The pre-Conquest Cathedral at Canterbury

THE PRE-CONQUEST CATHEDRAL AT CANTERBURY By DAVID PARSONS, B.A. lliTRODUOTION THE only evidence whlch exists for the Anglo-Saxon cathedral at Canterbury is the well-known description by the precentor Eadmer.1 Eadmer's account, although it is among the most lucid and detailed of early medieval descriptions of buildings, must be treated with some caution; like the writers of similar accounts at this period, he was neither architect nor architectural historian, and the lucidity of his prose does not necessarily imply accuracy of observation. It should also be remembered that Eadmer was a boy at the cathedral school when Lanfranc undertook the rebuilding;2 it may be assumed that he wa.'3 quite young at the time of the destruction of the Saxon church in 1067. Childhood impression.a of important buildings are apt to be somewhat exaggerated and distorted, and Eadmer's impressions were not written down until much later, by which time he had become thoroughly familiar with Lanfranc's building. The details of his description, then, should not be regarded as sacrosanct and capable of only one interpretation, nor should his vaguer statements, such as the passing comparison with old St. Peter's, be insisted upon. The present attempt to reinterpret parts of Eadmer's account has produced hypothetical ground plans different from Willis's, which has been accepted in broad outline by most other commentators. This is not to suggest that Willis was 'wrong'; until full archreological evidence is available no reasonable interpretation can be rejected. The intention here is to show that there are lines of approach other than Willis's and that alternative interpretations of Eadmer's description are possible. lliTERPRETATION General All previous interpretations have relied heavily on the -plans of old St. Peter's and of the small basilica at Silchester, largely because Bede mentioned that St. Augustine re-used a Roman building as his cathedral church,s and because Eadmer referred to St. Peter's in his 1 Text and translation in R. Willis, Architectural History of Canterbury Cathedral (London, 1845), 9-12. i Willis, op. cu., 15 and 14 note d. 3 HJJJ I, 33. 175 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY description. It is debatable, however, how far one may press Eadmer's comparison. It is true that the works in which Eadmer described the pre-Conquest church at Canterbury were probably written after his two visits to Rome, and it, is clear that he had first.hand knowledge of St. Peter's: it was there that the Vatican Council of 1099 took place, a description of which appears in his Historia Novorum in Anglia. There is nothing in that description to suggest that Eadmer paid special attention to his architectural surroundings, and his later memory of them need not have been particularly accurate. At the same time, St. Peter's must have had a considerable impact on any visitor, and the extent of Eadmer's impressions should not be underestimated. He does, however, qualify his comparison with the phrase ex quadam parte. This may be taken generally to mean 'to some extent', indicating that the comparison was a relativc>ly superficial one, or more specifically to mean 'in some respects', limiting the comparison to certain features only. Since Eadmer refers again to St. Peter's when describing the ma.in altars of Christ Church as being raised above a confe ssio and approached by steps, it may be argued that these are the only respects in which Eadmer found the buildings similar. To as1mme a fully basilican building is not only to go beyond the text but to ignore the archreological evidence. It is clear that from time to time the Anglo-Saxon builders introduced some elements of plan and some architectural details from continental sources, but it i.s not yet possible to demonstrate that any overall building plan, such as the basilica, was ever com,istently adopted by them. In the seventh century, it is true, the builders themselves were imported and presumably brought with them ready-made plans. The Northumbrian churches of the type attributed to Benedict Biscop bear no resemblance to the Roman basilicas of the previous century, but may be compared with sm11,ll contemporary Merovingian buildingo, which normally consisted of a limited number of box-lilte cells. This is not surprising in view of Bede's explicit statement that the masons concerned were brought from Gaul.4 It is the Augustinian churches which may have been built by Italians, and here the basilica might have been introduced. What is known of Kentish churches of the seventh century, however, suggests that it was not.5 The sm11,ll basilican building excavated at SHchester was unkno"'ll ' Hiatoria Abbatum, 5, in C. Plummer (ed.), Ven. Bedae Opera Hiatorica (Oxford, 1896), 368. 6 E. C. Gilbert's attempt in Art Bulletin, xlvii (1965), 1-20, to define a so• called 'English basilica' in the eighth century is a piece of special plea.ding whfoh only serves to cloud the issue. Those of his examples for which there ia archreologica.l evidence are composite buildings essentially different from the true, formal basilica, cf. infra, 180. The evidence for a. miniature basilica. at Lydd (E. D. C. Jackson and E. G. M. Fletcher, 'Excavations at the Lydd basilica., 1966', J.B.A.A., Third Series, xxxi (1968), 19-26) is both enigma.tic and undatable. 176 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY to Willis, but subsequent commentators6 seized upon its discovery as providing justification for the western part of Willis's plan with its western apse. However, if the Silchester building was a Christian church, which is not proven, it cannot be regarded as typical of its period because so little is known about Christian buildings in lateRoman Britain. There is, then, no real evidence that the building adapted by St. August,ine in Canterbury was of the Silchester type. The present attempt to interpret parts of Ea.dmer's description is based on what is actually known about church building in Kent during the Anglo-Saxon period itself,7 and assume.s that the Augustinian building may have been of the standard seventh-century type, with a nave flanked by single porticus and separated by a triple arcade from a stilted apse. The ground plan of St. Pancras, Canterbury, is fairly representative of this type before the addition of further porticus; ultimately (but at a date not yet established) such portious were to surround completely the naves of Reculver and Deerhurst. Fig. la is a half-plan based on that of St. Pancras, showing the essential features of the type; the western porch, which is not relevant to the present discussion, and the porticus overlapping nave and chancel, for which the evidence is slight, are omitted. Figs. lb and 2 represent hypothetical developments of this basic plan, which in some respects fulfil Eadmer's description. The East End The crypt is the clue to the arrangements at the east end, as Willis saw.a First, Ea.dmer's words curvatura cryptae ipsius and his specific reference to Old St. Pater's allow one to infer an apsidal end. Second, the full description implies that the crypt did not extend west under the choir, but was entirely contained within the apse, as the plans projected by Scott, Brown and St. John Hope have shown.9 Figs. I b and 2 suggest that the two units, the sanctuary above its crypt and the choir, could both be contained within an exaggeratedly stilted apse of the St. Pancras type, '\\ith the chord of the apse defining the 6 Summarized by W. St. John Hope, 'The plan a.nd arrangement of the first cathedral church of Canterbury', P.S.A.L., Second Series, xxx (1917-18), 137-56. 7 The most recent survey is E. G. M. Fletcher, 'Early Kentish churches', Med. Arch., ix (1965), 16-31. s Op. oit., 25-6. o Cf. St. John Hope, op. cit., fig. 6, facing 162. For the purposes of this paper it is not necessary to discuss the crypt in detail; it may be asswned that Eadmer was referring to some form of annular, or ring-crypt. A brief discussion of the ring• crypt at Christ Church and thl') extant example at Brixworth is given by H. M. Taylor in the Programme of the Summer Meeting at Oanterbury (RAX, 1969), 19. Dr. Taylor discusses the annular crypt in more detail in 'Corridor crypts in En􀃩and and on the Continent,' N. Staffs. Journ. of Field Studies, ix (1969), forthcommg .. 177 a b Fig 1 X!-1 ::===.-:..-:..-{..􀀘􀀙 §􀀚=.-:..-:..-:..􀀛,...-􀀜 _-_ -_-_ -_ -_ ....; rY ,- ,, --􀀍􀀎􀀏--7-::..::..::..::..::..􀀐 y DP 1969 ... - CJ [ apse demolished r==-====CJ====[ a b Fig 2 Fig 3 (/) 0:: <( 1-­.-1 <( z <( 2'. + - ,, .. - C (:)) u .c -I:'-- - > 0:: <( 􀂏 0:: a.. 􀀈 .c Ill - :J I,,.. - 0 0:: z <( w 0:: <( 0:: 0 0 .-1 LL 0:: w a.. a.. :::, > 0:: <( Cl z 0 u w (/) - ('-• - C (:)) u .c - 0 .... I -0 E - (/) z 0 I-­ Cl Cl <( J <( u 1- w :r: l­ o a.. > :r: I􀀃□ · Fro. I. (a) Half-plan of St. Pancras, Canterbury, after Hope and Clapham. (b) Half-plan based on St. Pancras with hyPothetical .Additions fulfilling Parts of Eadmer's Description. F10. 2. Alternative Interpretation: (a) at ground level; (b) at first-floor leveL Flo. 3. St. Pancras-type Nave as the Basis of a Westwork in a muoh extended Church. THE PRE-CONQUEST CATHEDRAL AT CANTERBURY line of the sanctuary steps and the western extent of the crypt.lo This is very similar to the arrangement at Brixworth, where the main body of the church was clearly divided into threellllits: nave, presbytery and sanctuary;n the presbytery was divided from the nave by a triple arcade of the Kentish type, so that the original form of Brixworth was not unlike what is here proposed for Christ Church. The main points of difference are that in the St. Pancras type of church used for the Christ Church reconstruction the apse and choir are of equal width and slightly narrower than the nave, while at Brixworth the nave and presbytery are of equal width but with a markedly narrower apse. Now the form of that apse is clearly not original, and it is interesting to speculate whether the outer wall of the present external ring-crypt may conceivably perpetuate the line of an earlier apse. This apse, together with the presbytery, would have been merely an enlarged version of a stilted apse such as is shown in Figs. I and 2. If the ringcrypt had existed at the same time, it would have been an internal feature and comparable with previous commentators' plans of the crypt at Christ Church . .Assuming that the choir was housed in the rectangular space between the sanctuary steps and the triple arcade, how can this be said to fulfil the next part of Eadmer's description 1 This concerns the choir itself: incle ad oocidentem clwrus psallentium in aitlam eoclesiae porrigebatur, decenti fabrica a Jrequentia turbae seclu-sus.12 The interpretation breaks down unless the next stage is also assumed, namely the addition of flanking aisles or porticus. Fig. 2, which may be used in toto to gain an impression of the overall reconstruction plan, shows· these enclosing the east end as far as the chord of the apse.13 Once they had been built (as they clearly had by 1067) the body of the church (aula) . could be thought of as comprising everything except the apse, that is the whole building from the west wall of the nave (X-X) to the line of the east walls of the aisles and the sanctuary steps (Y-Y). Thus the choir as defined could be described as penetrating into the aula. The east wall of the nave with its t"iple arcade, together with t,he north and south walls of the choir space, would have served much the same purpose as the closure screen of the sixth-century Roman basilica (e.g. St. Sabina, St. Clemente) and would have stood in a similar 10 Willis, op. cit., 11, refers in translation to 'the single passage (of entrance)' of the crypt; the Latin text does not specify that via una is an entrance passage, so figs. lb and 2a. show the crypt entrances flanking the sanctuary steps, which ia in keeping with the assumption of a ring-crypt (see note 9). 11 The most recent plan is in H. M. and Joan Taylor, Anglc-Saa:on Architecture (Cambridge, 1965), 110. 11 Willis, op. cit., 10, translated 11. 18 Slight indications of north-south walls adjoining the apse at its chord were found a.t St. Pancras; see plan, Fletcher, op. cit., 21. 179 THE PRE-CONQUEST CATHEDRAL AT CATERBURY spatial relationship to the building as a whole. This wall complex could be regarded as the 'appropriate structure' dividing the choir from the thronging congregation. An objection to this interpretation may be found in the plan of St. Pancras itself (Fig. la), where the minor north and south archos of the triple arcade are blocked by secondary walls. Such blockings would effectively separate choir from nave, and the barrier which they would form does not allow the all-embracing interpretation of aula suggested above. Since at this point only the lower parts of the walling of St. Pancras survive, however, there is no proof that the secondary walls completely blocked the arches; they may equally have been lcw partitions forming a primitive choir-screen. In this case, nave and choir would have been fully intervisible and the choir could reasonably be described as forming part of the aul,a,. The A isles and Towers Eadmer next mentions that the church had aisles. Previous interpretations have taken this reference literally and proposed a fully basilican plan, taking into account neither the rarity of Anglo-Saxon churches with aisles in the strict sense, nor the possible ambiguity in medieval Latin of the word ala. William of Malmesbury, writing not long after Eadmer, refers to al,a,s vel porticus when describing the preConquest church · at Glastonbury. The published ground plan of Gla.stonburyl4 reveals a building unlike a properly aisled basilica, and the flanking chambers are usually regarded as porticus. William, howe�er, was not prepared to make the precise distinction of t;he modern architectural historian, and offered alternative descriptions. The likelihood is that at the time a church could easily become quasibasilican by the addition of :flanking structures (either continuous aisles or inter-connecting porticus) and the piercing of the solid nave walls,15 and that William regarded al,a,s and porticus as synonymous. It is possible, then, that Eadmer used alas similarly of non-basilican accretions where today porticus would be specified. In any case, there is no reason to assume that the church described by Eadmer wa.s of one build or ever conceived and planned as a, whole, although he naturally described it as an entity. There is no good reason, then, to assume either a full basilican building or even continuous aisles, de.spite the word al,a,s in the original. Figs. 1 b and 2 show interconnecting porticus of different periods. The building sequence proposed is based on that of St. Pancras, where the primary features are the nave and the stilted u H. M. and Joan Taylor, op. cit., 253. 16 There are, of course, innumerable examples of this procedure in the post• Oonquest period. 180 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY apse. The porticus midway along the nave are secondary, but by analogy with the west porch (not shown on the plans here) may have been added during the building of the primary structure.16 These primary and secondary features form the basis for the reconst,ructions and are shown in half-plan in Fig. la. To this basic plan are added the hypothetical 'aisles' in the form of separate but interconnecting porticus, which are tertiary in the overall sequence. If it is thought desirable to follow St. John Hope's interpretation of the east end,17 the easternmost porticus may be extended to the north and south, respectively, so as to produce a pseudo-transept as shown in Fig. 2. St. John Hope's argument for a transept, however, rests partly on an invalid comparison of the Anglo-Saxon cathech·al with Lanfranc's church; this will be discussed briefly in another context. Additions of this sort to a previously aisleless structure would fit quite well with the building activities of Archbishop Odo (942-58} mentioned by Eadmer in his Vita Odonis,18 Odo's rebuilding is supposed to have taken three years and to have included the raising of the walls. Willis10 suggests that the heightening of the walls may indicate the addition of a clerestory, but does not draw the further conclusion that this must have been the consequence of adding aisles or porticus, which are likely to have covered the original nave windows. The addition of the aisles (which in any case must have occurred after the early seventh century) is therefore tentatively attributed to the midtenth century, but since the interpretation is pure conjecture, firm dates should not be insisted upon. The towers may be a.ssumed to belong to the same period of construction. St. John Hope quotes Scott's interpretation of the towers as representing the eastern limit of the Silchester-type western structure.20 But as Clapham points out21 towers would be quite out of place in either a late-Roman or an Augustinian context; his suggestion that they were an addition above pre-existing porticus has the weight of archreological evidence behind it.22 Olapham's interpretation is adopted here, and the towers are assumed to ha.ve been erected over the two secondary porticus at the same time as the aisles were added. Of the aisles and towers Eadmer writes: Dein sub medio kngitudinis 10 H. M. e.nd Joan Te.ylor, op. cit., 148; Flet-Oher, op. cit., 22, makes the interesting suggestion the.t the portious me.y ha.ve been Augustinie.n additions to a. pre-Saxon building. 17 Op. cit., 155 and fig. 5, facing 152. 18 Text and translation in Willia, op. cit., 3-4. 19 Op. cit., 30. 20 Op. cit., 140, 142. 21 A. W. Cle.pham, English Romanesque Architecture before the Oonquest {Oxford, 1930), 86, 11 Western porches were frequently raised to form towers. in the late-Saxon period, e.g. :Bardsey, Corbridge, Monkwea.rmouth. 181 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY aulae ipsius duae turres erant, prominentes ultra ecclesiae alas.12 The position of the towers sub m,e,dio longitudinis q,ulae has been variously interpreted as 'beyond the middle' and 'about the middle' of the nave. The controversy over this detail of translation IB singularly arid; suffice it to say that in the reconstruction plans proposed in Figs. lb and 2 the towers are placed half-way along the nave, but, in relation to the east end, beyond the middle of the aula as defined above. Slightly more important is the translation of prominentes ultra . . alas. Willis took this in a strictly lateral sense and drew his plan with flanking towers standing illogically outside the 􀃾ain structure. St. John Hope pointed out, however, that ultra could mean 'above'23 and incorporated the towers into the aisles. This idea of vertical penetration is well suited to the present interpretation. The towers, raised above the earlier porticus, would be seen in elevation to 'rise up through' the tertiary porticus system which abuts them. Should this interpretation not be acceptable, Fig. 2 has been drawn with the aisles narrower than the porticus-towers, so that the towers extend beyond them laterally as well as vertically. The final detail in the interpretation of the south tower concerns the altar in medw sui. St.John Hope agrees with Scott in objecting to Willis's apse to the south tower, saying that the altar should be placed against the middle of the east wall.24 This is a reasonable view, but just the same Willis's instinct was sound. An altar against the east wall would interfere with the use of the suthdure, which must have been inserted into the south porticus when the aisle was added. Fig. lb resolves this difficulty by assuming that the east wall of the porticus was pierced by an arch to throw it open to the adjoining part of the new south aisle complex. The altar could have been housed beneath this arch, perhaps with a screen behind it. The West End The most unsatisfactory feature of previous interpretations is the western apse. Willis suggested it as the neatest way of fulfilling Eadmer's description, and later commentators were confirmed in this view by the spurious analogy of Silohester. There are no other examples in AngloSaxon Engla.nd, apart possibly from Abingdon for which there is only literary evidence, although there are examples on the continent (e.g. Trier, Paderborn, Cologne). It is not easy to justify the idea of a western apse, particularly since it is not absolutely required by the text. Eadmer makes two major points about the Lady Chapel housed at the west end of the church. First, the altar stood at the east end of the ss Op. cit., 153. " Op. cit., 142. 182 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY chapel and the celebrant faced the people who were standing below. Second, the chapel could only be approached by steps: structura ... talis erat, non nisi per gradi(,8 cuivis patebat access11,8, .All that may be safely assumed is that the chapel was built at a higher level than the nave. It may have taken the form of a western gallery, for which there is structural evidence in the .Anglo-Saxon period,25 in which case the steps by which it was approached might well not have been in the nave, but in one of the flanking chambers as indicated in Fig. 2b. Nothing in the text demands a flight of steps like those leading to the sanctuary at the east end. The arrangement suggested in Fig. 2 is a timber gallery supported on a north-south wall inserted across the nave. Fig. 2a shows pillar bases at ground level to indicate a triple arcade which would have allowed access under the gallery along the axis of the nave. If this interpretation is admissible, the gallery must have been a substantial structure in order to support the pontifical chair, which Eadmer describes as being a ponderous affair of stone and mortar. He does say, though, that the chair was contiguous with the main wall of the church, and it may have been actually cemented to the wall, which ,vould have provided additfonal support; indeed, it could have, been a permanent feature built into a wall recess such as the gable-headed niche in the tower at Barnack. Undoubtedly, the best support would have been provided by a gallery carried on a masonry vault, but neither the text nor the archreological evidence gives the slightest excuse for assuming this. Fig. 3 represents a very tentative extension of this interpretation of the west end. Here most of the nave and the secondary porticus of the St. Pancras-type church are shown incorporated in true AngloSaxon fashion into a much larger structuL'e, of which they form the entire west end. The necessary staircases are added as turrets in the re-entrant angles between the nave and porticus. The hypothetical plan of Fig. 3 is very similar to that of the west end of St. Pantaleon, Cologne,26 which is a westwork in the Carolingian tradition. There is as yet no archreological evidence for the construction of westworks in England, and this interpretation may be as objectionable as Willis's apse; however, other descriptions of late-Saxon churches suggest that westworks may have been introduced into this country. It may also be objected that Fig. 3 is too ingenious and represents a building more sophibtioated than most pre-Conquest churches. Canterbury O!l,thedi:al, however, is likely above all others to have been developed ihto an elaborate. structure. 2& E.g. at Deerhurst, Tredington, etc. 28 Of. L. Grodecki, L'Architectu1·e ottonicnnc (Paris, 1958), 51, fig, 16. 183 THE PRE-CONQUEST CATHEDRAL AT CANTERBURY The Scale of the Church The interpretation represented by Fig. 3 raises the whole question of the scale of the late-Saxon cathedral. Although the type of church suggested in Fig. 3 may be thought too grandiose for the pre-Conquest period it must be borne in mind that Canterbury Cathedra.l was the Primate's church and therefore of some importance; further, it was the seat of St. Dunstan during the tenth-century reform period, and is likely to have been the subject of radical rebuilding and extemiion in the same way as the Old Minster at Winchester, the other major centre of monastic reform. The Old Minster wltich emerged from the building operations of the tenth century was clearly a structure of some size and complexity27 and may provide a better parallel with Christ Church than does the smaller type of Anglo-Saxon cathedral such as North Elmham. For this reason it may be thought that the churches proposed in Figs. lb and 2, based as they are on St. Pancras (overall length without west porch: c. 7 5 ft.), are too small in scale. At the same time, the larger Anglo-Saxon churches of the tenth and eleventh centuries, such as Winchester and St. Augustine's, Canterbury, seem to have grown from much smaller beginnings. The scale of St. Augustine's was achieved by throwing together SS. Peter and Paul and St. Mary's by the interpolation of Wulfric's octagon. Neither of the original churches was of any great size. There is no reason to suppose that the Augustinian cathedral at Canterbury was any larger, particularly since the original intention was to move the metropolitan see to London. To this extent the interpretations in Figs. I b and 2 are acceptable so long as it is realized that they do not necessarily represent the whole church complex in the eleventh century. There is no direct information about the absolute size of the eleventhcentury cathedral. St. John Hope argued28 that it was much the same size as La.nfranc's church, because it had been set out in relation to the Anglo-Saxon conventual buildings still remaining. But Eadmer expressly states20 that La.nfranc's buildings were larger than their predecessors, and that he went to the trouble not only of demolishing the preConquest remains but of digging out their foundations. This he would hardly have done had he intended to rebuild along the same lines. The most that can be said, then, is that the pre-Conquest cathedral was probably smaller than Lanfranc's church, but by how much is not known. 27 Textual and archreologfoal evidence for this may be found in the following: R. N. Quirk, 'Winchester cathedral-in the tenth century', Arch. Journ., cxiv (1957), 26-68; M. Biddle and R. N. Quirk, 'Excavations near Winchester Cathedral, 1961', ibid., cxix (1962), 150-94; and M. Biddle's subsequent excavation reports annually in Antiq. J ourn. (1963 ff.). :s Op. cit., 154-5. 29 Text and translation in Willis, op. cit., 13-14. 184