, A., 2015 ‘‘Qui a bâti les fortifica- (eds), 201-48. tions de Pont-de-l’Arche?’ (http://pontdelarche.overblog.com/article-qui-a- Bradley, J., Ó Drisceoil, C. and Potterton, bati-les-fortifications-de-pont-de-l-arche- M. (eds), 2017 William Marshal and Ire- 108299939.html). land (Dublin: Four Courts Press).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16261 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Brakspear, H., 1931 ‘A West Country capétien et archétype des tours-maitresses School of Masons’, Archaeologia 81, 1-18. de Philippe Auguste’, Bulletin de la So- Brindle, S., 2015 Dover Castle (London: ciété Historique de Compiègne 38, 9-45. English Heritage). Crouch, D., 2016 William Marshal: Third Brindle, S. and Sadraei, A., 2015 Conis- Edition (Abingdon: Routledge). brough Castle (London: English Heritage). Curnow, P. E., 1980 ‘Some Develop- Brown, R. A., 1976 English Castles (Lon- ments in Military Architecture c.1200: Le don: Batsford). Coudray-Salbart’, in Brown, R. A. (ed.), Proceedings of the Battle Conference on Caple, C., 2016 ‘Nevern Castle 2008 – Anglo-Norman Studies 2, 1979 (Wood- 2015: closing in on the first Welsh stone bridge: Boydell), 42-62. castle?’, Medieval Archaeol. 60/2, 382-91. Davies, J. A., Riley, A., Levesque, J-M. Chapman, A., 2007 ‘The gatehouse of Pe- and Lapiche, C. (eds), Castles and the vensey Castle’, Sussex Archaeol. Collec- Anglo-Norman World (Oxford and Phila- tions 145 (2007), 97–118. delphia: Oxbow Books). Châtelain, A., 1983 Châteaux Forts et Féo- Davis, P., 2016 ‘Interpretations of the eme eme dalité en Île de France du XI au XIII roof and upper works in the great tower of siècle (Nonette: Éditions Créer). Conisbrough Castle’, Castle Studies Châtelain, A., 1991 ‘Recherche sur les châ- Group Journ. 29, 250-62. teaux de Philippe Auguste’, Archéologie Dempsey, K., 2017 ‘Lea Castle, Co. Lao- Médiévale 21, 115-61. is: the story so far’, Castles Studies Group Chollet, S. and Gousset, J-M., 2012 'Laval: Journ. 30, 237-52. nouvelle datation dendrochronologique de Deschamps, P., 1972 ‘Forteresses de la la tour maîtresse du château et de son France médiévale’, Journal des Savants, hourd', Bulletin Monumental 170/3, 261-2. Année 1971/2, 106-18. Church, S. D., 2015 King John: England, Deyres, M. M., 1984 ‘Le donjon de Magna Carta and the Making of a Tyrant Châtillon-sur-Indre’, Bulletin trimestriel (London: McMillan). de la Societé Archéologique de Touraine Coldstream, N. and Morris, R. K., 2006 36, 364-77. ‘The Architecture and Decoration of the Dixon, P., 2002 ‘The myth of the keep’, Marshals’ Great Tower’, in R. Turner and in G. Meirion-Jones, E. Impey and M. A. Johnson (eds), 101-12. Jones (eds), 913. Colfer, B., 2013 Wexford Castles: Land- Dufaÿ, B., 2011 ‘Un nouveau regard sur scape, Context and Settlement (Cork Uni- la forteresse de Chinon: 1er bilan de 7 versity Press). années de fouilles archéologiques’, in Les Colvin, H. M. (ed.), 1963 A History of the Amis du Vieux Chinon (eds), Chinon, Le King’s Works 2: The Middle Ages (London: Destin d’une Forteresse (Chinon: Les HMSO). Amis du Vieux Chinon), 85-103. Corvisier, C., 2002 ‘La Grosse Tour du Roi Duffy, P., 2017 ‘From Carcassonne to à Compiègne: premier donjon circulaire Carrickfergus: The Legacy of de Lacy’s

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16262 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Crusade experience in Britain and Ireland’, Goodall, J., 2015 Warkworth Castle (Lon- in P. Duffy, T. O’Keeffe and J-M. Picard don: English Heritage). (eds), 295-328. Grant, L., 1994, ‘Le patronage architectur- Duffy, P., O’Keeffe, T. and Picard, J-M. al d’Henri II et de son entourage’, Cahiers (eds), 2017 From Carrickfergus to Carcas- de Civilisation Médiévale 37 (145-6), 73- sonne: The Epic Deeds of Hugh de Lacy 84. during the Albigensian Crusade (Turn- Guy, N. (ed.), 2009 ‘Castle Studies Group hout: Brepols). Conference, Abergavenny’, Castle Studies Emery, A., 2006 Greater Medieval Houses Group Journ. 22, 4-127. of England and Wales, 1300–1500: Vol. 3, Guy, N., 2013a ‘Conisbrough and Morte- Southern England (Cambridge University mer Castles’, Castle Studies Group Journ. Press). 26, 203-6. Erlande-Brandenburg, A., 1970 ‘Le donjon Guy, N. (ed.), 2013b ‘Castle Studies de Châteaudun’, Bulletin Monumental Group Conference, ‘Castles of Co. Dur- 128/2, 141-2. ham’’, Castle Studies Group Journ. 26, Fichet de Clairfontaine, F., Mastrolorenzo, 4-150. J. and Brown, R., 2016 ‘Le château de Guy, N. (ed.), 2015 ‘Castle Studies Group Falaise (Calvados): état des connaissances Conference, ‘Castles of Northern Ireland”, sur l’evolution du site castral du dixième Castle Studies Group Journ. 28, 4-119. siècle au treizième siècle’, in J. A. Davies, A. Riley, J-M. Levesque and C. Lapiche Guy, N., 2016a ‘The portcullis – design (eds), 231-55. and development – Chepstow Outer Gate’, Castle Studies Group Journ. 29, 192-201. Flanagan, M. T., 2004 ‘Lacy, Walter de (d. 1241), magnate’, Oxford Dictionary of Na- Guy, N. (ed.), 2017 ‘Castle Studies Group tional Biography (Oxford University Press: Conference, ‘Castles of the Welsh March- accessed 30 March 2018). es’’, Castle Studies Group Journ. 30, 4- 150. Fleury, M. and Kruta, V., 2000 Le Château du Louvre: guide et monographie (Dijon: Guy, N. (ed.), 2018 Castles: History, Ar- Éditions Faton). chaeology, Landscape, Architecture and Symbolism (Daventry: Castle Studies Gillingham, J., 2001 The Angevin Empire Group). (London: Hodder Arnold). Héliot, P., and Deyres, M., 1987 ‘Le châ- Gillingham, J. (ed.), Anglo-Norman Stud- teau de Loches’, Bulletin Monumental ies 24: Proceedings of the Battle Confer- 145/1, 15-85. ence, 2001 (Woodbridge: Boydell). Heslop, T. A., 2003 'Orford Castle: nostal- Goodall, J., 2001 Richmond Castle/St Ag- gia and sophisticated living', in R. Liddi- atha’s Abbey, Easby (London: English ard (ed.), Anglo-Norman Castles Heritage). (Woodbridge: Boydell), 273-296. Goodall, J., 2011 The English Castle 1066- Hilling, J., 2000 Cilgerran Castle/St Dog- 1650 (New Haven and London: Yale Uni- maels Abbey (Cardiff: Cadw). versity Press).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16263 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Hislop, M., 2016 Castle Builders: Ap- King, D. J. C., 1978 ‘Pembroke Castle’, proaches to Castle Design and Construc- Archaeol. Cambrensis 127, 75–121. tion in the Middle Ages (Barnsley: Pen & King, D. J. C., 1983 Castellarium Angli- Sword). canum (New York: Kraus International). Hislop, M., 2018 ‘The Round Tower of King, D. J. C., and Perks, J. C., 1970 Barnard Castle’, in N. Guy (ed.), 81-104. ‘Manorbier Castle, Pembrokeshire’, Ar- Hodkinson, B., 2003 ‘A summary of recent chaeol. Cambrensis 119, 83-118. work at the Rock of Dunamase, Co. Laois’, Knight, J. K., 1977 ‘Usk Castle and its in J. R. Kenyon and K. O’Conor (eds), affinities’, in M. R. Apted, R. Gilyard- 32-49. Beer and A. D. Saunders (eds), Ancient Hourihane, C. P. (ed.), 2012, The Grove Monuments and their Interpretation: Es- Encyclopedia of Medieval Art and Archi- says presented to A. J. Taylor (Chichester: tecture, 2 (Oxford University Press). Phillimore), 139-54. Hulme, R., 2014 ‘The Impact of Château Knight, J. K., 1993 ‘Excavations at Mont- Gaillard and Crusading on English Castle gomery Castle – Part 1’, Archaeol. Camb- Architecture’, Castle Studies Group Journ. rensis 141, 97180 27, 203-33. Knight, J. K., 2008 ‘Welsh Space and Impey, E. and Parnell, G., 2011 The Tower Norman Invaders: Usk Castle 1136-1245’, of London: The Official Illustrated History in J. K. Knight and A. Johnson (eds), Usk (London and New York: Merrell). Castle, Priory and Town (Almeley: Lo- Jost, B., 2002 ‘Meunzenberg (Hesse) and gaston), 55-68. its relationship to later twelfth-century cas- Knight, J. K., 2009 The Three Castles/Hen tles’, in G. Meirion-Jones, E. Impey and M. Gwrt Medieval Moated Site (Cardiff: Jones (eds), 179-87. Cadw). Kennedy, H., 1994 Crusader Castles Knight, J. K., 2018 ‘Fog in the Channel? (Cambridge University Press). Influences in Anglo-French castle build- Kenyon, J. R., 1990 Medieval Fortifica- ing’, in N. Guy (ed.), 155-73. tions (Leicester University Press). Langeuin, P., 2002 ‘Les campagnes de construction du château d'Arques-la-Ba- Kenyon, J. R., 2007 Kidwelly Castle (Car- e e diff: Cadw). taille (XI -XV siècles)’, Bulletin Monu- mental 160/4, 345-378. Kenyon, J. R., 2017 Helmsley Castle (Lon- don: English Heritage). Leask, H. G., 1964 Irish Castles and Cas- tellated Houses (Dundalk: Dundalgan Kenyon J. R. and O’Conor, K. (eds), 2003 Press). The Medieval Castle in Ireland and Wales: Essays in honour of Jeremy Knight (Dub- Lloyd, T., Orbach, J. and Scourfield, R., lin: Four Courts Press). 2004 The Buildings of Wales: Pembroke- shire (New Haven and London: Yale Uni- King, D. J. C., 1977 ‘Pembroke Castle: versity Press). Derivations and relationships of the domed vault of the donjon, and of the Horseshoe Ludlow, N., 1991 ‘Pembroke Castle and Gate’ Château Gaillard 8, 159-69. Town Walls’, Fortress 8, 25-30.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16264 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Ludlow, N., 2003 ‘The Castle and Lord- en hommage à André Matthys (Namur: ship of Narberth’, Journ. Pembs. Hist. Soc. Les Cahiers de l’Urbanisme), 122-7. 12, 5-43. Meirion-Jones, G., Impey E. and Jones, Ludlow, N., 2014 Carmarthen Castle: the M. (eds), 2002 The Seignurial Residence Archaeology of Government (Cardiff: Uni- in Western Europe AD c 800-1600 (Ox- versity of Wales Press). ford: Archaeopress; British Archaeologi- Ludlow, N., 2018 ‘Bothwell: a Welsh cal Reports International Series S1088). Marches castle in Scotland?’, in N. Guy Mesqui, J., 1993 Châteaux et Enceintes de (ed.), 236-80. la France Médiévale 2: la Résidence et les Manning, C., 2003 ‘The Record Tower, Éléments d’Architecture (Paris: Éditions Dublin Castle’, in J. R. Kenyon and K. Picard). O’Conor (eds), 7295. Mesqui, J., 1997 Châteaux forts et fortifi- Manning, C., 2013 Clogh Oughter Castle, cations en France (Paris: Flammarion). Co. Cavan: Archaeology, History and Ar- Mesqui, J., 2011 ‘La Tour des Archives et chitecture (Dublin: Stationery Office). le Fort des Tourelles de Vernon (Eure)’, Marshall, P., 2002a ‘The Ceremonial Bulletin Monumental 169/4, 291-318. Function of the Donjon in the Twelfth Cen- Mesqui, J., 2013 Châteaux et Enceintes de tury’, Château Gaillard 20, 141-51. la France Médiévale 1: les Organes de la Marshall, P., 2002b ‘The Great Tower as Défense (Paris: Éditions Picard). residence in the territories of the Norman Mesqui, J., 2018 ‘Philippe Auguste and Angevin kings of England’, in G. ‘tours”, in N. Guy (ed.), 129-139. Meirion-Jones, E. Impey and M. Jones Morris, R. K., 2010 Kenilworth Castle (eds), 27-44. (London: English Heritage). McDonald, P., 2017 ‘Identifying Hugh II Murtagh, B., 2016 ‘The medieval tower of de Lacy’s Contribution to Dundrum Castle Hook, Co. Wexford: A beacon of com- (Co. Down)’, in P. Duffy, T. O’Keeffe and merce and symbol of Marshal power and J-M. Picard (eds), 263-76. wealth in Ireland’, Château Gaillard 27, McNeill, T. E., 1990 ‘Trim Castle, Co. 223-236. Meath: the first three generations’, Archae- Murtagh, B., 2017 ‘William Marshal’s ol. Journ. 147, 308-36. Great Tower at Pembroke, Wales: a view McNeill, T., 1997 Castles in Ireland: Feu- from Ireland’, in J. Bradley, C. Ó Dris- dal Power in a Gaelic World (London and ceoil and M. Potterton (eds), 111-82. New York: Routledge). O’Conor, K., 1997, ‘The Origins of Car- McNeill, T. E., 2003 ‘Squaring circles: low Castle’, Archaeology Ireland 11/3, flooring round towers in Wales and Ire- 13-16. land’, in J. R. Kenyon and K. O’Conor O’Keeffe, T., 1990, ‘The Archaeology of (eds), 96-106. Norman Castles in Ireland 2: Stone Cas- McNeill, T. E., 2006 ‘The view from the tles’, Archaeology Ireland 4/4, 20-22. top’, in J. de Meulemeester (ed.), Mélanges d’archéologie médiévale: liber amicorum

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16265 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

O’Keeffe, T., 2009 ‘Dublin Castle’s donjon Pringle, D. (ed.), 1988 Crusader Castles in context’, in J. Bradley, A. J. Fletcher and by T. E. Lawrence (Oxford University A. Simms (eds), Dublin in the Medieval Press). World: Studies in Honour of Howard B. Pringle, D., 1989 ‘Crusader Castles: The Clarke (Dublin: Four Courts Press), 277-94. First Generation’, Fortress 1, 14-25. O’Keeffe, T., 2017 ‘Trim before 1224: Pringle, D., 2014a ‘Crusader Castles and New Thoughts on the caput of de Lacy Fortifications: The Armenian Connec- Lordship in Ireland’, in P. Duffy, T. tion’, in C. Mutafian (ed.), La Méditer- O’Keeffe and J-M. Picard (eds), 31-56. ranée des Arméniens, xiie–xve siècle (Paris: O’Keeffe, T. and Coughlan, M., 2003 ‘The Geuthner), 353–72. chronology and formal affinities of the Pringle, D., 2014b ‘Edward I, Castle- Ferns donjon, Co. Wexford’, in J. R. Ke- building and the Tower of the English in nyon and K. O’Conor (eds), 133-48. Acre’, in C. Donovan (ed.), A Fresh Ap- O’Keeffe, T. and Whelan, D., forthcoming proach: Essays Presented to Colin Platt in ‘A Little-Known Anglo-Norman Castle of Celebration of his Eightieth Birthday the First Rank: Mocollop, Co. Waterford’, (Bristol: Trouser Press), 48-56. in B. Murtagh (ed.), Castles and defences Radford, C. A. R., 1958 Goodrich Castle in Ireland and abroad: essays in honour of (London: HMSO). David Newman Johnson. Rees, S. and Caple, C., 2007 Dinefwr Page, W. (ed.), 1907 A History of the Castle/Dryslwyn Castle (Cardiff: Cadw). County of Leicestershire, 1 (London: Vic- toria County History). Remfry, P., 2011 ‘White Castle and the Dating of the Towers’, Castle Studies Painter, S., 1933 William Marshal: Knight- Group Journ. 24, 213-26. Errant, Baron, and Regent of England (Baltimore: Johns Hopkins Press). Renn, D. F., 1961 ‘The Round Keeps of the Brecon Region’, Archaeol. Cambren- Pitte, D., 2002 ‘Château-Gaillard dans la sis 110, 129-43. défense de la Normandie orientale (1196- 1204)’, in J. Gillingham (ed.), 163-75. Renn, D. F., 1968 ‘The Donjon at Pem- broke Castle’, Trans. Ancient Monument Pitte, D., 2007 ‘Arques-la-Bataille, sond- Soc., new series 15, 35-47. age archéologique dans le château’, Bulle- tin Monumental 165/1, 113-14 Renn, D., 1973 Norman Castles in Britain (London: John Baker). Power, D., 2003 ‘The French interests of the Marshal earls of Striguil and Pembroke, Rolland, D., 1984 ‘Le château et les châte- 11891234’, in J. Gillingham (ed.), Anglo- lains de Vic-sur-Aisne’, Fédération des Norman Studies 25: Proceedings of the Societés d’Histoire et d’Archéologie de Battle Conference, 2002 (Woodbridge: l’Aisne 29, 139-76. Boydell), 199-225. Runciman, S., 1951 A History of the Cru- Powicke, F. M., 1913 The Loss of Normandy sades 3: the Kingdom of Acre and the later (1189-1204): Studies in the History of the An- Crusades (Cambridge University Press). gevin Empire (Manchester University Press).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16266 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - References

Saunders, A. D., 2002 Launceston Castle Castle: A New Interpretation’, Journ. (London: English Heritage). Pembs. Historical Soc. 3, 5766. Shoesmith, R. (ed.), 2014 Goodrich Castle: Vincent, N., 2000 ‘William Marshal, King its History and Buildings (Almeley: Logas- Henry II and the Honour of Châteauroux’, ton). Archives: Journ. Brit. Record Assoc. Smith, B., 2006 ‘Lacy, Hugh de, earl of 25/102, 1-15. Ulster (d. 1242)’, Oxford Dictionary of Na- Walker, R. F., 1994 ‘The supporters of tional Biography (Oxford University Press: Richard Marshal, earl of Pembroke, in the accessed 30 March 2018). rebellion of 1233-1234’, Welsh History Smith, S. G., 2014 ‘Dolbadarn Castle, Caer- Review 17/1, 41-65. narfonshire: A Thirteenth-Century Royal Wiles, J., 2014, “Marshal towers’ in Landscape’, Archaeol. in Wales 53, 63-72. South-west Wales: Innovation, Emulation Sweetman, D., 1999 The Medieval Castles and Mimicry’, Castle Studies Group of Ireland (Woodbridge: Boydell). Journ. 27, 181-202. Thomas, W. G., 1993 ‘The Walls of Ten- by’, Archaeol. Cambrensis 142, 1-39. Thurlby, M., 2006 Romanesque Architec- ture and Sculpture in Wales (Almeley: Lo- gaston). Tietzsch-Tyler, D., 2013 ‘King John’s Cas- tle: Staged Development, Imperfect Real- ization’, North Munster Antiq. Journ. 53, 135-71. Tietzsch-Tyler, D., 2017 ‘William Mar- shal’s castle at Kilkenny in about 1395: a new reconstruction’, in J. Bradley, C. Ó Drisceoil and M. Potterton (eds), 183-200. Tietzsch-Tyler, D., 2018 ‘Innovative Castle Design on the Western Fringe of the An- gevin Empire’, in N. Guy (ed.), 105-28. Turner, R., 2006 ‘The Upper Bailey’, in R. Turner and A. Johnson (eds), 71-80. Turner, R. and Johnson, A. (eds), 2006 Chepstow Castle: its History and Buildings (Almeley: Logaston). Turner, R., Priestley, S., Coldstream, N. and Sale, B., 2006 ‘The ‘Gloriette’ in the Lower Bailey’, in R. Turner and A. Johnson (eds), 135-50. Turvey, R. K., 1989 ‘Nevern

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16267 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

Notes 5 ‘The wall in front of the tower’ is mentioned 1 Christian Corvisier has however questioned in the exchequer account of 1180 (Stapleton the archaeological evidence, and prefers a 1840, 73); the tower is unusual in showing date in the 1150s under the influence of the four large buttresses, much smaller however Counts of Blois; other authorities have pro- than the turrets employed in French donjons posed a date in the 1160s, under the Angevin (see below). Henry’s ‘tower’ at Lyons-la- king Henry II (Corvisier 1998(2), 318-19; Forêt (Eure, Normandy), also mentioned in Corvisier 2002, 44 n. 60). the 1180 account (ibid.), may similarly have been a cylindrical donjon (Corvisier 2 Those at Simiane-la-Rotonde (Alpes-de- 1998(2), 662-3). Haute-Provence) and Ottrott (Bas-Rhin, in 6 Alsace), lay beyond the borders of medieval When £95 was spent on ‘operacionibus tur- France, within the Empire. The latter are ris Bonavilla plancandis’ (Stapleton 1840, cylindrical ‘bergfriede’ like those built in 233). The account was rendered by Robert Germany from c.1160 onwards (Biller 1998, de Roos, lord of Helmsley Castle in York- 130-7; Jost 2002, 181-2; Mesqui 2013, 185; shire and hereditary custodian of Bonneville Renn 1961, 133, et al.). (Stapleton 1840, 142, 233-5). 7 3 The 1180 roll records the outlay of £13 on Corvisier and Michel de Boüard favour a date the ‘repair of houses and the tower [ie. don- under King John, 1199-1204, for the mural jon]’ at Neaufles, and £10 on the ‘repair of towers at Bonneville (Boüard 1966, 371; Cor- the houses, tower and gate’ at Château-sur- visier 1998(2), 62-3); Corvisier however con- Epte (Stapleton 1840, 72). In the 1184 ac- siders the curtain wall itself to belong to Henry count, £195 4s 8d was spent on ‘works on the II’s reign, though it is not clear how the donjon tower and houses, and heightening the wall might have been inserted. The latter may have and making the base of the wall’ at Neaufles, been complete by 1198 and is not mentioned in while works at Château-sur-Epte, costing the account roll of that year, which instead £301, were confined to heightening the che- records small sums spent on ‘repairing the mise around the donjon and work in the royal dwellings at Bonneville’ (Stapleton bailey (Stapleton 1840, 110). 1844, lxxv); the roll is an isolated survival, and only a few detached membranes survive from 4 The internal arcade at Château-sur-Epte was John’s reign (see Stapleton 1844, passim). purely decorative, and never intended to car- 8 ry vaulting (Châtelain 1983, 204). The large Although the latter may be thirteenth-centu- twin-light window is clearly a later insertion, ry (Corvisier 1998(2), 619-20). which interrupts the arcade while being sty- 9 The considerable expenditure at Pont-de- listically later, possibly early/mid thirteenth- l’Arche, recorded in the mid-1190s (see be- century (Corvisier 1998(2), 145; Mesqui low), represented a comprehensive rebuild. 1993, 196; also see Châtelain 1983, 204). It The castle, now gone, is known to have was probably inserted under French rule, included a large, free-standing round tower after 1204. The unusual circular oculi at which, like the rest of the castle, cannot be Neaufles show influence from ecclesiastical dated (Launay 2015; Fig. 18); the erection of building. The only other castle within which a ‘turret’ (tourelle) was commissioned by they have been recorded, as principal lights, King Philip in 1210 (Châtelain 1991, 115- is at Troyes (Aube), where they are probably 61), but the term is scarcely appropriate for mid-twelfth century (Corvisier 1998(2), the round tower which, moreover, is free- 490); they do however also occur in the gable standing within the bailey unlike the vast wall of the eleventh-century great tower at majority of his post-1200 ‘Tours Philippi- Chepstow, Mon. ennes’ (among which it is not normally in- cluded; see Mesqui 2013, passim).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16268 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

10 Marie-Pierre Baudry notes the Angevin char- Tietzsch-Tyler regards the ‘dominant tower’, acteristics of the tower at Montreuil-Bonnin in general, as having ‘taken the place of the . . (Baudry 2002, 57-9; also see Mesqui 1997, . keep’ as ‘essentially a public space’ (Tietz- 254-5); the rib-vault was added by King Philip, sch-Tyler 2017, 186). by whom – alone amongst the castles of Poitou 15 – it was held in 1206-10 (Baudry 2002, 46; The date of the Barnard chamber-tower is still Mesqui 1997, 254). The cylindrical tour-mixte a matter of debate. In a recent paper, Malcolm at Charlieu (Loire), is suggested to have been Hislop generally favoured David Austin’s built by Philip Augustus after his final acquisi- date in the 1190s. However, he noted that tion of the Auvergne from the Angevins in financial circumstances may instead argue a 1210 (Deschamps 1972, 111; also see Gilling- date-range of c. 1208-28 (Hislop 2018, 99) – ham 2001, 31, 89), but shows niched loops, an which is compatible with the stylistic evi- upper tier of large windows like Chinon’s Tour dence and preferred here. du Moulin and, though otherwise unvaulted, a 16 Affinities with the buttressed cathedrals of barrel-vaulted basement: could it have been Languedoc have also been suggested, eg. Albi built by Richard I or John? in Tarn and Agde in Hérault (Mesqui 1993, 11 Baronial Courville-sur-Eure and Montland- 329-30), though Richard I is not recorded the on (Eure-et-Loir), though under French over- region (see Landon 1935). lordship, may show Angevin influence in 17 And Richard I may have built a polygonal their later twelfth-century donjons (Corvisier donjon at Radepont (Eure), in Normandy 1998(2), 276-8, 475-8). Angevin influence (Corvisier 1998(2), 133 n. 38). may also be apparent in the free-standing, 18 The curtain wall inturns from its oval trace to circular ‘tours-beffrois’ of central Aquitaine connect with the tower, which sits well back (Mesqui 2013, 103), which appear to have from the edge of the enclosure. been built during the late twelfth and early thirteenth centuries. 19 As it is relative to the buttresses added to the tower at Gisors, by Henry II, in c.1170-1184 12 Perhaps surprisingly, King Stephen of Blois (Mesqui 2013, 199; Stapleton 1840, 72, 110). (1135-54) is not known to have built one in Conisbrough and Mortemer probably influenced either Britain or France, and the cylindrical similarly buttressed donjons in Normandy and tower at New Buckenham (Norfolk), built by Maine, at baronial Condé-sur-Noireau (Calva- his follower William d’Albini in 1138-46, is dos), Le Plessis-Lastelle (Manche) and Saint- thought to be purely local in context (Hislop Rémy-du-Plain, Sarthe (Corvisier 1998(2), 256- 2016, 113-14, et al.). A cylindrical donjon 62, 507-10, 581-5). has been suggested at Sauvey Castle (Leics.), which was begun by King John around 1211, 20 The tower at Gournay-sur-Marne (Seine-Saint- but the site is little-understood (Colvin 1963, Denis, Île-de-France) seems also to have been a 829; King 1983, 255; Page 1907, 249-50; pure cylinder; it is suggested to have been royal, Renn 1961, 131). and earlier twelfth-century (Corviser 1998(2), 352-5), but has now disappeared. 13 Towers of this kind can be distinguished from earlier rectangular donjons such as Portchester 21 An isolated circular tower, of unknown form, (Hants.), which were built before the principle was also built at the royal palace in Paris of flanking was fully appreciated and are only (Mesqui 2011, 311 et al.). incidentally peripheral and engaged. 22 As in Angevin tours-mixtes, however, multi- 14 The Record Tower’s definition as a donjon ple functions are implicit in their design and it has been challenged by, among others, Har- is possible that some saw royal use on an ad old Leask and Con Manning (Manning 2003, hoc basis (Jean Mesqui, pers. comm.); the 72; also see Murtagh 2017, 162, 175). Dan tower at Vernon (the ‘Tour des Archives’), for

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16269 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

example, may have housed King Philip’s 29 Marshal may have spent time in the Capetian treasury during the late 1190s (see Powicke lands with Henry the Young King in the early 161-2 and n., 401-2, 430). 1180s, while he attended two tournaments near 23 John was in Paris, with Philip, in early 1193 Compiègne in winter 1182-3 (Holden et al. (Holden et al. 2006, 126-7; Landon 1935, 723; 2002, 257-81, 305), when he may conceivably Powicke 1913, 143-4). He again visited Paris have seen the pre-Philippienne donjon there – at during the Le Goulet peace in July 1201 (Church least from the outside. He had, in 1188, been 2015, 97), but this was his last time in the French promised the hand of the heiress to Châteaur- royal domain, and many aspects of Philippienne oux, in Berry, for which Henry II negotiated planning had yet to properly emerge. Marshal with the canons of Bourges (Vincent 2000, 5-8), was not present on either occasion. where the Tour Philippienne was under way; there is however no evidence that Marshal him- 24 While most authorities agree that rebuilding self accompanied the king. But it is possible that in masonry commenced at Kilkenny in 1207, he saw Philip’s tower at Gisors being built while it is possible that work had already began campaigning there in September 1198 (Holden under Marshal’s seneschal Geoffrey FitzRob- et al. 2004, 47-55), though Richard I’s forces ert (see Crouch 2016, 120 and n. 34): the are not recorded near Vernon. castle was strong enough to withstand siege 30 in early 1208. But while its crenellations His absence meant he was unable to influence (‘kerneals’) are mentioned in a description of his Normandy lordships, and while it is clear the attack (Holden et al. 2004, 193) – sug- that his followers maintained his interests, he gesting it was more-or-less complete – the was reliant on ‘the good will (or otherwise) of account was written nearly twenty years later, Capetian officers and his local dependants’; and therefore unreliable. contact was minimal, and there was no return to the cross-channel lifestyle Marshal had 25 Nevertheless it has been suggested that it enjoyed before 1204 (Power 2003, 211-13, may, at first, have been isolated (Tietzsch- 216, 224). Tyler 2018, 126 n. 21). 31 Simple loops were employed in Plantagenet 26 Con Manning felt that the circular southwest Gascony until the end of the thirteenth centu- ‘Bermingham’ tower was the first element to be ry, for instance at Sauveterre-la-Lémance built at Dublin (Manning 2003, 90), and that the (Lot-et-Garonne), built by the English Crown Record Tower may be rather later; ‘construction during the 1280s (Mesqui 2013, 69 and fig.). at Dublin mainly spanned the period 1213- 32 c.1230’ (Manning 2003, 72). Tadhg O’Keeffe The order to strengthen Arques was given in has suggested that instead it was the Record March 1201, to Jordan de Sauqueville who Tower that was begun soon after 1204, and that was Marshal’s tenant in Normandy and one of it influenced Kilkenny (O’Keeffe 2009, 283-4), his leading mesnie knights (Hardy 1835b, 35; but keeps an open mind. Power 2003, 219, 224). Some of this work will, moreover, have been undertaken when 27 An attribute shared by the later dominant the castle was in Marshal’s direct custody towers at Cilgerran, Laugharne, Manorbier April-December 1202. and Narberth in west Wales (see below). With 33 the exception of Narberth, these towers – like Marshal was possibly with King John during the Pembroke donjon – also lack latrines further stays at Bonneville in October 1201 (Wiles 2014, 190, 192). and March 1203 (Hardy 1835a, 2, 27). 34 28 John Goodall has for example observed that The isolated Norman exchequer roll of 1198 no Tour Philippienne, including the Louvre, records £20 spent on ‘repairing the dwellings can have influenced Pembroke’s donjon of the castle and the bridge’ at Domfront (Goodall 2011, 162). (Stapleton 1844, lix), which may have been

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16270 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

the start of a wider programme on the castle 41 They are more frequent in French baronial gate; no expenditure had been recorded in the work of the thirteenth century e.g. at La Fère- roll of 1195 (Stapleton 1840). Work may have en-Tardenois (Aisne), in Picardy, begun c.1206 been more-or-less complete by summer 1202 (Mesqui 1997, 168 and fig.), Saint-Vérain when £100 was spent on the ‘towers and (Nièvre), where the slightly later towers show hourding’, though unspecified expenditure at ‘Angevin’-style niched embrasures (Corvisier the castle continued through 1203 (Hardy 1998(2), 594-5), and in Coucy’s outer defenc- 1835b, 50, 79, 84). es, 1220-30s (Mesqui 1997, 135-8). Their use in the outer ward at Harcourt (Eure), soon after 35 Triangular loops were also employed in the 1204, is thought to reflect continued Angevin gatehouse at Carrickfergus Castle, probably influence (Mesqui 1997, 457). built by King John (see below). 42 Its employment at Pembroke, by William de 36 They similarly appear in baronial work from Valence, may have been a conscious reference 1199-1214 at Warkworth (Goodall 2015, 26), to his marriage bonds and Marshal inheri- and elsewhere. tance, in the heartland of the Marshal building 37 It is quite possible that the ‘Hanbury Tower’ at tradition where it would be fully understood. Caerleon, within which these loops occur, is the 43 Richard’s itinerary does not, however, allow work of the younger William Marshal, who held for a stay at Druyes; he was at Donzy on 1 July Caerleon 1219-23, and again 1226-36 (Holden and had reached Vézelay, over 30 miles away, et al. 2006, 187). It may belong to the same by 2 July (Landon 1935, 36). campaign as the structure on the motte, which is normally considered to have been a shell-keep. 44 Marshal was not with them: he remained in William Marshal II is thought to have built the Britain as one of four co-justiciars during Rich- shell-keep at Wiston, Pembs., soon after 1220, ard’s absence (Crouch 2016, 90-4). and may have been responsible for the one at 45 They include Deir al-Balah (‘Darum’), Tel-es- Carmarthen Castle during his tenure there 1223- Safi (‘Blanchegard’), Yazur (‘Casel des 26 (discussed in Ludlow 2014, 18-23). A Plains’ and ‘Casel de Maen’), Yibna (‘Ibelin’) Younger Marshal liking for the shell-keep – and Ashkelon itself (Landon 1935, 57-62). which is not otherwise a feature of west Wales 46 – may be confirmed by its possible use at St As one of only two surviving rolls for the Clears Castle, Carms., held by the Marshals 1190s, it may not record the total expenditure from 1230 until 1245 (Pat. Rolls 1226-32, 339). there, while no particulars are given. 47 38 Offsets are absent from Philippienne work at, Complicating the issue is the statement, by the for example, Beauvais, Dourdan, Gisors, La- chronicler William the Breton, that the castle on, Lillebonne, Montdidier, Péronne and Vil- was slighted by its Norman defenders in leneuve-sur-Yonne. spring 1204; it is far far from certain however that any damage was extensive (Powicke 39 Neither does any Marshal work show the 1913, 373). string-courses that are depicted, in a late-me- 48 dieval illustration, throughout the Louvre (Fl- Recorded expenditure on Richard’s eastern eury and Kruta 2000, 68). Normandy castles, which is doubtless incom- plete, amounted to around £21,500; an addi- 40 The form was also adopted by Hubert de tional £46,000 was spent on the combined Burgh, who had led the defence of Angevin works at Les Andelys, and over £5000 on Poitou in 1202-4 and was later seneschal of fortifications at Eu, Seine-Maritime (Stapleton the province, in the inner ward towers at 1840 and 1844, passim). Major work by Rich- White Castle (Mon.), built c. 1229-39 (see ard I has also been suggested in Poitou Remfry 2011, 221-2). (Baudry 2002, 45, 67-9; see above).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16271 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

49 Nevertheless, whilst there Marshal had com- 2006, 157); two years later, it was confiscated mitted his body to the Templars (Holden et al. by King John (Holden et al. 2004, 217-21) 2004, 415), meaning that he may have spent and was not returned until 1216 (Hardy 1835a, time with the order. 180, 184). It is therefore unlikely that Marshal 50 The influence of existing Roman fortifications was responsible for any building there: the must be taken into account (Knight 2018, 157; hall, and possibly the inner curtain, were al- Mesqui 2013, 37, et al.); it is manifest at ready in existence, while his priorities lay Portchester Castle (Hants.) where the rectangu- elsewhere 1216-19 (Hodkinson 2003, 47-8; lar outline of the early twelfth-century inner Holden et al. 2006, 157). Brian Hodkinson ward was clearly dictated by the Roman fort believes the inner barbican and gatehouse – within which it was built. The use of symmet- which was equipped with a drawbridge – were rical, quadrangular planning in monastic build- instead built between 1219 and 1231 (Hodkin- ing may also be significant: it was probably the son 2003, 48). model for Bishop Roger of Salisbury’s court- 56 Also distinctly Angevin is the system of yard plans at Old Sarum and Sherborne Old tunnels beneath the defences at Arques, ex- Castle, both from the 1120s-30s. tending into the castle ditch (Langeuin 2002, 51 King John stayed at Niort on various dates 364-5), cf. Dover, Nottingham etc. (Baudry from April through to September 1214, when 2002, 61; Hislop 2016, 19, 155). he is also recorded at nearby Parthenay (Har- 57 Their position astride the curtain is analo- dy 1833, 166, 168, 170-2; Hardy 1835a, 113- gous with ‘Marshal’ towers (see above) but, 14, 119, 122, 139). Marshal remained in Brit- like the similarity of these gate-towers to the ain during both 1206 and 1214, though his double-entried Tours Philippiennes, is prob- son Richard was possibly with the king dur- ably incidental (see Mesqui 2013, 318). ing the latter campaign (Crouch 2015, 11; 58 It is possible that grooves were present but, Holden et al. 2004, 237). as in Warkworth’s gatehouse of 1199-1214, 52 Baileys with varying degrees of rectilinearity did not descend to ground level and so not are seen in other earthwork castles from the preserved. late eleventh and twelfth centuries, for in- 59 Cf. the similar gate-tower in the Armenian stance Carmarthen, Lincoln, St Clears castle at Korykos, Turkey, built c.1200 (De- (Carms.), Warkworth and Warwick (Ludlow nys Pringle, pers. comm.). 2014, 178). 60 These ‘horseshoe’ gate-towers can be distin- 53 Shoesmith’s plan shows this early tower with guished from semicircular barbicans, which the same spur-buttresses as the late were open enclosures and perhaps primarily thirteenth/fourteenth century towers (Shoe- administrative, for holding courts etc. (see smith 2014, 130 fig.), for which however Day and Ludlow 2016, 94). The earliest in there appears to be no excavated evidence. Britain appears to be at the Tower of Lon- 54 Marshal may have seen geometric castles in don, from 1275-81 (Impey and Parnell 2011, France during his 1216 visit although, arguably, 34-6), followed at Goodrich and Pembroke’s he is most likely to have met King Philip at outer gate (c.1290-1320). A similar barbican Pont-de-l’Arche (see above), an appropriately had been built by the mid-thirteenth century ‘borderland’ location which was also the venue at Carcassonne, Aude (Mesqui 2013, 354 for his widow’s meeting with Philip in July 1219 fig.). (Crouch 2015, 190-3; Holden et al. 2006, 195). 61 Neil Guy also notes their use by Emperor 55 The elder Marshal did not receive Dunamase Frederick II in early thirteenth-century until 1208 (Hardy 1835a, 80; Holden et al. southern Italy (Neil Guy, pers. comm.).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16272 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

62 Pivot-stones have yet to be reported from 68 But it, too, may post-1219 (Tietzsch-Tyler other contemporary work in the British Isles. 2018, 127). However, they were clearly vulnerable to 69 The remains were sketched and photographed robbing and loss; empty sockets in the gate- by the author in 2016 and 2017. Thanks to way at Manorbier (Pembs.), built by a Mar- Dan Tietzsch-Tyler and Jean Mesqui for dis- shal tenant c.1230, may have held something cussing the results with me. similar (Neil Guy, pers. comm.). Marshal influence presumably lies behind their adop- 70 He had acquired Longueville through mar- tion in Ireland, which persisted in tower- riage in 1189 (Crouch 2015, 465-6). The Mar- houses into the seventeenth century (Leask shals were again resident lords in Normandy 1964, 97-8). between 1220 and 1231, through William the elder’s son Richard (see below), and while 63 The solid-fronted, semicircular towers in the solid-towered gatehouses were still under con- main gatehouse to Montgomery’s inner ward struction at this time eg. at Launceston (Corn- recurve sharply at their junction with both the wall), Longtown (Herefs.), Montgomery and curtain and entrance arch, prompting com- elsewhere, they generally show portcullises parisons with Marshal’s Chepstow gatehouse (Knight 1993, 127-8; Saunders 2002, 6). (Hislop 2016, 167-8); a similar, if less-pro- nounced shape was again employed by de 71 The gatehouse at baronial Bricquebec, from the Burgh at White Castle, c. 1229-32 (see Rem- end of the twelfth century (see above), shows fry 2011, 219-21). However, we saw above similar towers but with open backs. that this shape had been used by the Angevin 72 The baronial twin-towered gatehouses at kings, as were solid rounded gate-towers; the Helmsley and Warkworth are from the same Chepstow gate-towers, moreover, do not re- general period but cannot be closely dated. curve at the entry. De Burgh held Montgom- See Neil Guy’s recent paper for a well-bal- ery for the Crown. Its gate-towers are anced review of contemporary gatehouse de- elongated as at King John’s Kenilworth and velopment in England and Wales (Guy in the slightly earlier gate at royal Pevensey 2016a). (see below), while nearly three-quarter round gate-towers were used by the Crown in the 73 King John formally thanked Marshal for his late thirteenth century, at Harlech and Rhud- intervention with the Pope in summer 1212 dlan. (Hardy 1833, 132; Sweetman 1875, 72-4), but was reluctant to allow his return, commanding 64 This lasting influence is confirmed by the him to remain in Ireland while warning crown large window in Caldicot’s southeast tower, officers to be vigilant against any harm that which is a close copy of the late thirteenth- ‘may be worked against you from the lands or century windows in Roger Bigod’s ‘Glori- jurisdiction of Earl William Marshal’ (Hardy ette’ at Chepstow. 1833, 121-2; Sweetman 1875, 72-3). Mar- 65 An earlier date in the 1240s has however shal’s hostage sons were not released by the been suggested (Goodall 2011, 183, 201), ie. king until late 1212 or early 1213 (Holden et not long after Caldicot. al. 2006, 159). 66 Hislop points out that Barnard Castle was held, 74 Guy de Cultura, a mesnie knight of William between 1278 and 1296, by John de Warenne’s Marshal, in Britain after 1204 (Crouch 2015, son-in-law John Balliol the Younger. 80-1, 108-11, 182-3), was from Les Andelys (Cultura = Petit Andely; see Pitte 2002, 167-9). 67 It may be significant that Marshal was at Chepstow in July 1217 (Hardy 1833, 314; 75 The witnesses narrow the date to 1203-10, and Pat. Rolls 1216-25, 79), and September 1218 include Nicholas Avenel, Marshal’s steward (Hardy 1833, 370). of Pembroke in 1202-3 (Brewer 1863, 227;

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16273 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

Davies 1946, 322, 327) – probably still in 81 Additional saucer-vaults have however been office in 1204 – and Robert Fitz-Richard, lord suggested at the Barnard chamber-tower, of Haverfordwest until 1210. Marshal is not based on antiquarian evidence from the six- otherwise known to have visited west Wales teenth and early nineteenth centuries (Hislop during these years. 2018, 93 and n. 6, 96, 102). 76 A loose fragment of dogtooth is also re-used 82 Similar vaulting was used in the polygonal at Usk Castle (Knight 1977, 147), and proba- donjon at Châtillon-Coligny (Loiret, Île-de- bly also Marshal-period. France), in 1180-90, inspired from Blois (Mesqui 2003, 111 fig.). 77 Dogtooth seems nevertheless to have its ori- gins in northern France where it was em- 83 It may be noted moreover that the alternating ployed from the 1130s until the end of the arrangement of stairways at Conisbrough twelfth century, and could formerly be seen at seems designed to place the maximum social Thérouanne Cathedral, Nord (Hourihane restriction on access to the upper floors (Mc- 2012, 300), demolished in the 1550s. Neill 2006, 123). 78 The West Country was the main architectural 84 It has been suggested that the donjon at Or- influence in twelfth-century Wales and, via ford Castle, Suffolk, carried a conical wood- the Marcher lords, Bristol and the Welsh en ceiling or vault as early as the 1170s, ports, also gained considerable agency in Ire- perhaps supported on a series of arcade posts land (Brakspear 1931, 4; Thurlby 2006, pas- (Heslop 2003, 292-5). sim). Marshal’s closest ties were moreover 85 Derek Renn noted the general similarity be- with this region, which represented his main tween the three (Renn 1968, 41-3). power base and the source of many of his household knights and followers (Crouch 86 Marshal’s eldest daughter Matilda married 2016, 217-18, 231-4); some of these, for in- Hamelin’s son William, but not until 1225 stance the Berkeleys of Gloucestershire, built (Crouch 2015, 39). Marshal’s only attested in a notable West Country style in their own appearance north of Nottingham was in houses and castles eg. the Great Hall at Berke- York, in February 1206 (Hardy 1837, 162). ley Castle, from c.1180 (Emery 2006, 58). 87 Also absent from Château Gaillard and Pem- 79 Marshal’s grant to Monkton Priory, of the tithes broke are the joggled lintels seen at Conis- of three mills (Caley et al. 1846, 320-1; Crouch brough; the same technique was used by 2015, 161-2), may have contributed to this work. Henry II in his donjons at Dover, Orford and Tickhill, Yorks. (Brindle and Sadraei 2015, 80 The tower at Charlieu (Loire), which may be 8), demonstrating Conisbrough’s close rela- Angevin, also has a barrel-vaulted basement tionship with works of the Crown. (see Table 1), while the Tour du Moulin fea- tures a ‘hybrid’ barrel/rib-vault as used by 88 Rib-vaults were infrequent in English round Richard I in the Bell Tower at London. The towers until the mid-thirteenth century, and entrance-floor vault at Châtillon-sur-Indre is were never widely used in donjons or tours- either primary, under influence from Blois mixtes, being largely confined to mural tow- (Corvisier 1998(2), 197, 201), or a secondary ers – and those of D-shaped, rather than insertion when the donjon was heightened in circular plan (see Ludlow 2018, 254 and n. the thirteenth century (Deyres 1984, 366, 376- 11). 7). The ground-floor vault in the cylindrical 89 The vaults in the late twelfth-century hall- donjon at Inchiquin, Co. Cork, also appears to block at Manorbier Castle, Pembs., are sec- be secondary (McNeill 1997, 241), as is the ondary insertions (King and Perks 1970, 95- ribbed vault at Conches-en-Ouche (Corvisier 6, 116). 1998(2), 255).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16274 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

90 In addition, several late-medieval church tow- in the Constable’s Gate at Dover. This dual ers in south Pembrokeshire show ‘domed’ influence may also lie behind the similar summit vaults, eg. St Petrox and St Twynnells donjon on the motte at Chartley (Staffs.), (Lloyd et al. 2004, 446-7), while a similar which was probably built in the 1220s by tower dome in Tenby’s town walls belongs to Ranulf, earl of Chester (Goodall 2011, 180- the late fifteenth century (Thomas 1993, 13). 1; Hislop 2016, 125), and also shows a semi- 91 The Marshal/west Wales tradition, referred to circular turret. The earl had spent most of throughout, was probably not a formal school 1199–1204 in Normandy where he pos- with dedicated lodges, but may have been a sessed extensive estates, while his constable looser network of professional links through of Chester had defended Château Gaillard time, as well as space, and lasting over 200 (Powicke 1913, 365, 379-80, 446, 491). years. It was perhaps largely driven by hom- Chartley’s overall plan-form has been com- age and emulation (see Wiles 2014). pared with King Philip II’s castle at Montl- héry, Essone (as noted by Duffy 2017, 312 92 A date between 1216 and 1231 has however and fig., 314), but any similarity is slight and also been suggested (Guy 2017, 29). incidental: Montlhéry has no motte, and 93 Semicircular turrets, sometimes housing shows a Philippienne symmetry entirely ab- stairs, were also present in the French don- sent from Chartley where the towers and jons at La Queue-en-Brie (Val-de-Marne), gatehouse are also of very different form. and Montchauvet and Montfort-l’Amaury 97 The small, isolated round tower at Tenby (Yvelines); they belong to a group of unusu- Castle, which apparently has a summit-vault al, non-rectilinear towers built by the Mont- (King 1977, 164), with a ground-floor entry forts during the twelfth century (Corvisier and rectangular stair-annexe that may be 1998(2), 442-3, 461-3, 541-50, 580). secondary, is possibly analogous to Nevern, 94 Ardfinnan was periodically held of the Crown but in the absence of diagnostic detail cannot as a tenant lordship, but was in John’s own be closely dated. While it conceivably be- hands by 1210 (McNeill 1997, 31, 233). The longs to the earlier thirteenth century, most tower shows cruciform loops, as employed by authorities have followed the Royal Com- the Angevins by the mid-1190s (see above), in mission’s preference for a date in the mid- embrasures with slightly pointed heads. late thirteenth century (eg. Goodall 2011, 208). The possible small, half-timbered 95 It has been suggested that the simple chamfer round tower on the motte at Carmarthen can at Lyonshall may indicate an even earlier date, be neither fully-interpreted nor dated (Lud- in the 1180s (see Pamela Marshall’s com- low 2014, 179-81). ments in Guy 2017, 121). However, the cham- fered string-course was a very conservative 98 King Philip is attested at a number of loca- and persistent form seen, inter alia, in the tions during the period of the elder Mar- basal string-course of the dominant tower at shal’s visit, including Melun, which is 25km Narberth Castle, Pembs., which was almost north of Nemours (Samaran and Nortier certainly built by Roger Mortimer of Wig- 1979, 27-32). However, the meeting is more more – using masons from the Marches – after likely to have occurred at Compiègne or 1257 (Ludlow 2003, 13, 24). It was widely Pont-de-l’Arche (see above). used in Wales, particularly in church towers, 99 Richard Marshal may also have been with into the sixteenth century. Philip in summer 1220, between receiving and 96 Skenfrith’s conservative round-headed taking formal possession of his Longueville openings may be a deliberate reference to lordship, when the king again stayed at Nem- Angevin prototypes, like so much work of ours (Samaran and Nortier 1979, 304-5). de Burgh’s work including White Castle and

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16275 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Notes

100 The sources suggest that Richard Marshal and Coughlan 2003, 138-43) and probably made attempts, during the Anglo-French Lea (Dempsey 2017, 242). No central pier détente of 1219-24, to revive his father’s has been identified at Lea, Carlow or Ferns, cross-channel lifestyle (Hardy 1844a, 98-9; unlike Nemours where, however, it may be see Power 2003, 210-11). He had been secondary (Mesqui 2013, 128, 214). granted the manor of Long Crendon 103 The castle is not named. In 1230-31, William (Bucks.), by his elder brother, in 1219-22 Marshal II was quit of 80 days’ service, this (Power 2003, 210), while a number of man- time towards work on the castle at Castle- dates were issued in 1225-6, after the re- comer, Co. Kilkenny (Sweetman 1875, 270, sumption of hostilities, ordering his ships 278). from Normandy to be impounded (Hardy 104 1844a, 47-8, 149). Richard himself was probably the source for the poem’s glowing testimonials to his fa- 101 The towered keep at Terryglass, Co. Tipper- ther, made by French knights after the lat- ary, had been built by 1333 (McNeill 1997, ter’s death in 1219 (Holden et al. 2004, 469; 118-19). The other four are either lost (Wex- Holden et al. 2006, 25, 195). ford, which had been built by 1323), late 105 medieval (Dunmoe, Co. Meath and Enniscor- David King, while noting that the two are thy, Co. Wexford) or undated (Delvin, Co. bonded higher up (King 1978, 118), does not Westmeath); see Colfer 2013, 52; Sweetman mention that the curtain was heightened, 1999, 39, 60; McNeill 1997, 118. O’Keeffe where it joins the tower, when the latter was and Coughlan regarded Ferns as belonging to added (see Ludlow 1990, 28 and fig.; Fig. 35). the earlier part of William de Valence’s ten- 106 Also of interest is Richard Marshal’s alliance ure, during the mid-thirteenth century with Llywelyn in 1233-4 (Neil Guy, pers. (O’Keeffe and Coughlan 2003, 147). Howev- comm. See Walker 1994, 43-4, 62-3;). er, they allow that certain of its features – the 107 The Dolbadarn sockets have an alternative cross-loops and trilobe-headed twin-light interpretation as supports for a timber stair windows – would normally indicate a later leading to an appearance balcony (Smith date. To these might be added the large circu- 2014, 68-9); an appearance doorway is sug- lar oillets (cf. Valence’s Goodrich); the gested in Pembroke’s donjon (see Appendix colour-banded masonry (cf. Caernarfon and 1), but by itself this does not imply any Haverfordwest); and the similarity of the Marshal associations at Dolbadarn. A rela- windows to those in Roger Bigod’s Gloriette tionship between Dolbadarn and the slightly at Chepstow, from the early 1280s (Turner et earlier donjon at Clogh Oughter has been al. 2006, 135-7), and in Queen Eleanor’s suggested by Con Manning, via the marriage Gloriette at Leeds Castle, Kent, from 1278- of William Gorm de Lacy, of Clogh Ough- 90 (Goodall 2011, 236-7 and fig.). Filleted ter, to Llywelyn ap Iorwerth’s daughter corbel-shafts, as in the Ferns chapel, were Gwenllian (Manning 2013, 198-9). still being used into the 1320s in the great hall at Caerphilly, while their facetted springers closely resemble those in the cellar of Chep- stow’s Gloriette. Ferns may then belong to a later period of Valence’s tenure, perhaps the 1280s-early 1290s. 102 There are further similarities between Nem- ours and Irish examples, including the use of one corner turret as a chapel at Nemours (Mesqui 2013, 128, 207), Ferns (O’Keeffe

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16276 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Table 1

Table 1 – Circular and polygonal great towers in France before c. 1203 (compiled from Corvisier 1998, Baudry 2002, Mesqui 2013 and others) 1/3 Ang. - Angevin; FR. - French; Vest. - Vestigial; * - Tour Philippienne.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16277 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Table 1

Table 1 – Circular and polygonal great towers in France before c. 1203 (compiled from Corvisier 1998, Baudry 2002, Mesqui 2013 and others), continued, 2/3

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16278 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Table 1

Table 1 – Circular and polygonal great towers in France before c. 1203 (compiled from Corvisier 1998, Baudry 2002, Mesqui 2013 and others) continued 3/3

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16279 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1

APPENDIX 1: Situated for maximum effect on the Pembroke’s donjon: form and function highest part of the headland, and immedi- The circumstances behind the erection of ately behind the defences, Pembroke’s Pembroke’s donjon, discussed above, sug- donjon was also a high-visibility architec- gest that it was primarily celebratory and tural reminder of lordship, by an earl who commemorative like the donjons at e.g. was normally absentee (see Marshall Coucy (Aisne, in Picardy), Hedingham (Es- 2002a, 142; Marshall 2002b, 35; Marshall sex), Rochester (Kent), Castle Rising (Nor- 2016, 161). The donjon may not, in fact, folk) and Dundrum (Co. Down), all of have seen any use under William Marshal which were built to mark major acquisi- who was in Ireland 1207-13 and not tions, social advance and upturns in status – known to have visited west Wales 1 frequently, like Marshal, through marriage thereafter. Several visits by the younger (Dixon 1998, 55-6; Marshall 2002a, 146; Marshals are however known or surmised, McNeill 2003, 96, 98). when the donjon may have fulfilled its intended role. Nevertheless, even when Though fireplaces are present at first- and the lord was in residence, it is suggested second-floor level, there appears to have been that many larger donjons were used infre- no provision for a latrine, indicating that – quently – possibly limited, in some cases, like the great towers at e.g. Château Gaillard, to a single, specific occasion (Marshall Chepstow, Hedingham (Essex) and the Tow- 2016, 160).2 But, at the same time, Philip er of London – it was not intended for resi- Dixon’s reminder that designed function dential use (see Dixon 2002, 11; Marshall may have differed from actual use must be 2002b, 32; Marshall 2016, 171) and, in this, born in mind (Dixon 2002, 11). it may be like many circular donjons. The third, uppermost floor is unheated and shows At second-floor level is a narrow external no arrangements consistent with use as a doorway, beneath which are two large bedchamber. Neither of the two fireplaces joist-sockets in the external face, presum- served a food-reheating area, unlike the don- ably to support timber decking. A jettied, jons at e.g. Norwich, Castle Rising (Norfolk), timber latrine, as suggested in e.g. the Bowes (Yorks.) and Sainte-Suzanne in May- donjon at Caldicot, Mon. (Morgan and enne (Marshall 2002a, 149; Marshall 2002b, Wakeman 1854, 22), is unlikely – the 30), and many Tours Philippiennes including entrance passage is straight, without the Dourdan and Lillebonne (Mesqui 2013, 163); dog-leg normally seen in latrine passages neither is there is any evidence of a well. All including Caldicot (see Wood 1983, 378). these attributes, or lack of them, indicate Nor is it likely that the doorway led onto a highly restricted and mediated use, probably timber bridge to the inner curtain parapet, limited to ceremonial occasions (see Mar- providing access to the latrine turret at the shall 2002a; Marshall 2002b, 30). A residen- north end of the curtain.3 The curtain is tial chamber-block for the earl already secondary to the donjon (though probably existed at Pembroke, in the form of the Nor- commenced during its construction) and, man Hall (Fig. 16); Marshal’s communal hall as originally built, its parapet lay four and steward’s accommodation appear to have metres below the level of the doorway been rebuilt and repurposed in the late thir- threshold (Fig. 35).4 The entrance passage teenth century (Day and Ludlow 2016, 66-8). lies at an awkward, oblique angle to the

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16280 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1

Fig. 36 – Floor plans of the donjon at Pembroke (radial joist arrangement after Rick Turner).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16281 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 curtain (Fig. 36), and the necessary combi- gins (Marshall 2016, 165). It is evident in nation of stairway and bridge would have the early rectangular donjons of Anjou been a complex construction that is not (ibid.; Gregory and Liddiard 2016, 149), evident in the surviving fabric. The turret, and in the cylindrical donjons at Fréteval moreover, lies nearly 35 metres away from and Châtillon-sur-Indre (Corvisier 1998(2) the doorway (Fig. 16), while the latrines 199, 201; Mesqui 2013, 129, 131 fig.), within could not be accessed from parapet which Marshal may have seen in the level; a possible latrine in the Horseshoe 1180s-90s (see above). It was also seen in Gate is similarly distant. There is no roof- Britain during the twelfth century, where crease above the doorway, showing that the the balcony at Norwich, as at Pembroke, timber superstructure was open to the overlooked the marketplace (Gregory and elements,5 while the entry itself is not fur- Liddiard 2016, 149). It may be significant nished with a drawbar-socket, arguing that an appearance doorway already exist- against communication with any other part ed at Marshal’s Chepstow, in the earlier of the castle. It might moreover be asked great tower (Turner, Jones-Jenkins and why, if latrines were required, they were Priestley 2006, 39-40). not built into the wall-thickness as in the At Pembroke, a single spiral stair connects pre-existing donjons at e.g. Orford and all four storeys and the parapet; no separate Scarborough – where they discharged into line of communication existed between lev- the inner ward rather than outside the castle els, according to social rank, suggesting that – or latrines with intramural cess-pits (‘pit- access to the entire donjon was restricted to latrines’), as employed in the donjons at those above a certain status (see McNeill e.g. Neaufles, Coucy and Dover (Corvisier 2006, 123, 125). In addition, the first-floor 1998(2), 490; Mesqui 2013, 169-73; Brin- (entry-level) chamber had to be entered to dle 2015, 15-21), and perhaps at Marshal’s reach the stair, suggesting the two upper- Pembroke itself, in the junction between most floors – whose entries were draw- the Horseshoe gatetower and the curtain barred from inside, against the stair – were (Fig. 16).6 Finally, the curtain here was, out-of-bounds. The donjon therefore ap- until the mid-thirteenth century, the outer pears to represent a space within which defence of the castle lying alongside the access was controlled and mediated, to en- town – a very exposed, public location for ter which was a mark of great privilege (see such a private function. Marshall 2002b, 33). Its very visibility may however provide a The eight narrow-splayed openings in clue to the function of the entry: it is most the body of the donjon are frequently likely to have led to an appearance balcony, referred to as loops. The embrasures how- facing the main street of the town and the ever are four metres long, but most are probable site of the early marketplace, from less than a metre wide to the interior and which the earl could be seen above the average only 1.5m in height (some are curtain parapet. It is very tall, and rather much lower), while their sills lie up to a narrow – consistent with this role – and has metre above floor level. No functional an oolite surround of high quality. The ap- traverse is possible within them, while pearance doorway and balcony were a long- they clearly could not be used by an arch- established tradition, with Carolingian ori- er standing within the interior of the don-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16282 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 jon itself. And while the sills show treasury in Wales and the west was at plunges, they are very shallow: the ground Chepstow (Crouch 2006, 49; Holden et al. within a 25-metre radius could be neither 2006, 36). Nor was it particularly secure. seen nor controlled. Jean Mesqui notes It is possible that, in addition to its prima- that, in general, donjons ‘do not include . . ry function of adding height to the donjon, . arrowloops; when there are external slots, it merely provided ventilation space and they are for lighting’, consistent with their insulation for the first floor (as suggested restricted access as stressed by Philip Dix- at Fréteval; Marshall 2002b, 31). Never- on (Dixon 2002, 10; Mesqui 1993, 252). theless the relative ease of access to the So the Pembroke openings are more likely basement suggests practical usage and it to be slit-lights or possibly, as Derek Renn may have been a wine-store, as in contem- suggested, for ventilation (Renn 1968, 39). porary chamber-blocks and, perhaps, the They are, nonetheless, an unusual feature donjons at Loches and Rochester (Blair above basement level in free-standing cir- 1993, 5; Marshall 2002a, 144, 147).7 cular donjons, the closest comparisons be- The first floor was carried on a system ing again in Angevin France e.g. Le of twenty-eight radial timbers, the sockets Brandon (Indre-et-Loire in Touraine), Le for which remain, probably supported at Châtelier and Neaufles, and at Caldicot their distal ends on a timber pier as in (Mon.) and Dundrum (Co. Down); at Pem- King John’s polygonal donjon, from broke, their design may have been chosen 1207-16, at Odiham, Hants. (Hislop 2016, as a deliberate foil to the two windows, the 39); cf . Marshal’s tower at Kilkenny, visual impact of which is enhanced by the where the pier was apparently of stone simplicity of the lights. (Murtagh 2017, 175), as it may have been The interior in the circular donjon at Llanhilleth, Mon., The ground floor forms a basement that, and perhaps at Penrice, Glam. (Renn originally, was entered solely from the first 1961, 142). A much simpler system of floor via the spiral stair. As in the donjons radial joists, without a central pier, was at e.g. Dinefwr and Dryslwyn (Carms.), used at Dundrum and Longtown (McNeill Dundrum (Co. Down) and Skenfrith 2003, 103-4). Rick Turner has suggested (Mon.), the present ground-floor entry is that the distal ends of the joists at Pem- secondary (King 1978, 99-100; Knight broke were sawn obliquely, narrowing 2009, 30; Guy 2015, 47-9; Renn 1968, 38; them gradually so that they could meet at Rees and Caple 2007, 30, 51); possibly the pier (Rick Turner, pers. correspon- late-medieval or sixteenth-century, it is an dence; see Figs. 36 and 37); Tom McNeill indication of the changed use and status of suggested a more complex arrangement the donjon also apparent in other altera- with a concentric ring of timbers to which tions discussed below. The chamber is un- their inner ends were attached (McNeill lit, and otherwise featureless. Pembroke’s 2003, 104). Either would be a very elabo- limited use by Marshal may argue against rate arrangement for something that it housing a treasury and/or archive store, would never be seen by most users of the as in the donjons at Rouen and elsewhere donjon. And while the timbers were (Holden et al. 2004, 89; Renn 1968, 37), 0.25m square, and would support a sub- while the sources suggest that their main stantial floor surface, the original wall-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16283 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1

Fig. 37 – Section, facing east, through the donjon at Pembroke.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16284 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 plaster terminates just above the top of the King John. However, while corbels at joists, showing that there was only one shoulder-height (represented by empty layer of floor-boarding. sockets at first-floor level) may have sup- The flooring arrangement at second-floor ported lintels, there is no toothing for the level was revealed in July 2018, when the hood itself, which survives at Vire (Fig. interior of the donjon was scaffolded for 38). As existing, the fireplaces are without remedial work. The sockets are somewhat close comparisons;8 are they another Mar- weathered at this level, but are around shal idiosyncrasy? 0.20m square and are increasingly oblique towards the east and west, showing that The first-floor, entry-level chamber was they received the ends of eleven more-or- lit by two slit-lights, and heated by a wall- less parallel joists (Fig. 36), as at Château- fireplace. It may have been the waiting sur-Epte (Corvisier 1998(3), Fig. 101) and area or anteroom for a second-floor audi- elsewhere. Two slightly lower sockets, to ence chamber, as it possibly did in the the east and west, appear to have received rectangular donjon at Richmond, Yorks. a cross-beam, not quite at right-angles to (Marshall 2016, 165) and, in an inverted the joists, providing additional support as sequence, in the great tower at Château in the towers at Clogh Oughter and Gaillard (Marshall 2002b, 32); like most Inchiquin, Co. Cork (McNeill’s Group 1A; cylindrical donjons, Pembroke lacks the McNeill 2003, 100 and fig.). forebuilding suggested to have housed the waiting area at e.g. Castle Rising, Dover It is difficult to envisage how the upper- and Rochester (Dixon 1998, 50; Marshall most storey was floored. The shallow sock- 2002a, 142, 146; Marshall 2002b, 28-9).9 ets are perhaps radial to the tower, but are Alternatively, the first floor was perhaps very ruinous and difficult to identify indi- itself a reception chamber, at a lower or vidually; moreover, they seem never to different societal level from a second re- have formed an uninterrupted series. Three ception chamber on the floor above, as at lie a slightly lower level, but do not form a Rochester and, possibly, Richmond (Mar- coherent group. Three surviving corbels, at shall 2002a, 144, 146; Marshall 2016, differing heights around the second-floor 165), perhaps presided over by an official chamber, possibly carried upright posts (or junior family member) as suggested at against the wall-face, into which braces Hedingham (Dixon 1998, 55). may have been jointed to support these third-floor timbers. The second-floor chamber is lit by an elaborate twin-light window of high qual- The two fireplaces, which have slightly ity, embellished externally with carved rounded backs, are unusual. They are very detail (see above) and showing window- tall, with segmental heads some 3.5 metres seats. There are two additional slit lights, above floor level (Fig. 38). They look like a wall-fireplace and the external doorway they have lost their hoods, being similar in noted above. It may represent a private form to the formerly-hooded fireplace at audience chamber, or perhaps an upper Vire (Calvados in Normandy), in the square reception room as at Hedingham, for use donjon built by Henry I (Châtelain 1983, 22 by the earl should he choose to visit. In and n. 11) and seemingly remodelled by either role, the chamber would be a fitting environment in which tenant lords and

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16285 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 guests could be received, providing a grand of the view, while the window in the more setting for formalities like receiving hom- ‘public’ chamber below faced the castle age, the award of grants and charters, and buildings, can be seen in other donjons diplomatic transactions (see Marshall (Gregory and Liddiard 2016, 149).12 10 2002a, 142; Marshall 2016, 164). An ap- Running around the summit of this pearance balcony will have enabled some chamber is an offset which supported the of these formalities to be witnessed and timber centring used in the construction confirmed by a wider audience. In the ab- of the dome, the underside of which pre- sence of a niche or recess, it is possible that serves the impressions of plank shuttering the earl’s chair was situated in front of the in its plaster finish.13 The offset was later window, a favoured location which placed taken advantage of by an inserted floor visitors at a psychological disadvantage (see below); two groups of three corbels (see Marshall 2002b, 31). which lie just below it, on opposite sides Lying beneath the dome, the uppermost of the interior, might have been be associ- chamber had the best lighting with a sec- ated either with dome construction or this ond large window, of high quality, and four flooring. slit-lights. It was unheated and, like for The summit example Hedingham (see Dixon 1998, 55), shows no arrangements for sleeping ac- The upper works have undergone consid- commodation. The dome was doubtless erable alteration. Just beneath summit lev- intended as the visual centrepiece, and its el is a ring of what appear to be sockets underside embellished with a decorative for a timber gallery or hourd. As David painted scheme. However, there is no pisci- King noted, many such sockets are too na and, while the window is correctly ori- small to take a functional timber and may ented on the west side, there is no often have been merely for drainage, but corresponding window on the liturgical at Pembroke ‘the arrangements for drain- east side to light an altar: the chamber ing the wall-walk, which are most effi- therefore was probably not a chapel or cient, are quite distinct from the row of oratory.11 It may then be an alternative holes beneath the parapet, each of them candidate for the role of audience chamber, capable of holding a massive timber’ while the lack of a fireplace militates (King 1978, 102). against prolonged episodes of use. Never- Hourding was used systematically in the theless, it may have been reserved for the outer curtain at Château Gaillard (Hislop entertainment of a special guest – perhaps 2016, 160; Mesqui 1993, 327, 334); it is the king, should he visit – as suggested absent however from the donjon. Unequiv- elsewhere in Britain and France (Marshall ocal evidence for hourding is infrequent in 2002b, 30). The window gave views to- Angevin France and Britain, and is best wards Pembroke Priory, at Monkton, the seen in the donjons at Pembroke and kind of ‘seigneurial marker’ that guided Conisbrough (see above), the donjon at donjon window planning elsewhere (Mar- Rochester, the donjon and towers at Caldi- shall 2016, 163; also see Wiles 2014, 183); cot, the inner gate at Corfe (Dorset) and the the arrangement at Pembroke, in which the north tower at Stokesay, Shrops. (Davis uppermost chamber window took advantage 2016, 257-60; Goodall 2011, 156-7 and

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16286 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 figs.; Hislop 2016, 161-2). The hourding at oillets,16 perhaps primarily for display as Stokesay, which survives and has been in the Wogan stair turret, where the lowest dated to the 1290s (Hislop 2016, 52-3), is and most prominent slit-light has an iden- domestic in character and was a perma- tical, but non-functional oillet. The para- nent, rather than temporary fixture, inte- pet nevertheless does not co-exist happily gral with the tower roof structure. This with the provision for hourding: if any may have been normal (see Hislop 2016, military intent is to be implied, the loops 51), and is seen elsewhere in, for example, could not be used once the hourding was the chamber-tower at Laval (Mayenne), in place while, if the loops were for dis- dated 1219-27 (Chollet and Gousset 2012, play only, they would be concealed from 261; also see Pré 1961, 353-72). the viewer behind the decking. While We have seen that the donjon at Pem- looped parapets may have co-existed with broke was not intended for defence and the hourding-sockets at e.g. Caldicot and perhaps never accessed by soldiers. As in Château Gaillard (Hislop 2016, 160), other donjons, this restriction may have Pembroke’s sockets lie well below the extended to the summit: as height was level of the parapet drainage chutes which equated with status (Marshall 2002a, 143), will have discharged, with damaging ef- it follows that the parapet should be the fect, over the decking (see Fig. 37). So a most exclusive – and excluded – zone in change of design may be implied, in the donjon.14 And recent studies suggest which the gallery was abandoned in fa- parapets were accessed primarily for their vour of a looped parapet; given that the views, limited to privileged individuals donjon may have reached parapet height and guests to whom the lord’s seigneurial by 1207 (see above), this change may ‘markers’, like religious houses, could be coincide with Marshal’s return from Lein- displayed (Gregory and Liddiard 2016, ster in 1213, after which the new design 148; McNeill 2006, 123, 126-7).15 It is was implemented (cf. Murtagh 2017, therefore possible that the gallery at Pem- 156).17 The looped upper gallery in the broke, and elsewhere, was similarly in- donjon at Kenilworth (Warwicks.) ap- tended for non-military uses: those at pears similarly to be an early thirteenth- Châlus-Chabrol (Haut-Vienne) and Salon- century alteration or addition and, again la-Tour (Corrèze) were entered through as at Pembroke, was probably for display decorative doorways from high-status rather than military use (Renn 2012, 175). rooms (Marshall 2002b, 33-4), and some- An inner wall running inside the wall- thing similar may be implied in the circular walk is built against the flank of the dome, donjon at Edwardian Flint, where the tim- rising up to oversail the parapet as at ber summit gallery was described, in 1301, Conisbrough (Brindle and Sadraei 2015, 9 as ‘noble and beautiful’ (Renn and Avent fig., 13; Davis 2016, 251); it may belong to 2001, 27). the change in design, although a ‘fossil- Above Pembroke’s sockets is the parapet, ised’ line of slabs, low down in its outer which is an accomplished work of some face (see King 1978, 103), is possibly asso- complexity, thicker at the merlons than the ciated with the original arrangement. The embrasures. The former are pierced by wall is now truncated, but openings (or plunging loops with rectangular basal embrasures?) may be suggested in illustra-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16287 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1

Fig. 38 – Left: the interior of the donjon at Pembroke, facing northwest, showing first- and second-floor fireplaces. Right: the fireplace in the donjon at Vire (Calvados), showing keystones for the hood.

Fig. 39 – John Speed’s plan of Pembroke, 1610 (National Library of Wales). Inset, top right: the castle as depicted in 1595 (from Owen 1897).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16288 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 1 tions from 1595 and 1610 (Fig. 39).18 The It is possible that the flagged drainage gul- elaborate flagged drainage gully running ly, mentioned above, also belongs to this within this wall (King 1978, 103 and Plate late work. The inner concentric wall may 11a; Figs. 36 and 37), which is inaccessible originally have supported a conical roof, as from the outer parapet and not a second suggested at Conisbrough and Château wall-walk, cuts through the surface of the Gaillard and a necessary covering for any dome and therefore secondary. timber dome (see above and Fig. 31). But, The spiral stair caphouse butts against the while vaulting was frequently roofed in the outer parapet, blocking its embrasures, and Middle Ages, domes were often left ex- is clearly also secondary but cannot be posed and there appears to be no provision closely dated; some kind of superstructure for roofing over the vaulted summits at e.g. at the head of the stair may be assumed Manorbier or Laugharne castles, nor else- from the start. Other later alteration, possi- where in Pembroke Castle itself. bly post-medieval, includes the crude inser- Not all of these later alterations, includ- tion of an entry through the dome, at ing the ground-floor entry, may be con- wall-walk level, and a light on the opposite temporary: the donjon may have been side. The light preserves the impression of redundant in its original role fairly soon a timber window-frame, of late form; the after it was completed, as at Chepstow entry however shows no provision for a (see Turner, Jones-Jenkins and Priestley door (see King 1978, 102; Renn 1968, 39) 2006, 42), with different uses found at and cannot therefore belong to a period different periods thereafter. But a radical when the tower was still a functional don- repurposing of the donjon in the fifteenth jon. The openings indicate that an addition- or early sixteenth century is suggested by al floor had been inserted beneath the the broad, curving flight of masonry steps dome, taking advantage of the ledge for its to the first-floor entry which, though centring and converting it into an attic largely rebuilt in the twentieth century, space.19 This in turn is probably contempo- was described in the nineteenth century rary with the insertion of timber partitions and depicted in antique prints and maps. in all three upper levels (King 1978, 102), APPENDIX 2: which were held in chases crudely hacked through the pre-existing plaster, and would The Marshals, Pembroke and Ireland provide additional support for the attic The records indicate that Chepstow Castle flooring. It is possible that the chases may was a favoured residence for both genera- postdate the removal of the third floor sur- tions of the Marshal family, for whom face, as the uppermost runs through the Pembroke was mainly a source of prestige, second and third floors, though perhaps not as their titular caput, and of revenue. uninterrupted (Fig. 37). Chepstow by contrast was central to the The central circular ‘turret’, the base of elder Marshal’s chief power bloc in which survives at the summit of the dome, Striguil, Usk, Gloucester and Dean, and may also belong to this later work. It was was the main Marshal port. Like his sons, also depicted in 1595 and 1610 (Fig. 39), William Marshal saw little of Pembroke – but only in vague outline; it cannot be only two visits are certainly known – accessed from the wall-walk and its func- whose castle was left to witness the deeds, tion remains indeterminate. mainly unrecorded, of its seneschals.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16289 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2

Despite Pembroke’s closer proximity to welyn ap Iorwerth, in 1218, Marshal relin- Ireland, moreover, Chepstow appears to quished Carmarthen and Cardigan to the have been the Marshal’s port of choice for Welsh prince in return for Caerleon Castle the crossing: while Milford Haven was a (Hardy 1833, 356; Holden et al. 2004, 397; busy waterway during the Middle Ages, Pat. Rolls 1216-25, 143): while the two Pembroke itself was not accessible to sea- west Wales castles were Pembrokeshire’s going vessels. Its maritime activity was first line of defence against the Welsh heart- restricted to local, coastal traffic throughout lands, Caerleon had posed a threat to the the historic period, and no quay was present security of Chepstow.20 until the nineteenth century. The elder Marshal crossed to Ireland Pembroke and Chepstow from Chepstow in 1207 (Holden et al. 2004, 169-73), and William Marshal II There can be no doubt that William Marshal chose the same route when he became treated Chepstow as his principal seat in justiciar in 1224 (Crouch 2016, 12). It was England and Wales, and that it remained the the base for their shipping: ‘galeas et alias centre of Marshal power (David Crouch, naves Marescalli . . . ad Striguil’ are men- pers. comm.; Crouch 2006, 49). When not at tioned in the 1230s (Crouch 2006, 48-9; court, in Ireland or active militarily and ad- Holden et al. 2006, 36, 152) and, as David ministratively, Marshal was recorded either Crouch points out, ‘it was at Chepstow in Normandy (until 1204), on his English that the Marshal fleet which commanded estates, or at Chepstow. The association be- the Severn estuary was based, the fleet comes most apparent after 1204, but he was that would have taken the younger earl apparently there in March 1194 when he Marshal to and from Ireland, as it took his heard of King Richard’s death (Holden et al. father before him’ (Holden et al. 2006, 2002, 509). He chose to reside there in 1206- 36). Pembroke may then have been a stop- 7 when he withdrew from King John’s court over during Marshal’s crossing of 1200 (Crouch 2015, 11; Crouch 2016, 116), and (see above), in order to visit his new lord- visited again in 1217 and 1218 (Pat. Rolls ship and commission the donjon, rather 1216-25, 79; Hardy 1833, 314, 370). Mar- than his port of embarkation. While Rich- shal probably kept his treasury and archive ard Marshal appears to have crossed to there (see above), while custody of the castle Ireland after visiting Pembroke in autumn and lordship may, by 1217, have been vested 1231 (see Crouch 2014, 396-7), it is likely in his closest companion John of Earley that, in 1233 and 1234, he used Chepstow (Crouch 2016, 171), whose son, another where he had spent much of his time. John, was in control in 1219 (Holden et al. After his death in 1234, Richard’s broth- 2004, 411). The younger Marshals are fre- ers seem to have returned from Ireland via quently recorded at Chepstow (Crouch Pembrokeshire (David Crouch, pers. 2015, 245-6, 251-2, 299-308, 337-8), while comm.). But while Walter Marshal almost Henry III visited in 1217 (Hardy 1833, 314; certainly crossed from Pembrokeshire Pat. Rolls 1216-25, 79) and 1232 (Cal. Pat. when he visited Leinster in 1244-5 Rolls 1232-47, 6). This stands in stark con- (Crouch 2015, 402-03, 415-17, 434-5), he trast to the limited evidence for any close may have used the port of Haverfordwest, relationship with Pembroke. It may also be where he is recorded in June-July 1245 significant that in the truce made with Lly- (Williams ab Ithel 1860, 85).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16290 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2

Pembroke and shipping the west side of Pennar Pill, near the vil- While many of the Marshal tenants in Ire- lage of West Pennar, and within it land also held lands in Pembrokeshire, it is ‘barques of 50 tons may ride safe and in far from certain that they chose to embark good anchor-hold at all times in low wa- from Pembroke. Along with their officials ter’ (Owen 1897, 545-6). Crow Pool is – and those of King John, 1207-c.121321 – also marked as an anchorage on Lewis they may, like the Marshals themselves, Morris’s chart of Milford Haven, from have used other ports such as Marshal-held 1748, with a depth of nine to twelve feet Tenby (Pembs.), which had a natural har- at the mouth of Pennar Pill (Morris 1748, bour suitable for sea-going vessels, or 15 and plate 24). As the historian Ron Haverfordwest. Walker recognized, the ‘Cross’ and Crow Pool are one and the same (Walker 1989, Pembroke’s borough charter, from 1102- 143-4); the latter name may be derived 30, mentions an anchorage called the from the Norman-French croiz/croix. ‘Cross’ (‘ad crucem’; Ballard 1913, 169; Cal. Pat. Rolls 1377-81, 106-7; Walker All of Pembroke’s maritime traffic ap- 1989, 132, 136-7). Next, the Norman- pears to have been conducted from this French epic poem The Song of Dermot and anchorage, from which, when tides per- the Earl, probably composed in the 1190s mitted, lighters or barges conveyed cargo (Crouch 2016, 123 and n. 40), tells us that to and from the town – and, during the 22 Henry II sailed for Ireland, in 1171, from Middle Ages, the castle. While Henry ‘The Cross’ near Pembroke (‘la croiz en II’s charter commanded that ‘all ships mer entra’; Orpen 1892, 188-9). Then in with merchandise that enter the port of 1210 King John assembled his fleet and Milford and wish to buy or sell on land personnel, for his Irish crossing, at ‘Cross- shall come to the bridge of Pembroke’, it below-Pembroke’ (‘crucem subtus Pen- seems that lighters are meant; larger ves- broc’; Hardy 1844b, 172-7). It appears to sels remained at the Cross where they paid have been the Pembrokeshire historian ‘their lawful customs’ (Ballard 1913, 169; Henry Owen who first suggested that the Cal. Pat. Rolls 1377-81, 106-7; Walker Cross was not a terrestrial location, but 1989, 137), making it clear that these 23 instead referred to an anchorage in Milford customs were collected at sea. Haven, downstream from the castle, from The identification of the Cross clarifies which ‘small boats . . . took out passengers the role of Pembroke, as a port, during the and cargo to ships lying in deeper water’ medieval period and later. The town itself (Owen 1897, 317 n. 5). was inaccessible to sea-going vessels, and Historically, the most important anchor- never properly developed as a port: an age in the Haven, for the town of Pem- account of 1818 tells us that ‘at low water broke, was Crow Pool in Pennar Pill, 1.5 the channel is narrow, intricate, and wind- km west of Pembroke Castle. The Pem- ing, from Crow Pool to Pembroke, and brokeshire topographer George Owen within which is an anchorage for small wrote, in 1595, that ‘a barque of 40 or 50 vessels; but those of burden are excluded tuns may enter this creek at low water and from this navigation, and thereby the trade ride at anchor at Crow Pool, but no further of Pembroke is checked’ (Pughe 1818, without help of the tide’; Crow Pool lay on 362). Its maritime trade was overwhelm-

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16291 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2 ingly coastal and, during the Middle Ages, Pembroke had enjoyed something of a wine from France was shipped from the recovery after the Restoration, with the transit port of Bristol, in vessels that like general growth of trade in the Bristol their crews were both small and local Channel area (Howell 1987, 28), but Hav- (Howells 2002, 471). By the thirteenth erfordwest increasingly took the bulk of century Pembroke, as a port, had been maritime traffic during the course of the superseded by Tenby; where the annual eighteenth century (Donovan 1805, 336; tolls from Pembrokeshire’s ports were re- Howell 1987, 294; Soulsby and Jones corded during the fourteenth to sixteenth 1975, 24).25 Pembroke’s trade was still a centuries, in most cases none were taken at low-key affair in c .1800, when a quay Pembroke ‘because no boats called in the seems to have been first established, in said port this year’ (Owen 1918, 112, 127, timber (Thompson 1983, 50 fig.), finally 143, 148). The sixteenth-century ‘Welsh allowing vessels to unload directly at the Port Books’ record only eight vessels be- town: in 1805 Edward Donovan hoped longing to Pembroke between 1566 and that, as a port, Pembroke ‘might one day 1603 – all small craft between seven and rise to consideration’ (Donovan 1805, sixteen tons – when trade with the town is 336). The present masonry quay was built mentioned on a mere six dates (Lewis in 1818 (Evans 1981, 84), but the narrow, 1927, 71-89, 134, 164, 198, 217, 226, 313); winding channel meant that Pembroke re- it features far more prominently as a mer- mained inaccessible to larger vessels (see chant seat than a port.24 Pembroke is not above) and it was soon eclipsed by Pem- listed among George Owen’s ports and broke Dock (Soulsby and Jones 1975, 24). havens in the county (Owen 1906, 349-50), Walter Marshal begun the lighthouse at while his description of Pembroke town, of Hook Point, Co. Wexford, which was still 1595, makes no mention of any maritime under construction in 1247 (Cal. Pat. activity (Owen 1897, 557-8); it is similarly Rolls 1232-47, 500; Murtagh 2016, 230; absent from John Leland’s account of his Sweetman 1875, 429); its foundation may visit in c .1538 (Smith 1906, 115-17), in therefore coincide with his visit to Lein- which Tenby, Fishguard, Carmarthen and ster in 1244-5 (see above). It served the Kidwelly are by contrast described as ports important Marshal port of New Ross, and (Smith 1906, 58, 61, 65). the River Barrow which was navigable for Moreover, no quay or wharfage is shown over 60 miles, connecting the leading at Pembroke on John Speed’s plan of 1610 Marshal towns at Kilkenny and Carlow (Fig. 39), nor in the wealth of artist’s views with the sea. Testifying to the importance spanning the late seventeenth century of the waterway, the lighthouse is still through to 1800 when the area of the pres- operational after nearly eight centuries ent quay – the only practical mooring sea- (Murtagh 2016, 223). ward of the town bridge – was a pool or The Pembroke River is a different matter embayment in which the river lapped up altogether. While the inlet through which against the town walls. A number of views, it flows is a ria , or drowned valley, the including the Buck prints of 1740 and river itself is a sluggish, meandering 1748, depict lighters drawn up here at low stream which, in sharp contrast to the Riv- tide (Fig. 40); an inlet beneath the Wogan er Barrow, extends inland for a mere 1½ is also shown.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16292 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2

Fig. 40 – Watercolour of Pembroke Castle from the north by Francis Place, 1678. Mudflats occupy the area of the present quay (National Library of Wales). miles without any further settlement along in the Year 1804, and the Four Preceding its banks. No lighthouse was therefore Summers, 2 (London: Edward Donovan). thought necessary at, or near Pembroke Green, F. (ed.), ‘Carmarthen Castle. A col- during the post-medieval period, when Mil- lection of historical documents relating to ford Haven was otherwise well-supplied Carmarthen Castle from the earliest times with navigational aids. All the above fac- to the close of the reign of Henry VIII’, tors, taken separately or together, would West Wales Historical Records, 4 (1914), militate against the suggestion that Mar- 1–71. shal, or any its lords, established a light at Pembroke Castle.26 Lewis, E. A., 1927 (ed.) The Welsh Port Books, 1550-1603 (London: Cymmrodor- Additional references ion Record Series 12). Primary sources Morris, L., 1748 Plans of Harbours, Bars, Ballard, A. (ed.), 1913 British Borough Bays and Roads in St George’s Channel Charters 1042-1216 (Cambridge Universi- (London: Lewis Morris). ty Press). Orpen, G. H. (ed.), 1892 The Song of Calendar of Patent Rolls, Hen. III 1232- Dermot and the Earl: An Old French 1247 (London: HMSO, 1906). Poem (Oxford: Clarendon Press). Calendar of Patent Rolls, Rich. II Vol. 1 Owen, H. (ed.), 1897 The Description of 1377-1381 (London: HMSO, 1895). Pembrokeshire by George Owen of Hen- llys, Lord of Kemes 2 (London: Cymmro- Cole, G. D. H. and Browning, D. C. (eds), dorion Record Series 1). 1962 Daniel Defoe: A Tour through the Whole Island of Great Britain, 2 (London: Owen, H. (ed.), 1906 The Description of J. M. Dent). Pembrokeshire by George Owen of Hen- llys, Lord of Kemes 3 (London: Cymmro- Donovan, E., 1805 Descriptive Excursions dorion Record Series 1). through South Wales and Monmouthshire

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16293 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2 -References

Owen, H. (ed.), 1918 A Calendar of Pem- Crouch, D., 2006 ‘Chepstow under the brokeshire Records 3: Pembroke (London: Marshals’, in R. Turner and A. Johnson Cymmrodorion Record Series 7). (eds), 43-50. Pughe, W. O. (ed.), 1818 The Cambrian Crouch, D., 2014 ‘Earl Gilbert Marshal Register, 3 (London: E. & T. Williams). and his mortal enemies’, Historical Re- Smith, L. T. (ed.), 1906 The Itinerary in search 87/237, 393-403. Wales of John Leland in or about the years Dixon, P., 1998 ‘Design in castle-build- 1536-1539, 3 (London: George Bell and ing: the control of access to the lord’, Sons). Château Gaillard 18, 47-57. Thompson, M. W. (ed.), 1983 The Journeys Gregory, J. and Liddiard, R., 2016 ‘Visi- of Sir Richard Colt Hoare through Wales ble from afar? The setting of the Anglo- and England, 1793-1810 (Stroud: Alan Sut- Norman donjon’, in J. A. Davies, A. Ri- ton). ley, J-M. Levesque and C. Lapiche (eds), Walker, R. F. (ed.), 1989 ‘Henry II’s Char- 147-58. ter to Pembroke’, Bulletin of the Board of Howell, D., 1987 ‘Society, 1660-1793’, in Celtic Studies 36, 132-45. B. Howells (ed.), Pembrokeshire County Williams ab Ithel, J. (ed.), 1860 Annales History 3: Early Modern Pembrokeshire Cambriae, (London: Rolls Series). (Haverfordwest: Pembrokeshire Histori- cal Society), 256-298. Secondary sources (unpublished) Howells, J., 2002 ‘Pembroke’, in R. F. Evans, B. H. (ed.), 1981 ‘Buildings of Spe- Walker (ed.), Pembrokeshire County His- cial Architectural or Historic Interest: Pem- tory 2: Medieval Pembrokeshire (Haver- broke and Pembroke Dock’ (Cardiff: fordwest: Pembrokeshire Historical Welsh Office). Society), 468-76. Soulsby, I. N., and Jones, D., 1975 The Ludlow, N., 2017 ‘Monkton Old Hall: a Archaeological Implications of Redevelop- fifteenth-century manorial courthouse?’, ment in the Historic Towns of South Pem- Journ. Pembs. Hist. Soc. 26, 9-29. brokeshire District (University College Cardiff). Ludlow, N., in prep. Pembroke: Castle, Town and Priory. Secondary sources (published) Marshall, P., 2016 ‘Some thoughts on the Blair, J., 1993 ‘Hall and Chamber: English use of the Anglo-Norman donjon’, in J. A. Domestic Planning 1000-1250’, in G. Davies, A. Riley, J-M. Levesque and C. Meiron-Jones and M. Jones (eds), Manori- Lapiche (eds), 159-74. al Domestic Building in England and Northern France (London: Society of An- Morgan, O. and Wakeman, T., 1854 Notes tiquaries Occasional Papers 15, 1-21). on the Architecture and History of Caldi- cot Castle, Monmouthshire (Newport: Bouet, P., 2016 ‘Châteaux et résidences Caerleon Antiq. Assoc.). princières dans le Tapisserie de Bayeux’, in J. A. Davies, A. Riley, J-M. Levesque and Pré, M., 1961 ‘Le château de Laval’, Con- C. Lapiche (eds), 135-46. grès Archéologique de France 119, 353- 72.

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16294 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2 - References & Notes

Renn, D., 2012 ‘Arrow-loops in the Great 4 The wall-walk was raised by 1.8m at its junc- Tower of Kenilworth Castle: Symbolism tion with the Dungeon Tower, when the latter vs Active/Passive ‘Defence”, Castle Stud- was added (see Ludlow 1990, 28 fig.). ies Group Journ. 25, 175-9. 5 In contrast with the donjon bridge at e.g. Renn, D. and Avent, R., 2001 Flint Tretower, Brecs. Castle/Ewloe Castle (Cardiff: Cadw). 6 David King speculated that this feature, now truncated, might be a postern (King 1978, Rowlands, I. W., 1996 ‘William Marshal, 106), but the external opening is only 0.8m Pembroke Castle and the historian’, Châ- wide, while it is not clear how it might have teau Gaillard 17, 151-5. been accessed from the interior (the upper Schellenberg, B. A., 1995 ‘Imagining the levels were ‘restored’ in the 1880s). Its over- all form is more consistent with that of a Nation in Defoe’s A Tour Thro’ the Whole latrine outfall, or a pit-latrine with an external Island of Great Britain’, English Literary opening for periodic cleaning. History 62/2, 295-311. 7 While prisoners were recorded in the donjons Stacey, N., 2011 Framlingham Castle at, for instance, Rouen in early 1203 (Bouet (London: English Heritage). 2016, 141-2; Holden et al. 2004, 137), and Turner, R., Jones-Jenkins, C. and Priest- Compiègne in 1204 (Corvisier 2002, 14), they were presumably of high status like Ran- ley, S., 2006 ‘The Norman Great Tower’, ulf Flambard, who was held in the White in R. Turner and A. Johnson (eds), 23-42. Tower, London, in 1100 (Marshall 2016, Wood, M., 1983 The English Mediaeval 170) – perhaps not entirely incompatible with House (London: Bracken Books). the donjon’s intended use – and therefore not kept in the basement. Notes 8 Fireplaces in pre-existing cylindrical donjons, 1 There are however two or three gaps in Mar- in Angevin France, were generally either never shal’s itinerary in 1213-14, and during winter present or have not survived; the fireplace at 1215-16, which were potentially long enough Le Brandon is truncated, but corbels suggest it for a visit. King John spent two weeks at was hooded (Corvisier 1998(3), Fig. 58). Pembroke in June 1210, preparing for his 9 Irish expedition, but seems to have stayed Although the presence of a projecting timber aboard his flagship anchored in Milford Ha- porch at Pembroke is indicated by sockets, it ven, from which all his acts and letters were would not be large enough to act as a waiting issued (Hardy 1844b, 172-7; see below). area. 10 2 As Pamela Marshall notes, ‘we may have to The Spanish term for donjon is ‘Tower of come to terms with a psyche that was prepared Homage’ (torre del homenaje). to spend a huge amount of money on a building 11 Only at this level are the rear-arches slightly that was used infrequently, perhaps only once, pointed, as in the external window surrounds, or even not at all’ (Marshall 2016, 171). though a change of masons may be indicated 3 The latrines in the curtain at Dundrum, simi- rather than specific function or status (see larly thought to have been accessed via a above). bridge from the donjon, were built to serve a 12 The deer-park at Monkton, however, was hall that was demolished when the donjon probably a creation of the fifteenth century was built (McDonald 2017, 267). (Ludlow 2017, 24).

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16295 STUDIES GROUP JOURNAL NO 32: 2018-19 William Marshal, Pembroke Castle and Angevin design - Appendix 2 - Notes

13 Similar impressions can be seen in the win- al. 2006, 35-6). Richard Marshal held a grand dow embrasures. tournament there in 1234 (Crouch 2006, 49), while Gilbert Marshal spent extensively on 14 Tom McNeill is however of the view that, on the castle (Coldstream and Morris 2006, 112). those rare occasions that a castle might come under siege, this strict restriction might be 21 It will be argued in Ludlow (in prep.) that waived (McNeill 2006, 123-4, 126). Pembroke was in King John’s hands 1207- c.1213 (also see Rowlands 1996, 155) 15 The Younger Marshal east curtain at Cilger- ran Castle, Pembs., from the 1220s, which 22 We see the same practice at Carmarthen provides extensive views over a wooded riv- where, by the 1430s, goods were trans- er valley, was parapetted on both sides and shipped onto lighters at the anchorage known possibly roofed (Hilling 2000, 21), possibly as Green Castle, some 3km downstream from functioning in a similar way. the town (Green 1914, 48), due to the pro- gressive silting described by Leland in the 16 Looped parapets had been introduced by the 1530s (Smith 1906, 61); the town quay had 1190s, e.g. at Château Gaillard and Fram- hitherto been accessible to sea-going vessels. lingham Castle, Suffolk (Hislop 2016, 160; Stacey 2011, 5, 10-11). 23 William Marshal II’s charter to Haverford- west demanded that ‘ships coming with mer- 17 An account from 1386 mentions ‘timber brat- chandise to Milford do not go elsewhere . .. tices (britagia de meremio) for the protection than to Pembroke and Haverford’ (Ballard of the walls of [Pembroke] castle, built on the and Tait 1923, 243), so that customs could be walls of the said castle’ (Owen 1918, 106). It collected at the Cross. is difficult to be certain what is meant; the donjon was, by 1386, far from being a for- 24 Although maritime legislation was re-organised ward defence of the castle, and unequivocal in 1559, and customs houses were established at evidence for hourding is absent elsewhere. Pembroke and Carmarthen (Lewis 1927, ix, As it appears in an account of repairs, it may xv), customs appear to have been collected, as instead relate to protection from the elements in the twelfth century, at Crow Pool. during the repair or rebuilding of walling. 25 Daniel Defoe’s claim that 200 ships belonged 18 The former is the earliest known depiction of to Pembroke town in the 1720s (Cole and Pembroke Castle, forming the background to Browning 1962, 57) must be treated with great the initial letter ‘T’ of George Owen’s De- scepticism; it has been pointed out that, for scription of Milford Haven (Owen 1897, 557). Defoe, the ‘discourse of improvement’ over- rode ‘all other rhetorics’ and that, particularly 19 The concept of the attic floor sits uneasily in Wales, he transformed visible deficits into within the donjon tradition in which height apparent surpluses (Schellenberg 1995, 300). was related to status, with the ‘lordliest areas at the top’ (Gregory and Liddiard 2016, 148); 26 Not only would a light at the summit of the the structural evidence confirms that it is Pembroke donjon be incompatible with its sta- secondary at Pembroke. tus and suggested purpose, but it would un- doubtedly have been furnished with a second 20 Marshal’s widow Isabel and son Anselm spiral stair – separate from that reserved for were both resident at Chepstow when they private use – to serve the keepers, irrespective died in 1220 and 1245 respectively and, like of their social standing cf. the segregation, with Walter Marshal in 1245, were buried at Tin- separate stairs, seen in the donjons at Appleby, tern Abbey (Crouch 2006, 47 n. 14; Crouch Carrickfergus and Trim (McNeill 2006, 124-6). 2015, 36). The biographical Histoire, com- posed in Marshal’s honour in the mid-1220s, was probably written at Chepstow (Holden et

THE CASTLE STUDIES GROUP JOURNAL THENO 29: CASTLE 2015-16296 STUDIES GROUP JOURNAL NO 32: 2018-19