The internal arrangement of the donjon at in : a reconsideration

178 The Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

The internal arrangement of the donjon century,7 which would only have acted as at Colchester in Essex: a reconsideration another propaganda spur to the Normans to Pamela Marshall build a castle there, underlining continuity of authority. Roman building material was Impressive in size, the great tower at liberally used in the construction of the Colchester represents one of the more enig- donjon, though it has to be borne in mind matic early donjons extant in the British that the dearth of good building stone in the Isles, for its early history, development and region, combined with a copious source of design are not entirely typical. While Wil- recyclable Roman ruins, would naturally liam the Conqueror instigated its building encourage such a course of action even around 1074 to 1076, only the ground floor without the added advantage of any sym- was completed, leaving the structure in an bolic connotations. architectural state of limbo. A quarter of a century passed before Henry I granted the The second peculiarity of the don- castle by charter to his steward Eudo, a jon surrounds the lengthy building break, man who had close personal connections clearly evidenced by crenellations fossil- with Colchester, and who may have over- ised within the walls at the top of basement 8 seen the earlier phase of building on behalf level (Phase Ia). Arguments that this was of the former king. It is likely that Eudo envisaged as a temporary measure are cer- 9 completed the Conqueror’s unfinished tainly justified, but resumption of the work on his own account, and it was not planned upper stories was delayed by about until his death in 1120 that the castle re- twenty years, which is an unusually long turned to royal ownership.1 For the purpos- hiatus. Corner turrets were built, doubtless es of this article and in the interests of to give the half-finished structure a better simplicity I shall divide the construction look (Phase Ib) but the upper levels of the regimes into just two phases, the first sub- building were not added until c.1101 10 divided, described as Phases Ia, Ib and (Phase II). This means that, although the Phase II.2 tower and its plan were started under Wil- liam I, it was not brought to fruition until The problems attendant on any the reign of Henry I, by which time some interpretation of this building are well developments were emerging in donjon known and can be summarised as follows. design. In the first place, the donjon is much larger than any other,3 which certainly accounts The third problem is the absence for some of the peculiarities in its plan. Its of all but a trace of the fabric above the first size, 46m by 33.5m can be attributed to a floor, due to a determined attempt by one desire to re-use the podium of the Roman Mr. Wheeley, beginning in 1683, to raze Temple of Claudius as a foundation for the the tower to the ground for its stone. How- donjon.4 Without doubt this decision repre- ever, he found that the effort required to sents more than simply a labour-saving take down any solid walls in the structure device, for the deliberate exploitation of made the enterprise economically unviable Roman connotations in both architecture and it was given up as hopeless within a and place for propaganda purposes has few years, though not, unfortunately, be- been noted here, as elsewhere.5 Indeed, this fore demolition of the upper storey had seems to have been a trend already preva- been achieved and other damage inflicted: lent in early eleventh-century France.6 “The tops of the towers and walls were There are also indications that the temple forced down with screws or blown up with site at Colchester had devolved into an gunpowder …... but [since] the profit did Anglo-Saxon villa regalis by the tenth not answer the charge of further demolition

The Castle Studies Group Journal No 23: 2009-10 179 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

0m 10m

Figure 1: Floor plans of the White Tower, London he [Wheeley] was forced to desist.” 11 the location of a chapel in the SE corner, the Subsequent alterations were made to the presence of mural galleries on all sides and surviving two floors when the S part was the continuation of the stairs in both of the used as a prison and later a museum, while W turrets. It is therefore possible to draw a the N part remains a ruinous shell.12 Nev- likely reconstruction of the final form of the ertheless, we have reasonably good evi- building from the surviving evidence, espe- dence for the basement and first floor and cially taking into account parallels with oth- the trace of the second floor has left enough er plans. to reconstruct certain key features, notably

180 The Castle Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

It has been assumed that Colchester, built for the same patron and initially in progress at the same time, would basical- ly have followed the plan at the White Tower, comprising two full floors above the basement.13 It will be argued here that the polite accommodation as finally built comprised an aisled Hall approached through a waiting room and rising through two sto- reys; that there was an audience chamber and private bedcham- ber at the same level as the Hall; that at top floor level a gallery gave access to a chapel and to chambers placed over the small- er rooms below. At Colchester the SE cell is, in plan, pretty much a replica of the White Tower de- sign, with a protruding apse for the vaulted chapel envisaged on the top floor and two vaulted Figure 2 Colchester: Ground floor plan stories below this to support its structure (compare Figure 1 and Figure 2). The chapel was almost certainly building, with a lateral cross-wall and designed to stand proud of the roofline, as chambers across the whole width of the S in the original White Tower arrangement.14 section; at the White Tower this occurred However, the scale of the enterprise made only where the chapel cell was situated. any further reproduction of the White Tow- Even the cross-wall was not straightfor- er plan problematical and there were bound ward, for the chambers differed in size, so to be differences from the outset. In the their N walls dog-legged to accommodate first place, the basement floor was not set the layout of the first floor. Despite the at ground level, as was normal, being al- outline footprints of Colchester and the ready raised some 3m by the re-use of the White Tower having been so celebrated for temple podium. More importantly, the size their similarity,15 it must have been clear to of the building required two spine walls, its builders that Colchester could not sim- structurally necessary because of a roof ply reproduce the Conqueror’s other great span that needed a triple pile rather than the donjon on this particular site. The layout of more common dual pile plan (see Figure 2). the basement reflects this from the very start and even more essential differences This left a long central cell that was emerged in the finished plan, delayed by difficult to light. In addition, the scale of two decades. By the time building resumed the building did not lend itself to accom- c.1101, work on Henry I’s great tower at modating one hall through its immense Norwich was advancing, marked by a de- length, as was normal. So the spine walls termined change of plan over that envis- did not even run the full length of the

The Castle Studies Group Journal No 23: 2009-10 181 The internal arrangement of the donjon at Colchester in Essex: a reconsideration aged by the architects of its instigator, secure, possibly envisaged as strong-rooms. William Rufus.16 There are signs of cross- The N wall of the central cell was deliber- fertilization between Henry’s modified ately set proud of the line of its SE counter- plan for Norwich and a revised plan for part, while the N wall of the next cell to the Colchester, which probably veered closer W was set back again. Thus a salient corner to the contemporary royal model than that protruded into the westernmost of the long which the Conqueror’s master masons had northern cells, made more pronounced by had in mind. the position of the westernmost spine, which does not align with the W wall of the Analysis of the Plan: Ground Floor (See vaulted central cell on the S side. This ar- Plan: Figure 2) rangement of supporting walls is very sig- Although circumstances led to its constitut- nificant in interpreting the first floor. ing the entire accommodation throughout In the N section of building the E Phase I (c. 1074-1101, the poor provision spine wall still stands but the W one was of lighting, no more than ventilation loops, demolished by Wheeley. Where this joined shows that the vast majority of the ground the N wall of the donjon, however, there is floor space was envisaged, as was normal, no scarring on the lower 2m, suggesting as a basement designed mainly for storage. that it was an arcade rather than a solid wall. The survival of the ground floor plan fortu- This would also account for this being the nately provides us with the basis for suppo- only wall at basement level that Wheeler sition about the arrangement of the lost managed to demolish.17 The interpretation superstructure, where the polite accommo- of the first floor works best with the wall dation was located. above also reconstructed as an open arcade. There were a total of six cells at Superimposed arcades instead of a central basement level: three long rectangles in the spine were used at Beaugency (Loiret) as N section to support a triple-piled roof and early as c.1015.18 Other examples of the use three smaller ones in the S section. The of short arcades as supporting members in roofing of the building must have been basements are known from Ivry-la-Bataille more complicated than at the White Tower, (Eure) c.100019 and Nogent-le-Rotrou with a second set of pitched roofs across (Eure-et-Loir) c.1005-1028).20 At Norwich, the S section. It is possible that the middle c.1100, a revision of the original plan incor- section on the S side was roofed in con- porated the use of an arcade in the junction with the chapel while the western- basement21 and this may have been directly most one had its own small roof that could influenced by Colchester. have been pitched, or perhaps more proba- An outstanding departure from the bly pyramidal. The roof as a whole would White Tower plan is that the entrance was have required a complicated system of gut- located on the ground floor although, be- ters, which could easily have accommodat- cause of the height of the temple podium, ed a cistern for water collection. this was well above natural ground level. The SE cell of the ground floor The main door, at the W end of the S wall, was determined by the shape of the chapel was reworked in Phase II22 and its elaborate and, like its counterpart on the next floor, decoration reflects architectural fashion at was vaulted in preparation for the ultimate that time23 (Figure 3). thrust of the chapel vaults. A second barrel- Nevertheless, scholarly opinion vaulted cell, with equally thick walls, was agrees the doorway had been in this posi- placed immediately to the W, so that the tion from the start. It was widened to com- ground floor potentially had two enclosed plement a new forebuilding belonging to rooms that would have been particularly

182 The Castle Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

Figure 3 Colchester: Main doorway in the S wall.

Phase II, an improvement upon an elev- First Floor (See Plan: Figure 4) enth-century timber original, both found 24 As one would expect, all the windows on by excavation. The new doorway incor- the first floor were larger than those in the porated a portcullis, for which a recess was basement, denoting more polite use. The provided in the Phase II chamber above. arrangement of the S side of the first floor This is another departure from the White has been partly obscured by modern altera- Tower plan, where defences were incon- tion and use of the building. Nevertheless, spicuous. However, the Colchester door- the surviving evidence permits some recon- way does resemble the entrance to the struction. The SW stair would have deposit- Tower in being very grand in its dimen- ed visitors into a room in the SW corner. sions. It led into an entrance vestibule in While the equivalent space below is divided the SE corner where there is a semi-circu- into two small cells, on the first floor it is lar niche in the wall, which may have been more likely that these two spaces were com- a place for a guard to stand; such niches are 25 bined into a single chamber, as at present. apparent in the great tower at Norwich. At basement level the immediate entrance No doubt anyone of status would have vestibule had no need to be large, but the been ushered straight up to the first floor room above would have served as a check- via a wide spiral stair in the SW turret off 26 ing-out area, waiting room and antecham- this vestibule. The stair shows signs of ber to the hall and was more likely to have reworking during resumption of the works been a reasonable size. The provision of in Phase II, at the same time that the door- 27 two garderobes in the adjoining turret way was widened. room28 supports the interpretation of its use, for these facilities, often two, are commonly

The Castle Studies Group Journal No 23: 2009-10 183 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

Figure 4 Colchester: First floor plan found associated with waiting rooms. The Hall winding gear for the portcullis would have On the N side of the donjon the arrange- been on show in a recess in the S wall next ment of supporting walls in the basement to the doorway and a much-altered recess W of the solid E spine was deliberately in the S wall may represent the position of disposed to create a narrow central section, 29 a fireplace, flanked by two windows, a even though it would have been easy to classic Romanesque room arrangement. align the W spine with the salient corner of From the waiting room one entered the the central vaulted cell in the basement, hall. There is evidence for a doorway in the almost equalising the available space. This NE corner, conveniently adjacent to a deliberate division into a wider and smaller semicircular wall niche of the sort that space, added to the evidence for an arcaded would accommodate a sentry, also found in spine at basement level, strongly argues for the ground floor entrance vestibule. a hall with an aisle. This solution would also solve the otherwise insurmountable problem of lighting an enclosed central

184 The Castle Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration space. The doorway from the antechamber 29m long by 18m wide maximum) that it shows that one entered through the aisle. required two fireplaces, both set into the W As the S wall of the hall main section has wall. The public nature of the room is un- gone we cannot be sure that there was not derlined by the provision of two more gard- another doorway directly from the ante- erobes in a turret at the NW angle.31 chamber, but this is unlikely. As already At the W end of the N wall was an noted, the supporting basement wall was external doorway to the hall; the marks of deliberately set back towards the S com- steps to reach it are visible on the exterior pared with its neighbouring cell where, N wall32 and it was overlooked by a protect- again, the easy solution would have been to ing arrow-loop in the adjoining NW build the N wall of the basement S cells in turret.33 However, this was never apparent- a continuous line. This very deliberate lay- ly planned as a main entrance. It may have out at basement level was clearly part of the been envisaged as a private entrance for the Phase I plan of , use of the lord: a comparable arrangement designed to create a recess at the S end of is found in the donjons at Loches (Indre-et- the hall nave that is reminiscent of the Loire), built 1013-35 for the count of recess at the S end of the first floor chamber Anjou34 and at Beaugency (Loiret on the E side at the White Tower (see c.1015).35 The donjon at Falaise, built by Figure 1, first floor). This looks like a place Henry I c.1120 also had one of these ‘back to set a throne. Although there was a door- doors’ and this is likely to have been the way in the comparable recess at the White direct model for one at Norwich. Examples Tower, this connected with a private, en- carry on into the second half of the twelfth closed room beneath the chapel, a vaulted century, for example Semblençay (Indre-et- space that may have provided a withdraw- Loire).36 Frequently these doorways com- ing room or dressing space for the king. mand a view of the castle courtyard (as was The more public nature of the antechamber the case at Colchester)37 or of the township at Colchester, which formed part of the (as at Norwich),38 often both. They might entry arrangement, is less likely to have have been used as ‘appearance’ doorways, been directly connected with the recess at where the lord could make an appearance to the high end of the hall. The general ar- a wider assembly than could be admitted to rangement of entering the hall via the aisle, the hall, either gathered in the courtyard or with the king’s throne set at the S end of an even to be seen by the wider populace be- adjoining section of the hall beyond an yond the castle walls. The presence of door- arcade is a very close approximation to the ways or large window openings in surviving arrangement of the second floor numerous Romanesque donjons attests to at Rochester, built a generation later. One such usage, possibly originally inspired by yearns for Mr. Wheeley to have left the Carolingian palace architecture and the arcade, if only so that we could see whether practice of kings at this period.39 there was a wider arch at the throne end, as is the case at Rochester, where the architec- Chambers tural detail strongly suggests ceremonial A doorway in the N section of the E wall of 30 use. The hall was lit by a series of paired the hall led into a long narrow space in the windows along the W wall and four more easternmost cell (28m by 6m). It is possible in the N wall. There was most likely anoth- that this was a complete withdrawing cham- er set of windows set at a higher level in ber but, despite this space being nowhere wall galleries (discussed below), another near as large as the adjoining hall, there parallel with the ceremonial space at Roch- were two fireplaces. This, combined with ester. The hall and its accompanying aisle the proportions of the space, suggests that it formed such a vast space (approximately

The Castle Studies Group Journal No 23: 2009-10 185 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

window and the doorway into a gar- derobe, which also had a vaulted closet;44 all of which suggests more personal use, a private chamber that may have been used as a bedchamber as well. There is some dispute con- cerning access to the heavily vaulted sub-chapel to the S of this chamber. The 1922 Royal Commission report states very clearly that it was ac- cessed from this chamber through the NW apsidal recess, the doorway having a drawbar. The commission- ers asserted that another doorway in the W wall of the vaulted room (marked as a dotted line in Figure 4) was a later insertion.45 Crummy, on the other hand, took this last to be original.46 On the whole, the evi- dence of the Royal Commission is more convincing, especially the Figure 5 Colchester: Second Floor or Gallery Level plan mention of the drawbar. Access to the vault from within the lord’s was divided into two rooms by a timber chamber makes sense if it was used as a partition to form an outer and inner cham- secure robing room or treasury, whereas 40 ber, for there is no second doorway. The there can be no very obvious reason why first room, connected immediately with the this secure space should have connected hall, would be more likely used as a with- directly with the waiting room. drawing room for private audiences and therefore is likely to be the larger. The Second Floor (see plan, Figure 5) arrangement of hall and adjacent chamber, While evidence for the missing upper floor used at the White Tower, is commonly at Colchester becomes more fragmentary, found in earlier donjons in France41 and it some elements can be reconstructed with was also adopted in Henry I’s donjons at certainty, including a chapel in the SE cor- Norwich and Falaise. Windows flanking a ner. Its walls stand to a maximum of 1.2m, fireplace seems to have been the preferred enough to show that they were thin by com- arrangement in contemporary surviving parison with the tower walls, maximising chambers, which gives a clue as to the on the space available. Plain internal re- position of the partition (Figure 4). A divi- sponds around the apse suggest a plan with sion here would have given a room approx- aisles and an ambulatory, like St. John’s imately 6m by15m, with four windows in chapel at the White Tower.47 In addition, total. There was a vaulted mural chamber in there was a side chapel in a projecting S the adjoining NE turret42 but there were no turret. A gallery within the thickness of the garderobes attached to the outer chamber,43 remaining tower walls can also be safely which may be another pointer to its use as reconstructed, along with its turret accesses a formal audience chamber. in the NW and SW corners. The gallery The inner chamber thus recon- windows on the reconstructed plan pro- structed was smaller (approximately 6m by posed here are, however, speculative, 13m). Its fireplace was flanked by only one though based on parallels at Norwich and

186 The Castle Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration

Rochester. Similarly, the case for mezza- visaged as private spaces, probably bed- nine-level chambers will be argued here, chambers. On the S side, the space over the though it cannot be proven conclusively. antechamber to the Hall would be large enough to be divided into two. The proxim- Access to the second floor was ity of this space to the chapel might suggest possible by either of the stairs in the turrets use by the resident chaplain. Both Norwich on the W side. During Phase I, while work and its derivative at Castle Rising had a was suspended on the upper floors and the mezzanine- level chamber close to the cha- building was temporarily single storey, the pel, thought to accommodate the priest, and NW turret had been raised as far as first each also provided a room beside the cha- floor level, where it would have led only to pel that was probably associated with it, the roof.48 Despite this, the turret was perhaps a vestry.53 Two such chambers equipped with two garderobes,49 presum- could have been accommodated in the pro- ably in readiness to serve the projected hall. posed mezzanine-level space at Colchester. It may be that this corner of the building had advanced further than the rest, for evi- Drury attributed the Colchester dence of the temporary crenellation is gallery to the influence of the White Tower weakest in this corner, with no sign of it on which, at the time he was writing, was the S wall of the turret itself. When work thought to have an internal gallery over- resumed in Phase II the decision was made looking the upper section of the second to continue the stair upwards, an apparent floor.54 This is now known to have been an change to the old plan that involved the external feature, originally at the level of blocking of one of the garderobes.50 the gutters, which only opened on to the upper chambers after the roof was raised in The mural gallery at second floor c. 1490.55 The White Tower gallery led level has been taken as evidence that there only to the triforium of the chapel (see was another full storey,51 but this is not Figure 1 parapet level). The upper storey necessarily so. The donjon was already arrangement at Colchester, put on hold for massive in scale and the king’s donjon at a couple of decades, benefited from an Norwich was setting a trend for a single, improved design. This gallery was built as large storey above the basement with some a proper, internal feature and was also set at additional accommodation at mezzanine the floor level of a second upper storey. level. There, the principal floor was encir- Here it overlooked the extensive space of cled by a gallery that overlooked the public the dual-storeyed aisled hall as well as pro- rooms and also gave access to the mezza- viding convenient access to the aisles of the nine-level chambers. This formula may chapel rather than its triforium. Thus it well have been adopted at Colchester, could work on a ceremonial level while where the gallery certainly led to the chapel also giving practical access to mezzanine- and quite possibly to rooms over the ante- level chambers. Its inspiration is less likely chamber on the S side and the chambers to be found in the White Tower than in a range on the E side (Figure 5). The propor- more contemporary model. Only three tions of the hall are consistent with its hav- English great towers have true internal gal- ing risen through the full height of the top leries: Norwich, Hedingham and Rochester storey to the rafters, but the proportions of and the last two post-date Colchester. the waiting room and E side chambers would suggest that they were single storey Back in the 1070s the intention rooms with extra chambers above. The E may have been to attempt a near twin to the side mezzanine chambers could each have White Tower but this was never going to be been accessed independently, like the W easy and circumstances, very possibly ex- side chambers at Norwich;52 these are en- acerbated by difficulties encountered in

The Castle Studies Group Journal No 23: 2009-10 187 The internal arrangement of the donjon at Colchester in Essex: a reconsideration achieving a workable plan, prevented the 6 The symbolic connotation associated progress of the build. When it resumed it with the re-use of Roman building materials seems times were changing. Norwich pro- has been noted, for example, at Langeais vided the inspiration for a the revised de- (Indre-et-Loire) c.1000; Impey and Lorans, sign at Colchester: quite apart from the 1998, 25 and in the donjon at Mayenne architectural parallels, the dating and close (Mayenne) c. 900; Early 2002, 254. personal connections between Henry I and 7 Drury 1982, 390 Count Eudo, and that Colchester was re- turned to royal ownership in 1120 add 8 Ibid. 393. weight to this hypothesis. I would also 9 Ibid. 393 contend that Colchester in its turn influ- enced the design at Rochester. Similarities 10 Ibid. 306 between the Colchester hall arrangement 11 Morant, 1768, 7-8 and the layout of the ceremonial space on 12 RCHM Essex, vol. III, 1922, 50 the second floor at Rochester have already been alluded to and Rochester, too, was 13 Brown 1976, 66; Drury 1982, 393 and given a gallery that overlooked the princi- 400 pal floor. It may be that Norwich and 14 Drury 1982, 395 Colchester present us with a bridge be- tween the great Norman donjon prototypes, 15 HKW I, 31 and fig. 6; Brown 1976, 66; such as Rouen and Ivry that inspired the Mesqui 1992, 116, 126 White Tower, and the next generation of 16 For alternative interpretations of the high status donjons. development of the great tower at Norwich, Endnotes see Drury 2002 and Dixon & Marshall 2002. 1 HKW I, 31; Drury 1982, 399-400 17 Crummy 1981, 80 and fig. 67; Drury 1982, 395 2 Ia relates to Phase VIIA in Paul Drury’s extensive report on the building and related 18 For plans of Beaugency see Valery- excavations published in Archaeological Radot, 1930 and Mataouchek 2004. Journal in 1982: his Phase VIIA2 will be 19 Impey 2002, 193 called Phase Ib here and his Phase VIIB, Phase II. For detailed analysis of the dating 20 Chatelain 1973, 129 and Planche VII evidence see Drury 1982, 398-399. 21 Dixon and Marshall 2002, 236 3 For comparative plans of Colchester 22 M.R.Hull in Drury et al, 1982, 322 and the White Tower, see HKW I, fig. 6; Drury 1982, Figure 40. For comparative 23 Zarnecki quoted by Crummy, 1981, 80 plans of major rectangular donjons see 24 Drury 1982, 315 and Fig 2; 396 Mesqui1992, Figure 124. 25 Dixon & Marshall 2002, 237-238. 4 Hull 1982, 319; Drury 1982, 391 26 A stairway situated immediately adja- 5 Wheatley, 2004, 123-128. In relation to cent to an entrance and leading to a more Colchester in particular, see Wheatley important social space on an upper floor is 2004, 41. For a comparable interpretation a common feature of donjon design. More of re-used Roman materials in William’s usually, however, the formal entrance is at donjon at Chepstow, see Turner 2004, 254- first floor level and the adjacent stair is 257; Creighton 2002, 67; Wheatley 2004, positioned to allow unimpeded access for 128-129. favoured visitors to a superior set of rooms on the second floor. For an early example

188 The Castle Studies Group Journal No 23: 2009-10 The internal arrangement of the donjon at Colchester in Essex: a reconsideration and explanation of this point see Loches 46 Crummy 1981, 80, quoting H. Jenkins, (Indre-et-Loire), built c.1013-35; Marshall Colchester Castle: the templed citadel, 2002, 143-145. Later examples include 1869, 59-61. Crummy reported being able Richmond (Yorkshire), Hedingham (Es- to see evidence of the doorway behind a sex), Rochester (Kent), amongst many oth- museum case, but this does not preclude its ers. being a later alteration. 27 Hull, 1982, 322 47 RCHM Essex vol. iii, 1922, 54. 28 Ibid. 320-321 48 Hull was fairly confident that first floor level in the turret belonged to Phase 29 Crummy 1981, 80 There was certainly Ib. Hull, 1982, 321. a fireplace in this position by the late fif- teenth or early sixteenth century; RCHM 49 Ibid. Essex, iii, 1922, 54 50 Hull 1982, 321. The building survey 30 Marshall 2002i, 145-149; Goodall on the SW transept at Ely Cathedral, con- 2006. temporary with the work at Colchester, showed a marked tendency to build up the 31 Hull 321. One of these was subse- corners first at each level. Fearn et al 1995 quently blocked in association with carry- ing the staircase up to the second floor. 51 Drury 1982, 395. Drury notes that mural galleries are only found on the top 32 RCHM Essex, vol. iii, 1922, 51 floor, citing various examples, including 33 Hull 1982, 321-322 Rochester, with Hedingham as an excep- tion. In fact Rochester is the only exception. 34 Mesqui 1998, 95 52 Dixon & Marshall 2002, 237-238 35 Marshall (2006), unpublished PhD thesis, 295 53 Allen Brown, 1978, 54-55 36 Ibid, 93 54 Brown and Curnow 1984, 66 and Figure 3. 37 Drury 1982, Figure 38 55 Impey et al 2009. 38 Marshall 2002, 150 Bibliography 39 For appearance doorways and their possible Carolingian origins see Marshall Brown R.A. (1976), English , Lon- 2006. don. 40 A strikingly similar room arrange- Brown R. A. (1987), Castle Rising Castle, ment is found in the great tower at Nor- English Heritage, London. ham in its second phase (c.1160). Dixon Brown R. A. and Curnow P. E. (1984), The and Marshall, 1993, 419-423. , HMSO, London. 41 Early examples can be cited at May- Chatelain A. (1973), Donjons Romans des enne (Mayenne), Loches (Indre-et-Loire), Pays d'Ouest, Paris. Beaugency (Loiret), amongst others. Creighton O. H. (2002), Castles and Land- 42 RHCM Essex, vol. iii, 1922, 51 scapes, Equinox, London. 43 Hull 1982, 322 Crummy P. (1981), Aspects of Anglo-Sax- 44 RHCM Essex, vol. iii, 1922, 51 on and Norman Colchester, CBA, London. 45 Ibid., 51-54 Dixon P. W., and Marshall P. E., (1993), ‘The Great Tower in the 12th Century: the

The Castle Studies Group Journal No 23: 2009-10 189 The internal arrangement of the donjon at Colchester in Essex: a reconsideration case of Norham Castle’, in Arch. J. 150, raine, Societe Francaise d’Archeologie, 9- 410-432. 59. Dixon P. W. and Marshall P. E. (2002), Impey E. and Parnell G. (2000), The Tower Norwich Castle and its Analogues’ in The of London: the Official Illustrated History, Seigneurial Residence in Europe AD Merrell, London, in association with His- c.800-1600, ed. G. Meirion-Jones, E. Im- toric Royal Palaces. pey and M Jones, BAR International Series Impey et al (2009), The White Tower, Yale 1088, Oxford, 235-243. University Press. Drury P. J. (1982), ‘Aspects of the origins Marshall P. E. (2002) ‘The Ceremonial and development of Colchester Castle’ Function of the Donjon’ in Château Gail- Arch. Jour. 139, 302-419. lard 20, Caen, 141-151. Drury P. J., (2002), ‘Norwich Castle Marshall, P. E (2006) Magna Turris: A ’, in The Seigneurial Residence in study of the development, planning and use Europe AD c.800-1600, ed. G. Meirion- of social space in donjons of the eleventh Jones et al (above) and twelfth centuries located in the geo- Early R. (2002), ‘Le Château de graphical territories of the Norman and Mayenne: les témoins archéologiques de Angevin kings of , unpublished l’évolution d’un centre de pouvoir entre le PhD thesis, University of Nottingham, Xe et le XIIe siècle’, Château Gaillard, 20, 2006. Caen, 247-262. Mataouchek V. (2004), ‘Beaugency, la Fearn K., Marshall P. E. and Simpson W. Tour César’, in Lumières de l’an mil en G. (1995), The South-West Transept of Ely Orléanais, Brepols, Turnhout. Cathedral: Archaeological Interpretation Mesqui J. (1991-1993a), Chateaux et En- and Site Archive. Unpublished report for ceintes de la France Medieval, 1 and 2, the Dean and Chapter and English Heritage. Paris. Goodall J. A. A. (2006), ‘The great tower Morant, P. (1768), The History and Antiq- of ’ in Medieval Art, Ar- uities of the County of Essex, T. Osborne et chitecture and Archaeology at Rochester, al, London BAA Transactions 28, 265-299, Maney Publishing, Leeds. RHMC Royal Commission on Historic Monuments Essex, vol.3, 1922. HKW History of the King’s Works (Brown, Colvin and Taylor, 1963) Turner R. C. (2004), ‘The Great Tower, Chepstow Castle, Wales’ in Antiquaries’ Hull M.R. (1982) ‘Observations on the Journal 84, 223-318. Keep’ in Drury 1982, 317-323. Vallery-Radot J. (1930), 'Beaugency - le Impey E. (2002), The turris famosa at Ivry- donjon' in Congrés Archéologique de la-Bataille, Normandy in The Seigneurial France, 93, 341-351. Residence in Europe AD c.800-1600, ed. G. Meirion-Jones, E. Impey and M Jones, Wheatley A. (2004), The Idea of the Castle BAR International Series 1088, Oxford, in Medieval England, York Medieval 189-210. Press, York. Impey E. and Lorans E. (1998), ‘Le donjon de Langeais et son environnement’, Deux donjons construits autour l’an mil en Tou-

190 The Castle Studies Group Journal No 23: 2009-10