Chapter VI

., p. 89, and Appendix O.

[1141] Piper states that the evidence of remains proves that the lower storey was a prison. But these remains probably belong to a later date, when the donjon had been abandoned as a residence, and was becoming the _dungeon_ to which prisoners were committed. The top storey of the keep was often used in early times as a prison for important offenders, such as Conan of Rouen, William, the brother of Duke Richard II., and Ranulf Flambard.

[1142] See Appendix P.

[1143] At Conisburgh and Orford castles there are ovens on the roofs, showing that the cooking was carried on there; these are keeps of Henry II.'s time.

[1144] De Caumont says these remains are on a motte, a strange statement, as they are only a foot or two above the surrounding level.

[1145] No stone castles in England are known to have been built by William Rufus; he built Carlisle Castle, but probably only in wood. As we have seen, several Welsh castles were built in his time, but all in earth and timber.

[1146] Built by Archbishop William of Corbeuil. _Gervase of Canterbury_, R. S., ii., 382.

[1147] Robert de Torigny, also called Robert de Monte, was Abbot of Mont St Michael during the lifetime of Henry II., and was a favoured courtier whose means of obtaining information were specially good. French writers are in the habit of discounting his statements, because they do not recognise the almost universal precedence of a wooden castle to the stone building, which when it is recognised, completely alters the perspective of castle dates. See Appendix Q.

[1148] The keep of Caen, which was square, was demolished in 1793. De Caumont, _Cours d'Antiquités_, v., 231. The keep of Alençon is also destroyed. There are fragments of castles at Argentan, Exmes, and St Jean-le-Thomas. The keep of Vernon or Vernonnet is embedded in a factory. _Guide Joanne_, p. 6.

[1149] The writer has also visited Vire and Le Mans, but even if the walls of the keep of Vire, of which only two sides remain, were the work of Henry I., the details, such as the corbelled lintel, the window benches, and the loop in the basement for a crossbow, point to a later period. At Le Mans, to the north of the cathedral, is a fragment of an ancient tower, built of the rudest rubble, with small quoins of ashlar; this may be the keep built by William I., which Wace says was of stone and lime (p. 234, Andresen's edition). It is difficult to examine, being built up with cottages. Domfront, like Langeais, is only a fragment, consisting of two walls and some foundations.

[1150] _Dictionnaire de l'Architecture._

[1151] _M. M. A._, i., 186.

[1152] In speaking of Falaise, of course we only mean the great square keep, and not the Little Donjon attached to it at a later period, nor the fine round keep added by Talbot in the 15th century.

[1153] Small spaces, such as the chapel, passages, and mural chambers, are vaulted in most keeps.

[1154] Colchester keep has only two storeys now, but Mr Round argues that it must have had three, as a stairway leads upward from the second floor, in the N.W. tower, and some fragments of window cases remain as evidence. _Colchester Castle_, p. 92.

[1155] The Tower and Colchester keep both have wells, which are seldom wanting in any keep. There was no appearance of a well at Langeais, but excavation might possibly reveal one.

[1156] The first castle at Corfe was built by William's half-brother, Robert, Count of Mortain. The keep of Corfe is sometimes attributed to him, but when we compare its masonry with that of the early hall or chapel in the middle bailey, we shall see that this date is most unlikely. Norwich was always a royal castle.

[1157] Part of the basement of Norwich keep has pillars, from which it has been assumed that it was vaulted; but no trace of vaulting is to be seen.

[1158] The only decoration at Corfe keep is in the oratory, which being at a vast height in one of the ruined walls is inaccessible to the ordinary visitor. Corfe was so much pulled about by Sir Christopher Hatton in Elizabeth's reign, and is now so ruinous, that many features are obscure. Norwich has suffered greatly from restorations, and from re-casing.

[1159] In 1184 Henry II. paid "for re-roofing the tower of Gisors." _Rotuli Scacc. Normanniæ_, i., 72.

[1160] It should be remembered that rude work is not invariably a sign of age; it may only show haste, or poverty of resources. It should also be mentioned that in the _Exchequer Rolls of Normandy_ there is an entry of £650 in 1184 for several works at Gisors, including "the wall round the motte" (murum circa motam). Possibly this may refer to a wall round the foot of the motte, which seems still to exist. The shell wall of Gisors should be compared with that of Lincoln, which is probably of the first half of the 12th century.

[1161] No decagonal tower of Henry I.'s work is known to exist; all his tower keeps are square.

[1162] Bower, _Scotichronicon_, v., 42. This passage was first pointed out by Mr George Neilson in _Notes and Queries_, 8th ser., viii., 321. The keep of Carlisle has been so much pulled about as to obscure most of its features. The present entrance to the basement is not original.

[1163] _M. M. A._, i., 353.

[1164] Unfortunately the greater part of these valuable _Rolls_ is still unpublished. The Pipe Roll Society is issuing a volume every year, and this year (1910) has reached the 28th Henry II.

[1165] The keeps of Richmond and Bowes were only finished by Henry II.; Richmond was begun by Earl Conan, who died in 1170, when Henry appears to have taken up the work. Bowes was another of Earl Conan's castles. Tickhill is now destroyed to the foundations, but it is clear that it was a tower. The writer has examined all the keeps mentioned in this list. It will be noticed that most of the towers took many years to build.

[1166] Henry built one shell keep of rubble and rag, that of Berkeley Castle, which is not mentioned in the _Pipe Rolls_, having been built before his accession. It is noteworthy that he did not build it for himself, but for his ally, Robert Fitz Hardinge.

[1167] The basement storey of Chester keep (the only part which now remains) is also vaulted, but this can scarcely be Henry's work, for though he spent £102 on this castle in 1159, it must have been begun by Ranulf, Earl of Chester, in Stephen's reign. Moreover, it is doubtful whether the vaulting, which is covered by whitewash, is really ancient.

[1168] Leland says of Wark, "the dongeon is made of foure howses hight," but probably he included the basement.

[1169] The earliest instance of a portcullis groove with which the writer is acquainted is in the basement entrance of Colchester. It is obvious to anyone who carefully examines this entrance and the great stair to the left of it that they are additions of a later time than William's work. The details seem to point to Henry I.'s reign. The keep of Rochester has also a portcullis groove which seems to be a later addition.

[1170] King, paper on Canterbury Castle in _Archæologia_, vi., 298. We have not observed in any English keeps (except in this single instance) any of the elaborate plans to entrap the enemy which M. Viollet le Duc describes in his article on Donjons. He was an imaginative writer, and many of his statements should not be accepted without reserve.

[1171] Wark was also an octagonal keep, but there is considerable doubt whether this octagonal building was the work of Henry II., as Lord Dacre wrote to Wolsey in 1519 concerning Wark that "the dongeon is clerely finished," and mentions that all the storeys but one were vaulted with stone. This makes it almost certain that the castle of Wark was entirely rebuilt at this time, after having been demolished by the Scots in 1460. It is now an utter ruin, and even the foundations of the keep are buried.

[1172] At Thorne, near Doncaster, where the great earls Warenne had a castle, there are the foundations, on a motte, of a keep which seems to resemble that of Orford; it ought to be thoroughly excavated.

[1173] These measurements are from Grose, _Antiquities_, v., 74.

[1174] See Payne Gallwey, _The Crossbow_, 309; Köhler, _Kriegswesen_, iii., 192. The trébuchet is first mentioned at the siege of Piacenza in 1199.

[1175] As far as we can tell, the tops of keeps having generally been ruined or altered, the common arrangement was either a simple gable, or two gables resting on a cross wall, such as all the larger keeps possessed.

[1176] Another consequence of the introduction of an engine of longer range was the widening of castle ditches. We frequently find works on ditches mentioned in John's accounts.

[1177] Payne Gallwey, _The Crossbow_, p. 3. We find it used by Louis VI. of France, before 1137. Suger's _Gesta Ludovici_, 10 (ed. Molinier). Ten balistarii are mentioned in Domesday Book, but they may have been engineers of the great balista, a siege machine. There is no representation of a crossbow in the Bayeux Tapestry. There are entries in the _Pipe Rolls_ of 6, 8, and 9 Henry II. of payments for arbelast, but these also may refer to the great balista.

[1178] _Guill. Brit. Armorici Philippides_, Bouquet xvii., line 315.

[1179] The bow brought by Richard from Palestine is believed to have been an improved form of crossbow, made of horn and yew, "light, elastic, and far more powerful than a bow of solid wood." Payne Gallwey, _The Crossbow_.

[1180] "Fenestris arcubalistaribus," Bouquet xvii., 75. The writer has never found a single defensive loophole in any of the keeps of Henry I. or Henry II. Köhler remarks that the loopholes up to this period do not seem to be intended for shooting (_Entwickelung des Kriegswesen_, iii., 409), and Clark has some similar observations.

[1181] _Dictionnaire de l'Architecture_, art. "Meurtrière."

[1182] Meyrick in his _Ancient Armour_ quotes a charter of 1239, in which the French king grants a castle to the Count de Montfort on condition "quod non possumus habere in eodem archeriam nec arbalisteriam," which Meyrick audaciously translates "any perpendicular loophole for archers, nor any cruciform loophole for crossbowmen." The quotation is unfortunately given by Sir R. Payne Gallwey without the Latin original. It is at any rate probable that the cruciform loophole was for _archers_; it does not appear till the time of the long-bow, which was improved and developed by Edward I., who made it the most formidable weapon of English warfare.

[1183] See Appendix H.

[1184] _Entwickelung des Kriegswesen_, iii., 417.

[1185] In 1186, the Duke of Burgundy caused the towers and walls of his castle of Chatillon to be "hoarded" (hordiari). This duke had been a companion of Richard's on the third crusade. William le Breton, _Philippides_, line 600. Richard's _hurdicia_ at Chateau Gaillard were two years earlier.

[1186] See Dieulafoy, _Le Chateau Gaillard et l'Architecture Militaire au Treizième Siècle_, p. 13.

[1187] The best French and German authorities are agreed about this. The holes in which the wooden beams supporting the hurdicia were placed may still be seen in many English castles, and so may the remains of the stone brackets. They would be good indications of date, were it not that hurdicia could so easily be added to a much older building.

[1188] Köhler gives the reign of Frederic Barbarossa (1155-1191) as the time of the first appearance of the round keep in Germany.

[1189] In spite of this, I cannot feel satisfied that the keep of Étampes is of so early a date. The decorative features appear early, but the second and third storeys are both vaulted, which is a late sign. The keep called Clifford's Tower at York, built by Henry III. 1245 to 1259, is on the same plan as Étampes.

[1190] This keep has been long destroyed.

[1191] Ground entrances occur in several much earlier keeps, as at Colchester (almost certainly an addition of Henry I.'s time), Bamborough (probably Henry II.'s reign), and Richmond, where Earl Conan seems to have used a former entrance gateway to make the basement entrance of his keep. See Milward, _Arch. Journ._, vol. v.

[1192] Built by Earl Hamelin, half-brother of Henry II., who died in 1201.

[1193] Viollet le Duc, art. "Donjon."

[1194] The walls of the Tower are from 12 to 15 feet thick at the base; those of Norwich 13; the four walls of Dover respectively, 17, 18, 19, and 21 feet; Carlisle, 15 feet on two sides. (Clark.) William of Worcester tells us that Bristol keep was 25 feet thick at the base! _Itin._, p. 260.

[1195] See Enlart, _Manuel d'Archæologie Française_, ii., 526.

[1196] MacGibbon and Ross, _Castellated Architecture of Scotland_, p. 159.

[1197] This type of castle was probably borrowed from the fortifications of Greek cities, which the Crusaders had observed in the East.

[1198] Conway and Carnarvon consist of two adjoining courts, without any external enclosure but a moat. Flint has a great tower outside the quadrangle, which is sometimes mistakenly called a keep, but its internal arrangements show that it was not so, and it is doubtful whether it was ever roofed over. It was simply a tower to protect the entrance, taking the place of the 13th-century barbican.

[1199] Köhler states that the gatehouse palace is peculiar to England: "only at Perpignan is there anything like it." _Entwickelung des Kriegswesen_, iii., 480.

[1200] Köhler mentions the castle of Neu Leiningen as the first example in Germany, built in 1224. _Kriegswesen_, iii., 475. Frederic II.'s castles were of this type. The castle of Boulogne, finished in 1231, is one of the oldest examples of the keepless type in France. Enlart, _Archæologie Française_, ii., 534. The Bastille of Paris was a castle of this kind. According to Hartshorne, Barnwell Castle, in Northants, is of the keepless kind, and as the _Hundred Rolls_ state that it was built in 1264, we seem to have here a positive instance of a keepless castle in Henry III.'s reign. _Arch. Inst. Newcastle_, vol. 1852. And it appears to be certain that Gilbert de Clare, Earl of Gloucester, built the keepless castle of Caerphilly before Edward came to the throne. See Little's _Mediæval Wales_, p. 87.

[1201] French archæologists are enthusiastic over the keep of Chateau Gaillard, the scientific construction of the towers of the curtain, the avoidance of "dead angles," the continuous flanking, etc. See Viollet le Duc, art. "Chateau," and Dieulafoy, _Le Chateau Gaillard_.

[1202] This type is extremely rare: Trim, in Ireland, and Castle Rushen, in the Isle of Man, are the only other instances known to the writer. Trim is a square tower with square turrets in the middle of each face; Castle Rushen is on the same plan, but the central part appears to have been an open court.

[1203] Enlart, _Archæologie Française_, ii., 516.

[1204] Martène's _Thesaurus Anecdotorum_, iv., 118. "Nulli licuit in Normannia fossatum facere in planam terram, nisi tale quod de fundo potuisset terram jactare superius sine scabello. Et ibi nulli licuit facere palicium, nisi in una regula; et id sine propugnaculis et alatoriis. Et in rupe et in insula nulli licuit facere fortitudinem, et nulli licuit in Normannia castellum facere."

[1205] The document which calls itself _Leges Henrici Primi_, x., 1, declares the "castellatio trium scannorum" to be a right of the king. _Scannorum_ is clearly _scamnorum_, banks. It is noteworthy that a motte-and-bailey castle is actually a fortification with three banks: one round the top of the motte, one round the edge of the bailey, one on the counterscarp of the ditch.

[1206] See the case of Benhall, _Close Rolls_, ii., 52b (1225).

[1207] Aldreth and Burton are omitted from this list.

[1208] M. and B. stand for Motte and Bailey; K. and B. for Keep and Bailey; O. for Outside the Town.

Transcriber's Note:

Every effort has been made to replicate this text as faithfully as possible. The errata listed at the beginning of the book have been fixed, and some minor corrections of spelling and punctuation have been made.

Italic text has been marked with _underscores_. Bold text has been marked with =equals signs=. Letters preceded by a ^caret appeared as superscripts to the end of the word. OE ligatures have been expanded.