Chapter XII: The Age of Transition: the Fortified Dwelling-House (2)

[40] Domesday, i., f. 23; “Castrum Harundel Tempore Regis Edwardi reddebat de quodam molino xl solidos,” etc. “Castrum Harundel,” however, applies to the town, not the castle; and it does not follow that the name was given to the town before the Conquest.

[41] Ord. Vit., _Hist. Eccl._, iii. 14; “id castellum situm est in acutissima rupe mari contigua.” The phrase may be used generally to describe a site which, in Ordericus’ own day, had become famous for its castle.

[42] Ord. Vit., _Hist. Eccl._ iv. 4.

[43] The Tower of London was outside the east wall of the medieval city. Baynard’s castle was at the point where the west wall approached the Thames.

[44] Ord. Vit., _op. cit._, iv. 4; “pinnas ac turres ... in munimentis addebant vel restaurabant ... Portæ offirmatæ erant, densæque turbæ in propugnaculis et per totum muri ambitum prostabant.”

[45] The foundation of these castles is noted by Ord. Vit., iv. 4, 5.

[46] The word “bailey” (_ballium_) literally means a palisaded enclosure. The synonym “ward,” applied to the various enclosed divisions of a medieval castle, means a guarded enclosure. The term “base-court” (_basse-cour_) is also applied to the bailey.

[47] It should be noted that at York there were not two distinct _burhs_ or fortified towns, such as are found in the earlier cases. The river passed through and bisected the _burh_, which was surrounded by an earthen bank, save at the point where the Foss formed the boundary of the city.

[48] Domesday, i. 248 _b_.

[49] An example of this is the fine earthwork at Lilbourne, in Northamptonshire. There are many other instances, and the lesser bailey at Clun partakes of this character.

[50] There are cases, of course, which give rise to perplexity. Thus at Earls Barton, in Northamptonshire, the famous pre-Conquest church tower stands on a site which appears to be within the original limit of the ditch of the adjacent castle mount. It is doubtful, however, whether the mount was ever ditched on this side; and the church does not encroach upon the mount.

[51] Cæsar, _De Bell. Gall._, vii. 73; “huic [vallo] loricam pinnasque adiecit, grandibus cervis eminentibus ad commissuras pluteorum atque aggeris, qui ascensum hostium tardarent.” See p. 60 below.

[52] See Enlart, ii. 494.

[53] Domfront, however, on its rocky site, may, like Richmond, have been surrounded by a stone wall from the first.

[54] L. Blanchetière, _Le Donjon ou Château féodal de Domfront (Orne)_, 1893, pp. 29, 30.

[55] See note above, p. 45.

[56] Ord. Vit., iii. 5.

[57] The essential portions of these texts are quoted by Enlart, ii. 497-9.

[58] The “lesser donjon” at Falaise, which contained the great chamber, is a rectangular projection of two stories from the great donjon.

[59] Mrs Armitage in _Eng. Hist. Review_, xix. 443-7.

[60] Ord. Vit., viii. 12; “fossis et densis sepibus.”

[61] _Ibid._, viii. 24; “Hic machinas construxit, contra munimentum hostile super rotulas egit, ingentia saxa in oppidum et oppidanos projecit, bellatores assultus dare docuit, quibus vallum et sepes circumcingentes diruit, et culmina domorum super inhabitantes dejecit.”

[62] Ord. Vit., viii. 13; “Callidi enim obsessores in fabrili fornace, quæ in promptu structa fuerat, ferrum missilium callefaciebant, subitoque super tectum principalis aulæ in munimentis jaciebant, et sic ferrum candens sagittarum atque pilorum in arida veterum lanugine imbricum totis nisibus figebant.”

[63] See J. H Round, _Castles of the Conquest_ (_Archæologia_, lviii. 333).

[64] _Adulterinus_ = spurious, counterfeit.

[65] Cæsar, _Bell. Gall._, vii. 68 _seq._ Alesia, near the modern village of Alise-la-Reine, is in the Côte d’Or department, some 36 miles N.W. of Dijon.

[66] Cæsar, _De Bell. Civ._, ii. 1 _seq._

[67] A detailed account of this siege is given by Oman, _Art of War_, pp. 140-7.

[68] Enlart, ii. 413, 414.

[69] Ord. Vit., vii. 10.

[70] _Ibid._ “Rex itaque quoddam municipium in valle Beugici construxit ibique magnam militum copiam ad arcendum hostem constituit.”

[71] _Ibid._, viii. 2.

[72] _Ibid._, viii. 23; Roger of Wendover.

[73] Thus Henry I., in his wars with Louis VI., conducted one blockade by building two castles, which the enemy called derisively Malassis and Gête-aux-Lièvres (Ord. Vit., xii. 1). So also (_ibid._, xii. 22) his castle of Mäte-Putain near Rouen. Many other instances might be named.

[74] Oman, _Art of War_, pp. 135, 139: his authority is Guy of Amiens, whose poetical rhetoric, however, may not be altogether accurate in description.

[75] Ord. Vit., viii. 24. _Cf._ viii. 16, where Robert of Normandy, another great Crusader, besieging Courcy-sur-Dives in 1091, caused a great wooden tower or belfry (_berfredum_) to be built, which was burned by the defenders. Robert of Bellême was also present at this siege.

[76] See below, p. 99.

[77] Suger, _Gesta Ludovici Grossi_ (ed. Molinier, pp. 63-66).

[78] Pent-houses were sometimes elaborately defended. Thus Joinville describes the large “cats” made by St Louis’ engineers to protect the soldiers who were making a causeway across an arm of the Nile near Mansurah (1249-50). These had towers at either end, with covered guard-houses behind the towers, and were called _chats-châteaux_.

[79] See the account of the sieges of Boves and Château-Gaillard by Guillaume le Breton, _Philippis_, books ii. and vii. At the siege of Zara in the fourth Crusade, after five days of fruitless stone-throwing, the Crusaders began to undermine a tower which led to the surrender of the city (Villehardouin).

[80] Abbo: see the account of the siege of Paris above.

[81] Ord. Vit., ix. 15: “Machinam, quam ligneum possumus vocitare castellum.” It was strictly a belfry (see below).

[82] _Ibid._

[83] _Cf._ the account of the operations at the siege of Marseilles (Cæsar, _De Bell. Civ._, ii. 11): “Musculus ex turri latericia a nostris telis tormentisque defenditur.”

[84] The _porte-coulis_ is literally a sliding door. Its outer bars fitted into grooves in the walls on either side. See pp. 227, 229.

[85] Vitruvius, _De Architectura_, x. 13, § 3, mentions among Roman scaling-machines, an inclined plane, “ascendentem machinam qua ad murum plano pede transitus esse posset.”

[86] Guillaume le Breton, _Philippis_, book vii. This poem is an important source of information for the wars of Philip Augustus, and for the siege of Château-Gaillard in particular.

[87] Ord. Vit., ix. 13.

[88] _Ibid._, ix. 11.

[89] _Ibid._, xii. 36.

[90] This is the usual distinction. But the use of the names varies. In Vitruvius (_op. cit._, x. 10, 11) the _catapulta_ or _scorpio_ is a machine for shooting arrows, while the _ballista_ is used for throwing stones. The pointed stakes at the siege of Marseilles (Cæsar, _De Bell. Civ._, ii. 2) were shot from _ballistae_. Vitruvius indicates several methods of working the _ballista_ by torsion: “aliae enim vectibus et suculis (levers and winches), nonnullae polyspastis (pulleys), aliae ergatis (windlasses), quaedam etiam tympanorum (wheels) torquentur rationibus.”

[91] For the injuries inflicted by stone-throwing machines, see Villehardouin’s mention of the wounding of Guillaume de Champlitte at Constantinople, and of Pierre de Bracieux at Adrianople.

[92] Oman, _op. cit._, 139, quotes Anna Comnena to this effect.

[93] Stone-throwing engines and _ballistae_ alike were employed by the Saracens at Mansurah (1250), for hurling Greek fire at the towers constructed by St Louis to protect his causeway-makers (Joinville).

[94] Thus, in the first siege of Constantinople by the Crusaders (1203), Villehardouin emphasises the number of siege-machines used by the besiegers upon shipboard and on land, but gives no account of their use by the defenders. They were employed, however, by the defence, as we have seen at Marseilles; see also Chapter I. above, for possible traces of their use in the stations of the Roman wall. A special platform might in some cases be constructed for them and wheeled to the back of the rampart-walk.

[95] Such crenellations are indicated even in the timber defences at Alesia and Trebonius’ second rampart at Marseilles. They are familiar features of oriental fortification, _e.g._, of the great wall of China or the walls and gates of Delhi.

[96] This roof was sometimes gabled, the timbers, as in the donjon at Coucy, following and resting on the slope of the coping of the parapet.

[97] Sometimes, as at Constantinople in 1204 (Villehardouin), towers were heightened by the addition of one or more stages of wood. _Cf._ the heightening of the unfinished _tête-du-pont_ at Paris in 885-6.

[98] Clark, i. 68-120, gives an elaborate list of castles in England and Wales at this date. A large number, however, of those which he mentions, had been already destroyed; and many were of later foundation.

[99] Accounts of this rebellion are given by Benedict of Peterborough and Roger of Hoveden.

[100] Nottingham was a foundation of the Conqueror: Newark was not founded until after 1123.

[101] Ord. Vit., xi. 2, mentions the capture of the castle of Blyth (Blida castrum) by Henry I. from Robert de Bellême. By this Tickhill is probably meant. It is four miles from Blyth, where was a Benedictine priory founded by Roger de Busli, the first Norman lord of Tickhill, and granted by him to the priory of Ste-Trinité-du-Mont at Rouen. Ordericus, who, as a monk of a Norman abbey, was familiar with the name of Blyth priory, may have supposed the castle of Roger de Busli to have been at Blyth.

[102] See Rymer, _Fœdera_ (Rec. Com., 1816), vol. i. pt. i. p. 429: “castrum de Pontefracto, quod est quasi clavis in comitatu Eborum.”

[103] The remains are chiefly of the second quarter of the fifteenth century; but it was a residence of the archbishops as early as the twelfth century.

[104] A. Harvey, _Bristol_ (Ancient Cities), pp. 35, 116.

[105] Rob. de Monte, quoted by Stubbs, _Select Charters_, 8th ed., 1905, p. 128: “Rex Henricus coepit revocare in jus proprium urbes, castella, villas, quae ad coronam regni pertinebant, castella noviter facta destruendo.”

[106] The curtain (Lat. _cortina_, Fr. _courtine_) is a general name for the wall enclosing a courtyard, and is thus applied to the wall round the castle enclosure.

[107] Martène, _Thesaurus Anecdotorum_, iv. 47, quoted by Enlart, ii. 418. From _alatorium_ is derived the word _allure_, often employed as a technical term for a rampart-walk.

[108] Ord. Vit., v. 19: “Lapideam munitionem, qua prudens Ansoldus domum suam cinxerat, cum ipsa domo dejecit.” In this case the wall seems to have been built, not round an open courtyard, but round a house or tower. The French term for a fortified wall, forming the outer defence of a single building, is _chemise_. Thus, in a mount-and-bailey castle, the palisade round the tower on the mount was, strictly speaking, a _chemise_, while that round the bailey was a curtain.

[109] Ord. Vit., vii. 10.

[110] _Ibid._, viii. 23.

[111] _Ibid._, viii. 5. Robert, son of Giroie, “castellum Sancti Cerenici ... muris et vallis speculisque munivit.”

[112] “Herring-bone” masonry consists of courses of rubble bedded diagonally in mortar, alternating with horizontal courses of thin stones, the whole arrangement resembling the disposition of the bones in the back of a fish. The horizontal courses are frequently omitted, and their place is taken by thick layers of mortar.

[113] See _Yorks. Archæol. Journal_, xx. 132, where the evidence quoted points to the conclusion “that the doorway was not erected later than about 1075.” Harvey, _Castles and Walled Towns_, p. 85, assumes that the doorway was cut through the south wall of the tower at a later date: the evidence of the masonry is decisively against this idea.

[114] The architectural history of Ludlow castle has been thoroughly examined by Mr W. H. St John Hope in an invaluable paper in _Archæologia_, lxi. 258-328.

[115] The original design probably included an upper chamber of moderate height. There was, however, a considerable interval between the completion of the gateway and the building of the upper stage.

[116] The large outer bailey at Ludlow was an addition to the original castle, late in the twelfth century, and is contemporary with the blocking of the gatehouse entrance. Originally the castle consisted merely of the present inner ward. The outer bailey or base-court gave enlarged accommodation for the garrison, and contained stables, barns, and other offices for which there was no room in the inner ward.

[117] The explanation of this passage through the wall was long a mystery. Clark, ii. 278, recognised that it led from an outer to an inner “room,” but was puzzled by the bar-holes which showed that the doors had been carefully defended.

[118] Mr Hope thinks that it was originally intended to cover the gateway with a semicircular barrel-vault. The lower stage of the keep at Richmond has a ribbed vault with central column. This, however, with the vice, now blocked, in the south-west corner, was inserted many years after the building of the great tower on the site of the gatehouse.

[119] The string-courses of the upper stages of the tower, and the windows of the southern chamber, which was of the full height of the two upper stories, and probably formed the chapel of the castle, have further enrichment; but the detail is nowhere elaborate. See T. M. Blagg, F.S.A., _A Guide to Newark, &c._, 2nd ed., pp. 19-22.

[120] Harvey, _op. cit._, p. 98, says that Newark castle “has now no trace of a keep, and possibly never possessed one.” The gatehouse, however, may fairly be considered as belonging to the category of tower-keeps, and has one characteristic of that type of building—viz., the cross-wall which divides the upper stages, and is borne by an archway in the centre of the gateway passage.

[121] The churches of Upton, near Gainsborough, Burghwallis, near Doncaster, and Lois Weedon in Northamptonshire, are examples of this type. “Herring-bone” work occurs at Brixworth, in a portion of the tower to which a pre-Conquest date cannot safely be attributed. At Marton, near Gainsborough, it occurs in a tower of “Saxon” type, which was probably not built until after the Conquest. It is found twice at York, but the date of the so-called Saxon work in the crypt of the minster is very doubtful; while the tower of St Mary Bishophill Junior, although Saxon in type, is more likely to be Norman in date. Examples of “herring-bone” work in the churches of Normandy are found, _e.g._, at Périers and in the apse at Cérisy-la-Forêt (Calvados).

[122] The donjon of Falaise belongs to the early part of the twelfth century, and is therefore a late example of “herring-bone” work. The “herring-bone” work in the keep at Guildford is probably still later, and that in the curtain wall at Lincoln, raised on the top of earthen banks, can hardly be attributed to a very early date.

[123] It has also been noted in the tower of Marton church, near Gainsborough.

[124] The lodge which now occupies its site was built in 1815, while the present main entrance to the castle, south-west of the mount, was made in 1810, and is quite outside the original _enceinte_.

[125] See note 122 on p. 100.

[126] A curtain is said to be flanked when its line is broken at intervals by projections, so near one another that the whole face of the piece of curtain between them can be covered by the fire of the defenders stationed in them.

[127] Much of the curtain of Lancaster castle is of fairly early date. For the supposed Roman origin of the castle and its probable history, see note 354 on p. 327 below.

[128] These additions have given rise to the common theory that this hall is a work of late twelfth century date.

[129] Other examples of early stone halls will be mentioned in a later chapter.

[130] This is very noticeable in Shropshire, where a large number of parish churches, to which rectors were presented and instituted in the ordinary way, are described as free chapels in the registers of the bishops of Lichfield and Hereford during the thirteenth and fourteenth centuries.

[131] See Pat. Rolls, 18 Rich. II., pt. 1, m. 28; 3 Hen. IV., pt. 1, m. 6.

[132] Pat. 2 Edw. III., pt. 2, m. 4. The walls of this chapel, dedicated to St Peter, remain. In the fifteenth century it was enlarged as far as the west curtain by a western annexe, and in the sixteenth century it was divided into two floors, the upper floor being the court-house, and the lower floor the record-room of the court of the Marches.

[133] Pat. 2 Edw. II., pt. 2, m. 24.

[134] The word _keep_ is a comparatively modern term, unknown to medieval castle-builders, to whom this part of the castle was the _donjon_ or _dungeon_, or the _great tower_.

[135] Other important shell keeps of the normal type are at Arundel, Cardiff (114), Carisbrooke (111), Farnham, Lewes, Pickering, Totnes, and Tonbridge—the last one of the most considerable and finest examples.

[136] Clifford’s tower at York is sometimes quoted as a shell keep. It was actually a tower with a forebuilding.

[137] See Enlart, ii. 500, 676: Anthyme Saint-Paul, _Histoire Monumentale_, p. 168, gives the date 993, with an expression of doubt. Fulk the Black was count of Anjou 987-1039.

[138] Enlart, ii. 685, says “début du xiiᵉ siècle.”

[139] Ord. Vit., xii. 14.

[140] _Ibid._, viii. 19.

[141] _Ibid._, x. 18.

[142] _Ibid._, xi. 20: _adulterina castella_ is the phrase used.

[143] Enlart, ii. 710. Blanchetière, _op. cit._, 83, mentions Henry’s operations in 1123, but believes in an earlier date for the donjon.

[144] Rad. de Diceto, _Abbrev. Chron._, sub anno.

[145] _Pipe Roll Soc._, vol. i., pp. 13, 14; iv. 23.

[146] _Ibid._, i. 27.

[147] _Ibid._, i. 29, 30, 31; ii. 14; iv. 36; v. 50; vi. 57, 58; vii. 11, 12; xii. 79; xiii. 31.

[148] _Ibid._, ii. 12; v. 49.

[149] _Ibid._, iv. 35.

[150] _Ibid._, iv. 39.

[151] _Ibid._, iv. 40.

[152] _Ibid._, viii. 89; ix. 59, etc.

[153] _Ibid._, xiii. 107, 108; xv. 132; xvi. 32.

[154] _E.g._, _ibid._, xiii. 140.

[155] _Ibid._, xvi. 32; xviii. 110.

[156] _Ibid._, xviii. 110.

[157] _Ibid._, xiii. 161.

[158] _Ibid._, v. 35.

[159] _Ibid._, xix. 53.

[160] Charles Dawson, _Hastings Castle_, ii. 524.

[161] _Pipe Roll Soc._, ix. 17; xi. 18; xii. 15; xiii. 95; xv. 2; xvi. 2.

[162] _Ibid._, xviii. 16; xix. 68.

[163] _Ibid._, xix. 167; xxi. 77; see also xvi. 92.

[164] _Pipe Roll Soc._, xvi. 118, 119.

[165] _Ibid._, xvi. 141.

[166] _Ibid._, xvi. 137.

[167] _Ibid._, xix. 81.

[168] _Ibid._, xviii. 7; xix. 173.

[169] _Ibid._, xviii. 66; xix. 110; xxii. 183. Malcolm, king of Scots, yielded Bamburgh, Carlisle, and Newcastle to Henry II. in 1157; and the towers at all three places were begun within a few years of this event. That at Bamburgh is mentioned in 1164.

[170] _Ibid._, xix. 2.

[171] See evidence brought by Mrs Armitage, _Eng. Hist. Rev._, xix. 443-7.

[172] Ord. Vit., iv. 1. He calls these strongholds _firmamenta quaedam_.

[173] _A.S. Chron._, sub anno.

[174] Such cross-walls, found in the larger towers, were not merely useful as partitions between the rooms. They enabled the builders to lay their floors more conveniently, as timber of sufficient scantling for so large an undivided space was obtainable with difficulty. In case of the great tower being taken by storm, the cross-wall on each floor formed a barrier to the besiegers, shutting off the tower as it did into two halves. This is well seen, for example, at Porchester.

[175] At Norham and Kenilworth the towers are at an angle of the inner ward where the two wards are adjacent. At Porchester it is at an outer angle of the inner ward, so that two of its sides are on the outer curtain of the castle.

[176] At Hedingham and Rochester there are mural galleries above the level of the second floor, the height of which therefore corresponds to that of two external stories. Both towers are exceptionally lofty, Rochester being 113, Hedingham 100 feet high.

[177] We know from the Pipe Roll for 1173-4 that work was being done at Guildford in that year (_Pipe Roll Soc._, xxi. 3).

[178] This points to two separate dates for the structure. The earlier masonry has been attributed to Bishop Flambard, who founded the castle in 1121; the later to Bishop Pudsey, who made additions to the castle about 1157. If this is so, the history of the tower is parallel to that of Porchester—a low stone tower, possibly of the reign of Henry I., heightened in the reign of Henry II.

[179] Porchester, in spite of its great size, is a tower which was apparently built for exclusively military purposes. The floors are feebly lighted, and there is no fireplace in the building.

[180] Both these castles belong to the class of cliff strongholds which were walled from their earliest foundation.

[181] Further alterations were made in the fifteenth century, when a new stair was inserted in the north-east angle, and the outer stair against the west wall was removed.

[182] For the reason, see note 174 on pp. 121, 122.

[183] Legends about the cruelties practised on prisoners, often connected with these basement chambers, need not be believed too readily. Specially constructed prison chambers in castles usually belong to a period later than the twelfth century. On the origin of the word “dungeon” see Chapter III.

[184] See the description of the tower at Ardres in Chapter III. Such upper floors were probably divided into rooms by wooden partitions.

[185] It was thus impossible to reach the roof from the first floor without passing through the second-floor chamber—a precaution which was adopted also in the cylindrical tower at Conisbrough.

[186] Here the basement was probably used as a prison. The upper part of the original stair still remains.

[187] There are indications, however, of a second chapel in the keep itself, occupying the south-east angle of the third floor.

[188] The recently excavated chapel of the great tower of Old Sarum was a vaulted building occupying the south-eastern part of the basement of the tower itself. It was entered directly from the bailey, and had no direct communication with the first floor of the tower.

[189] Such as the so-called oratories in the fore-buildings of Dover and Newcastle.

[190] At Old Sarum, the room in the basement, west of the chapel, was probably the kitchen.

[191] _Cf._ the employment of one of the angle towers at the later castle of Langley in Northumberland as a garde-robe tower. Some of the late medieval pele-towers of the north of England, _e.g._, Chipchase and Corbridge, provide excellent examples of mural garde-robes with corbelled-out seats.

[192] Roger of Wendover, ann. 1215.

[193] See the description of the fortifications of Antioch in Oman, _Art of War_, pp. 527-9; plan facing p. 283.

[194] _Ibid._, 526-7.

[195] Enlart, ii. 504.

[196] _Ibid._, ii. 508: it is attributed to Amaury, count of Evreux (1105-37): the masonry (_ibid._, 461) is of coursed rubble with bonding-courses of ashlar.

[197] See note 161, p. 119. The keep of Orford is described at some length by Harvey, _Castles and Walled Towns_, pp. 106-111.

[198] Enlart, ii. 505.

[199] Possibly there was a trap-door in the centre of each floor: see below. All the floors are gone above the entrance stage.

[200] An embrasure is the splay or inner opening of a window. The word is also applied to the openings between the _merlons_ or solid pieces of a crenellated parapet.

[201] See pp. 217, 230, 233.

[202] It may also be noted that the practice of placing windows immediately above one another would be naturally avoided, as tending to weaken the masonry of the whole wall at these points. This is well seen in the irregular position of the numerous loops which light the vice of the donjon at Coucy.

[203] Enlart, ii. 735, gives the date of the donjon (Tour Guinette) at Etampes as about 1140.

[204] Enlart, ii. 674, gives the date of completion at Issoudun as 1202.

[205] Or _mâchecoulis_. _Coulis_ = a groove. The first part of the word is probably derived from _mâcher_ = to break or crush, and implies the purpose effected by missiles sent through those openings.

[206] Drawing in Enlart, ii. 504. Here there are two rectangular towers, with rounded angle-turrets, connected by a lofty intermediate building.

[207] The same cause undoubtedly led, at an earlier date, to the covering of Syrian churches with roofs of stone.

[208] Château-Gaillard was on the French side of the Seine, in territory purchased by Richard I. from the archbishop of Rouen.

[209] E. Lefèvre-Pontalis, _Le Château de Coucy_, pp. 48, 49, shows that the donjon forms part of the latest work undertaken by Enguerrand III., lord of Coucy, the founder of the present castle, who died in 1242: it was evidently completed about 1240.

[210] The town walls appear to be rather earlier than the castle (_ibid._, 34).

[211] On the third floor, these niches are divided into two stages and connected by an upper gallery which pierces the abutments of the vault, and surrounds the whole apartment. The method of vaulting this gallery behind the abutments, so as to give additional resistance to the masonry of the tower, is described by Lefèvre-Pontalis, _op. cit._ 94: see plan _ibid._, p. 93.

[212] In the angle-towers at Coucy, however, the stairs take the form of vices, and do not curve with the wall, although ceasing at each floor.

[213] The gabled coping of the parapet formed the central support for the sloping roof of the outer gallery and of the corresponding _coursière_ on the inner side.

[214] It stands on a promontory between two creeks at the head of the inlet known as the Pembroke river.

[215] The domestic buildings may be in part earlier, but were largely reconstructed in the thirteenth century.

[216] The tower is sometimes described as being of five stages: the dome, however, was merely a vault, and did not form a separate stage.

[217] An account of Flint castle is given by Harvey, _Castles and Walled Towns_, p. 123 _seq._ Speed’s map of Flintshire, made _c._ 1604, shows that the tower was joined to the adjacent curtain by a wall, the rampart-walk of which probably gave access to the entrance on the first floor of the tower.

[218] In 1277 the castle of Flint was a timber structure, so that the present work cannot be earlier than the end of the thirteenth century. The masonry is composed of large blocks of yellow sandstone, decayed where they are exposed to the tide. There was an outer bailey, the platform of which alone remains, with a ditch between it and the castle proper.

[219] These holes do not, however, surround the tower, so that the passage may have been only partially roofed.

[220] The keep of Launceston was probably built about the close of the twelfth century: that at Flint later, as already noted.

[221] Reproduced in _Memorials of Old Yorkshire_, 1909, opposite p. 256.

[222] _I.e._, retaining walls used to face (_revêtir_) a sloping surface.

[223] A bartizan is a small turret or lookout corbelled out at an angle of a tower or on the surface of a wall. The word is connected with “brattice” (_bretèche_); and such turrets, like the machicolated parapet, are the stone counterpart of the bratticing and hoarding of timber applied to fortresses at an earlier date.

[224] Ventress’s model of the castle, made in 1852, shows the great hall near the north-east corner of the outer ward, its west end being nearly opposite the main entrance of the castle. The outer ward nearly surrounded the small inner ward, which contained the keep.

[225] At Richmond the hall and its adjacent buildings were unusually complete for their date, and the tower-keep was not planned as a dwelling-house. None of our tower-keeps, Porchester excepted, are so purely military in character.

[226] The origin of this term is doubtful; some think it to be a corruption of “barbican”—a work covering the entrance to the house or castle proper. Large outer baileys, as at Ludlow (96) and Coucy, correspond to the “barmkins” of the north of England.

[227] At Arundel, Cardiff, and Warwick, mount-and-bailey castles which are still inhabited, the present great halls stand on sites which were doubtless occupied by the original halls built by the founders. All three were largely rebuilt at a later date, and have been further restored in modern times. Warwick was one of the Conqueror’s earliest castles; Arundel was founded before 1086, Cardiff about 1093. A large portion of the _enceinte_ at Cardiff follows the line of the curtain of the Roman station (see _Archæologia_, lvii. pp. 335-52).

[228] At Boothby Pagnell there is a cylindrical chimney-shaft very similar to that of the hall at Christchurch.

[229] The usual arrangement even in small cottages: _cf._ Chaucer, _Cant. Tales_, B. 4022 (the house of the dairy-woman in the Nonne Preestes Tale), “Ful sooty was hir bour, and eek hir halle.”

[230] The word “solar” or “soller” (_solarium_ = a terrace exposed to the sun) was used indiscriminately of any room, gallery, or loft above the ground-level of a building: _e.g._, the loft or gallery above a chancel-screen was commonly known as a “solar,” and the same word should be applied to the chamber, inaccurately called a “parvise,” on the first floor of a church porch. The word, however, is sometimes applied to a well-lighted parlour facing south, without respect to the floor on which it stands, _e.g._, the abbot’s solar at Haughmond (_Archæol. Journal_, lxvi. 307) and at Jervaulx (_Yorks. Archæol. Journal_, xxi. 337).

[231] Ord. Vit., iv. 19: “Super solarium ... tesseris ludere ceperunt.” The word “solarium” may be used, of course, in this passage with reference merely to the site of the house—_i.e._, it may mean “the first floor above the ground.” In this case William and Henry may have been playing dice in the hall itself, which, as at Christchurch, may have occupied the whole “solarium.” Robert was evidently outside the house.

[232] Bates, _Border Holds of Northumberland_ attributes the walling, etc., of Warkworth castle “on its present general lines” to Robert, son of Roger (1169-1214), who obtained in 1199, for 300 marks, a confirmation of the grant of the castle and manor from John.

[233] So called in Clarkson’s survey, made in 1567. One explanation of the name is that the tower was similar to one in Carrickfergus castle, on Belfast Lough. Clarkson describes its polygonal form as “round of divers squares.”

[234] This entrance has been blocked, and the modern entrance has been cut through a window-opening, in the adjoining bay to the west.

[235] The aisle-walls are low and the whole building is covered by a single high-pitched roof, so that there is no clerestory.

[236] The same feature occurs at the west end of the great hall at Auckland, where the daïs was placed: there are regular responds at the east end, but the eastern bay was made somewhat wider than the rest, to give room for the screens.

[237] Bishop Bek (1284-1311) probably heightened the aisle-walls and inserted traceried windows. Cosin (1660-72) rebuilt the greater part of the outer walls, renewed Bek’s windows, and added the present clerestory and roof: the splendid screen, which divides the chapel from the ante-chapel, was also part of his work.

[238] The work of this late period is attributed to Bishop Tunstall (1530-59). Cosin at a later date made additions to the chapel.

[239] At the fortified manor-house of Drayton, some fourteen miles south-east of Rockingham, the great hall is a fabric of the later half of the thirteenth century, although the date has been obscured by later alterations. The vaulted cellar at the east end of the hall (_c._ 1270) is almost intact; but the great chamber above was rebuilt about the end of the seventeenth century.

[240] As at Penshurst. The hearth-stone remains at Stokesay. At Haddon the great fireplace in the west wall was inserted several years after the hall was built.

[241] At Harlech the kitchen was at right angles to the hall, against the south curtain.

[242] The words “horn-work,” “demilune,” or “ravelin,” were applied in later fortification to flanked outworks which presented a salient angle to the field, _i.e._, on the side of attack. To such defences in the middle ages the general name of “barbican” seems to have been given.

[243] The mining operations, so successful at Château-Gaillard, were not without their own danger to the miners. In the siege of Coucy by the count of Saint-Pol in 1411, the traditional method was used to undermine one of the towers of the base-court. A party of the besiegers descended to admire the preparations. The wooden stays, however, were not strong enough to support the weight of the tower, which fell unexpectedly, and buried the men in the mine. Their remains have never come to light.

[244] These are additions to the wall, probably made soon after the building of the great cylindrical tower. The wall seems to be of the earlier part of the twelfth century, and may have enclosed the bailey from the first. No traces of a mount remain.

[245] The position of Appleby town and castle, within a great sweep of the Eden, is somewhat similar.

[246] Apartments, known as the Constable’s lodging, were on the first floor of the gatehouse: the portcullis probably descended through the thickness of the south wall of this floor, which was not pierced for a window.

[247] The common idea that molten lead was poured through these holes on the besiegers is a mere legend. This valuable material would hardly have been employed for this purpose. Powdered quick-lime, however, may have been used, with even more deadly effect.

[248] This applies, of course, to almost all vaulted towers which are cylindrical in plan, and not to gatehouse towers alone: _e.g._, the towers of the inner ward of Coucy. But, even where there is no vaulting, the interior plan of cylindrical towers is sometimes polygonal—_e.g._, in the western angle-towers at Harlech, on all floors as well as in the basement. In the eastern angle-towers of the same castle, the interior of the basements is cylindrical. Clark, ii. 73, describes these angle-towers inaccurately.

[249] The entrances to such guard-rooms, where great thickness was given to the outer wall, took the form of narrow elbow-shaped lobbies, which would be a source of difficulty and deception to an attacking force.

[250] The Black gate was built in 1247: the entrance was protected by an outer barbican in 1358.

[251] Holes in the masonry for the beam to which the pulley was fixed may be seen, _e.g._, in the gateways at Conway and Rhuddlan.

[252] At Sandal (86) there was a barbican guarding the entrance to a shell-keep.

[253] Conisbrough is virtually a castle of one ward set on an isolated hill, not unlike Restormel in Cornwall.

[254] The entrance may be compared to the more perfect plan of the barbican and platform at Conway (254).

[255] The wall of _enceinte_ at Scarborough is probably in great part the wall which defended the castle from its foundation.

[256] They appear to have been a feature of the keep at Pontefract; _cf._ also Micklegate, Monk, and Bootham bars at York, which have bartizans at the outer angles. At Lincoln the wall of the upper floor of the gatehouse, between the bartizans, presents an obtuse angle to the field.

[257] The main gatehouse (Belle-Chaise) was built under abbot Tustin (1236-64); the _châtelet_ was added under Pierre Le Roy in 1393.

[258] The fortifications of Coucy were built in the thirteenth century: the round tower in front of the Porte de Laon was superseded in 1551 by a bastion of pentagonal form. The southern gate of Coucy (Porte de Soissons) was made in a re-entering angle of the town wall: the southern gate at Conway (Porth-y-Felin) shows the same disposition. The walls of Tenby were originally built early in the reign of Edward III.: letters patent, granting murage for seven years to the men of Tenby for the construction of their walls, were issued 6th March 1327-28 (Pat. 2 Edw. III., pt. 1, m. 22).

[259] Plan in Oman, _Art of War_, opposite p. 530.

[260] The northern rampart-walk at Coucy was widened by the building of an arcade of thirteen pointed arches against the inner face of the wall, connecting a series of internal buttresses. Part of the western wall of the town of Southampton was widened, some time later than the actual building of the wall, by the addition of eighteen arches upon the outer face (293). The soffits of the arches were pierced by long machicolations—a necessary precaution in so exceptional an arrangement.

[261] In the battlement of the donjon of Coucy, each piece of solid wall between the arched embrasures is pierced by an arrow-loop (177).

[262] Viollet-le-Duc, _La Cité de Carcassonne_, p. 27, has a drawing of a similar device with an upper and lower shutter (245): the upper shutter is propped open by iron guards: while the lower is hung in iron hooks fixed in the face of the wall.

[263] _Cf._ sections of church parapets in Bond, _Gothic Architecture in England_, pp. 385-8.

[264] At Kenilworth the Water tower, on the south curtain of the base-court, has a fireplace in the basement.

[265] Garde-robes built upon arches across re-entering angles of a wall occur on each side of a large buttress in the west wall of Southampton. A similar feature occurs at the junction of the north curtain of Porchester castle with one of the Roman towers. In both cases the addition was probably made in the fourteenth century.

[266] These towers appear to be of the fourteenth century, and are therefore much later in date than the towers of the inner curtain.

[267] At Flint, Rhuddlan, and several other castles, the angle-towers were three-quarter circles, the face towards the bailey being a flat wall, on which, at Rhuddlan as at Harlech, the rampart-walk was corbelled out.

[268] These walls, pierced by seven gates and flanked by thirty-nine rectangular towers, were begun under Pope Clement VI. in 1345, and finished _c._ 1380. The rampart is reached by stairs set against the inner face of the walls. The walls of Aigues-Mortes, built 1272-5, and of Carcassonne, begun earlier and completed later than Aigues-Mortes, belong to an earlier period of fortification, corresponding to that of our Edwardian castles. Of other well-known French examples, the walls of Mont-Saint-Michel are of various dates from the thirteenth to the fifteenth century: those of Domfront are partly of the thirteenth, those of Fougères (250) of the fifteenth century, and those of Saint-Malo chiefly of the fifteenth and sixteenth centuries. The thirteenth-century _enceinte_ of Coucy has already been referred to. A list of the numerous remains of town walls in France will be found in Enlart, ii. 623 _seq._, under the name of each department.

[269] Clark, i. 460, 312, 314.

[270] The cross-wall at Carnarvon is gone.

[271] The polygonal towers which flank the great gatehouse at Denbigh had the same characteristic of obtuse angles, as can be still seen where the masonry has not been stripped from the rubble core.

[272] The threshold of the gateway was from 35 to 40 feet above the bottom of the ditch.

[273] The eastern gateway was defended in the same way.

[274] Le Krak (Kala’at-el-Hosn) was rebuilt in 1202, and held by the Franks till 1271 (Enlart, ii. 536). It was a frontier fortress of the county of Tripoli in Syria, commanding the mountain country to the east, and must be distinguished from the great castle of Kerak in Moab, near the Dead sea, built about 1140, and surrendered in 1188, “the eastern bulwark of the kingdom of Jerusalem” (Oman, _Art of War_, 541). The entrance to the castle of Kerak has been described above, pp. 240, 241.

[275] One feature of the defences of Berkhampstead is the series of earthen bastions, applied to the outer bank on the north side of the castle, probably at a date long after the foundation of the stronghold.

[276] The breadth of the “lists” or intermediate defence of the town-walls at Carcassonne varies. On the steep western and south-western sides they are very narrow, and in one place are covered by the rectangular Bishop’s tower. The ground-floor of this was a gateway, which could be used to shut off one part of the lists from each other. Of the castle and its defences more will be said later.

[277] At Newcastle the plan was nearly concentric; but the curtains of the outer and inner ward met at one point, and the outer ward was a large space, containing the domestic buildings, while the inner was nearly filled by the keep. The concentric scheme was therefore almost accidental, and no simultaneous use of both lines of defence was possible.

[278] _Cf._ the outer ditch constructed to cover the barbican at Alnwick, where there was possibly a further outwork next the town.

[279] All these gatehouses, like the gatehouse at Rockingham and others of the same period, have a central passage, flanked by round towers towards the field. Traitors’ gate, however, has an entrance of great breadth, wide enough to admit a boat from the river; and the interior is an oblong pool, without flanking guard-rooms. The round towers cap the outer angles, but are of relatively small importance in the plan. The interior pool is actually part of the ditch between the outer ward and the Thames, and the gateway is “a barbican ... placed astride upon the ditch” (Clark, ii. 242).

[280] These angle-towers appear to belong in great part to the end of the twelfth century: the Beauchamp tower is generally attributed to the reign of Edward III.

[281] These and the adjacent curtain are largely of the twelfth century: the Bloody tower was added in the fourteenth century.

[282] Thus protecting the quay outside Traitors’ gate. _Cf._ the spur-wall at Beaumaris.

[283] The thirteenth-century work in the great hall (103) of Chepstow castle is unusually elaborate for military work of the period: nowhere in English castles have we such splendour and beauty of detail as that of which there remain many indications at Coucy.

[284] It was begun about 1267 by Gilbert de Clare, eighth earl of Gloucester and seventh earl of Hertford (d. 1295).

[285] The inner ward at Kenilworth lay at all points within an outer line of defence. The outer ward, narrow on the south and west, was very broad on the east and north, and its western half was cut up into sections by cross-walls: it was also crossed by a ditch in front of the inner ward. The lake did not surround the castle, and on the north its outer defence was a very deep dry ditch.

[286] The partisans of the Despensers held Caerphilly against Queen Isabel in 1326: its defenders were granted a general pardon, from which Hugh, son of Hugh le Despenser the younger, was excepted, 15th February 1326-7 (Pat. 1 Edw. III., pt. 1, m. 29). One of the defenders, John Cole, received a special pardon on 20th February (_ibid._, m. 32). There is no record of a definite siege.

[287] The earthwork or redoubt on the north-west side of the castle is probably of this period: no definite details of the destruction of the castle are preserved.

[288] The inner buildings at Rhuddlan have entirely disappeared: traces of one or two fireplaces are left in the curtain.

[289] At Rhuddlan a passage, protected by an outer wall ending in a square tower, descended the river-bank to the water-gate.

[290] Clark (i. 217) places the date of foundation about 1295.

[291] The outer drum towers are large and imposing, though low: the inner angles are capped by smaller towers, which bear much the same relation to the gatehouses as the outer round towers to Traitors’ gate in the Tower of London.

[292] Of these towers, that on the west has an outer salient or spur, on the sides of which two bartizans are corbelled out: these are united into one, so that the outer face of the upper stage of the tower is rounded into a semicircle. The eastern tower is smaller, with a solid base: the western part of the upper portion is corbelled off in the angle between the tower and a rectangular southern projection. The upper stages of the towers completely command the approach, while the projection just mentioned would conceal a small body of defenders posted between the gateway and the spur-wall (236).

[293] This was not founded by the Crown, like the great castles of North Wales, but, like Caerphilly, was a private foundation. It passed by marriage, early in the fourteenth century, into the possession of the house of Lancaster. Some of the most important English castles—_e.g._, Kenilworth, Knaresborough, Lancaster, Lincoln, Pontefract, and Pickering—came at various times into the possession of this royal house, and, at the accession of Henry IV., became castles of the Crown as seized of the duchy of Lancaster.

[294] The stair to the rampart-walk, built against the curtain, was, however, normal in the defences of towns (241).

[295] It may be compared with the division of the outer face of the polygonal tower at Stokesay into two smaller half-octagons (306).

[296] Viollet-le-Duc’s drawing (_La Cité de Carcassonne_, p.75) shows a rampart-walk on each of the enclosing walls of this passage. He also shows the passage crossed by a series of looped barriers, so placed that each formed a separate line of defence, guarded by a few soldiers, and compelled an enemy to pursue a zigzag course through the passage. Much allusion has been already made to oblique and elbow-shaped contrivances for impeding an enemy’s progress: the antiquity of these is evident from the entrances to earthworks like Maiden Castle (see