The First Norman Cathedral at Canterbury

THE FIRST NORMAN CATHEDRAL AT CANTERBURY By EDWARD GILBERT THE first Norman cathedral at Canterbury was begun in A.D. 1070 by WilHam's new archbishop, Lanfranc. Lanfranc came from Pavia originaUy, via Bee in Normandy, and St. Etienne, Caen. St. Etienne was William's own foundation, and Lanfranc was Abbot there when appointed to Canterbury. Lanfranc's new cathedral was the first of the Anglo-Norman cathedrals, and it laid down, to some extent, the pattern which these would take. It does not foUow, however, that aU its features were necessarfly derived from Normandy. The superior culture and prestige of the Engfish state at the time would make this thesis a dangerous one. Comparatively Httle attention has been paid to Lanfranc's cathedral. There has been no fuUy detailed consideration of it since WiUis wrote in 1845.1 His work was so exceUent that any analysis must use it extensively; and much of the ensuing study is an attempt to see where WiUis' conclusions need modification in the Hght of a further 125 years. The textual references to the history of the cathedral were carefuUy coUected by WiUis, and the historical facts given here, unless otherwise stated, rest on his authority. Such modifications of the picture as are here made arise mostly from a fresh study of the fabric, and partly represent a coUation of known facts with contemporary architectural history. It seems clear, amongst other things, that WiUis did not make the tedious climb into the north triforium of the nave, thereby missing some important evidence, while it has to be remembered that the architectural history of the West was even more obscure in his day than it is in ours. Other writers, like Gilbert Scott2 and St. John Hope,3 have not succeeded in invafidating WiUis' main conclusion, and they too seem to have found the cfimb into that triforium too much for them. Gilbert Scott's plan of the cathedral is formalized, and no substitute for Willis' thoughtful plan reproduced here (Fig. 1). Although much of Lanfranc's cathedral has been entirely denatured, a good deal of the fabric remains (Fig. 2), and coUating this with the careful description of the church by the monk Gervase, it is possible not only to reconstruct its form in some detaU, but to recover also 1 R. Willis, The Architectural History of Canterbury CatJiedral, London, 1846. * G. Gilbert Scott, jun., Essay on the History of Church Architecture, London, 1881. a W. St. John Hope and J. Wiokham Legg, Inventories of Christ Church, Canterbury and London, 1902. 23 EDWARD GILBERT i ,*J ftRlffcfc-W ATI n n TOWCKM trail CtefSffft. I > H i n ft H 3, fjm • K« 'K» p* NAV6. • • M t'tfP Itt !(«;{., fcmfcasstfswi w »(« waisctc «*<«#•>«*& Fia. 1. Willis' Plan of the Cathedral in 1174. 24 PLATE I mis Xationa! Building Record The Angle of Lanfranc's Transept and the Nave. face page 24 PLATE II The Canterbury Weal of'c. 1130. Photo: Cathedral Library PLATES III AND IV National Buildings Record The Podium under the Crossing from South. • "f > Xational Buildings Reeorit The West Wall of Ernulf's Crypt. PLATE V Photo: E. Gilbert The Stair Turret at Milborne Port, Som. THE FIRST NORMAN" CATHEDRAL AT CANTERBURY much of its spirit.4 The evidence suggests that it expressed a rehgious outlook, and architectural fashions more appropriate to the earHer Middle Ages, from about A.D. 450-1100, than to the later period to which most of our Romanesque cathedrals belong. The decisive change in spirit is difficult to analyse. One aspect of the earHer rehgious outlook was the greater emphasis on the promise of the Second Coming. This rose to a last climax as the year A.D. 1000 approached, and the |c.U71-78 jc.1100-15 J c.m'i-s.t .ii&"r.v*."r.v.9r.\'.r."^.\".,s".r.^.v.i"+r.v.rs\v."[&" lo )$ -io metre* Fio. 2. Plan of Canterbury Cathedral showing Work of 1170, according to A. W. Clapham. failure ofthis great hope involved profound psychological reorientation, including a greater stress on the hope of heaven. It is perhaps symbohc of this hope that Gothic cathedrals soar skyward, while the great Romanesque achievements seem to cling closer to the earth. The earHer outlook reflects itself also in a loose architectural unity to which the generic name 'Carolingian' is often given. Lanfranc's church is to some degree Carolingian in this sense, and nowhere in England can one come so close to the spirit and form of that architecture as at Canterbury. The other great interest of Lanfranc's cathedral is its relation to its immediate predecessor on the site. We are iU-informed about the late Anglo-Saxon cathedrals and any evidence, however Httle, is valuable. The question is how much Lanfranc's cathedral reflects its predecessor. The Canterbury historian Eadmer asserts that Lanfranc * Gervase's descriptions of the Norman work at Canterbury cathedral ooour in more than one of his works. The references are collected and translated by Willis, op. cit., in n. 1, oh. 3. I have used this version as being easily the most convenient. Gervase wrote in the late twelfth century. 25 EDWARD GILBERT rebuflt a fundamento.5 Such claims are general medieval practice and rarely mean exactly what they say. The matter is further discussed below (pp. 47-8). In any case, Eadmer's statement does not dispose of the possibihty that Lanfranc's cathedral was on the site of the Saxon cathedral, that it reproduced some of its details, or even that it re-used some of the foundations. The medieval buUder was inclined to do this last unless there was good reason for not doing so. It was cheaper, to begin with. The usual reason for not doing so was the desire to enlarge. But the Saxon cathedral at Canterbury, one may safely say, cannot have been much smaUer than Lanfranc's. The enlargement of it referred to by Eadmer, was probably mostly at the east end.6 Attitudes are changing today about the influence of the Saxons and their churches on the Anglo-Norman churches of England. Such influence has always been admitted abstractly, as imphed in the phrase 'Saxo-Norman overlap'. This influence has not been much admitted in respect to the Anglo-Norman cathedrals, but it is now becoming clear that the late Saxon cathedrals affected their Norman successors more than we have hitherto realized. An increasing number of major Norman churches can be shown to have been influenced by their immediate predecessors. Examples occur at Sherborne Abbey,7 Wimborne Minster,8 Rochester Cathedral,9 Bath Abbey10 and, as we shaU see, Lanfranc's own cathedral at Canterbury. Almost certainly the same could be shown of other early Norman cathedrals and minsters. Canterbury cathedral was originaUy founded as such by St. Augustine about A.D. 600. According to Bede,11 Augustine re-used a Roman church, by which he may have meant a later Romano-British church, as is suggested by the fact that the axis of the cathedral is far out of 6 Eadmer, the precentor, wrote his memories of the Anglo-Saxon cathedral, with some glances at the Norman work of Lanfranc, about A.D. 1100, also in several different works, which are likewise summarized and translated by Willis, op. cit., 13-19. Eadmer's comments were reprinted and retranslated by Dr. H. Taylor, who gave also the sources of each part in 'The Anglo-Saxon Cathedral Church at Canterbury', Arch. Journ., cxxvi (1969), 125-9. Por a translation of Eadmer's Historia Novorum, see G. Bosanquet, London, 1964. 8 Eadmer actually says that Lanfranc rebuilt the cathedral 'augustiorem'. This is usually taken to imply enlargement, but does not necessarily have to do so. Cf. Taylor, op. cit., 127. ' For Sherborne, see Royal Commission on Historical Monuments (West Dorset), R.C.H.M., London, 1952. 8 For Wimborne, ibid., East Dorset, ii, pt. I, xliii-xliv. 0 The most relevant facts come from an unpublished study by Dr. 0. A. Raleigh Radford. 10 There is unpublished evidence about Bath Abbey in the City Library where Irvine's notes on his excavations are kept. I owe my knowledge of this to Mrs. Helen Panter, of Norton St. Philip. 11 Bede, Historia Ecclesiae, i, 8. 26 THE FIRST NORMAN CATHEDRAL AT CANTERBURY aHgnment with the Roman street system.12 Baldwin Brown13 and Clapham both beheved that Augustine's church must have been largely rebuflt before A.D. 1066. In accordance with the only text,14 then known, claiming any sort of the buUding work at the cathedral subsequent to Augustine; they considered Archbishop Odo to have been the rebuUder in about A.D. 950. There is no text suggesting that Augustine himseh0 added to his 'Roman' church. It is possible, nevertheless, that the church on which Odo worked was not the original church, but a rebuUd by Archbishop Wulfred of A.D. 813.15 This would make the late Saxon cathedral, as Clapham believed it to be, essentiaUy a Carolingian church, somewhat modified by Odo. Such a church may have existed before Lanfranc's church, and it may reasonably be suspected that, where Lanfranc's church departs notoriously from current Norman practice, a possible cause may be the influence of the lost Saxon cathedral. In A.D. 1067 the Saxon cathedral was burnt, and aUegedly aUowed to he derehct for three years by the Saxon bishop Stigand.16 Lanfranc, arriving in A.D. 1070, is said by the Canterbury historian Eadmer to have buflt an entirely new cathedral in about seven years. From about A.D. 1096-1130 a great new east end, stiU largely extant, was buUt by the Priors Ernulf and Conrad. After the murder of Becket in A.D. 1170, the choir was again rebuflt and extended, the work being planned by WilHam of Sens, and finished by an Enghsh mason, also caUed WiUiam. No record of the rebuUding of Lanfranc's nave at this time exists. Just before the Peasants' RebeUion of A.D. 1381, Archbishop Simon of Sudbury prepared to rebuild the nave, and possibly the transept, of Lanfranc's church. He is alleged to have puUed down the nave preparatory to rebuilding, but examination of the fabric shows that this claim is false. Simon himself died in the revolt, and the renewal of the nave was actuaUy carried out by Prior Chillenden, who succeeded him, but who left at least the two flanking towers at the west end untouched. THE WEST END OF THE NAVE This part of the cathedral was not described either by Eadmer, who wrote about A.D. 1100, nor by Gervase who wrote an account of the 11 See Sonia Chadwick Hawkes, 'Early Anglo-Saxon Kent', Arch. Journ., cxxvi (1969), 190, fig. 2. 13 G. Baldwin Brown, The Arts in Early England, II, Edinburgh, 1925, 76 ff. 14 A. W. Clapham, English Romanesque Architecture, Oxford, 1930, I, fig. 16. 16 E. C. Gilbert, 'The Date of the Late Saxon Cathedral at Canterbury', Arch. Journ., cxxvii (1970), 202. 18 This and the following historical facts are all given by Willis, op. cit., in n. 1. He deals with events to 1130 in Ch. 1, from 1130 to 1180 in Ch. 3 and for subsequent years, he discusses the choir in Ch. 6 and the nave in Ch. 7. 27 EDWARD GILBERT cathedral after the reconstruction of the late twelfth century. Hence, we are more than usuaUy dependent on architectural evidence for our knowledge of Lanfranc's work. The usual early Norman fabric was of smaU rectangular stonework facing a rubble core. This can be seen, for mstance, at the relevant parts of Winchester, Hereford and Chester cathedrals. At Canterbury cathedral the earhest stonework stiU extant is a rubble waU with a facing of smaU squarish stones. These stones are not actuaUy square in face, but measure about 8 in. X 6 in. I shaU refer to this as 'smaU square stonework', though at times it verges on rubble. It reflects Roman practice, especiaUy in France, as for example in the town waUs at Le Mans, but also in England in the London waU, and at York. A simUar technique was used in the Roman revival of the seventh to eighth centuries, as at Jarrow in England and Beauvais in mm?***. « fMMm .: m -'! X*. \! \ • ! i I * Si =. ; • M W W W W I ! .! ; ! « \ I i ! I :. I < ? m 1 * am m l I ii m • > sw» i i m « »; •: it I. • • ' W i t f ! I! I ! i» : tefe* > ^i u FIG. 3. Ernulf's Fabrio and Arcading (Exterior of Choir). 28 THE FIRST NORMAN CATHEDRAL AT CANTERBURY France.17 At Canterbury this fabric has always been attributed to Lanfranc, but it must be emphasized that Ernulf also used it, though his stones are often shghtly larger, about 9 in. X 7 in. (Fig. 3). Lanfranc's work cannot be certainly distinguished from Ernulf's merely by fabric—a vital point sometimes neglected. There was, after all, only some thirty years between the two works. The two WiUiams preferred a larger ashlar often about 12 in. X 9 in., but they both reused old stones where convenient and avaUable. So also did Prior ChiUenden, whose own stonework is best seen in the cloisters, where it is seen to be a megahthic ashlar. The smaU square ashlar is never at any place regularly and consistently used. Many irregular stones intrude, some of which must be repair work, others possibly re-used Saxon; but some of them may weU be original. The distribution of this early stonework makes a re-use of its material in the cathedral certain. The two west towers flanking the nave were certainly part of Lanfranc's scheme, as Gervase teUs us.18 The north-west tower survived tiU 1834 (Fig. 4). This drawing shows it as having five storeys, of which the first four appear Norman and the top one transitional. Such drawings can, however, be unrehable in regard to detaU, especially of the window arch-heads. InternaUy, in both towers, the Norman fabric stiU exists on the inner east waUs, but only up to the top of the third stage, after which a new fabric supervenes. It is possible therefore that both the top two stages were added in the transitional-Norman period, and presumably by the WiUiams, in spite of the drawing. THE NAVE The visible fabric has many smaU, squarish stones in the nave, especially in the clerestory, suggesting that this is the original fabric. The aisle waUs are largely rebuflt in Chfllenden's large ashlar, especiaUy round the aisle windows, while the quatrefofl triforium windows are also, externaUy in the larger stonework, presumably Chfllenden's. It foUows that either Simon of Sudbury did not take down aU the Norman walls, as aUeged, or that ChiUenden re-used a great many stones coming from Lanfranc's fabric. Either conclusion is possible, but both WiUis and P. M. Johnston felt certain that part of the existing nave wafls is Lanfranc's work.19 WUHs described this part as the internal plinths of Lanfranc's side waUs. By the word 'plinth', he 17 The 'basse ceuvre' is given as either of the eighth or tenth centuries by K. J. Conant, Carolingian and Romanesque ArcMtecture, Harmondsworth, 1969. The west wall, however, has cordons different to, and unlined with, those of the nave. It is clearly an addition, so that the relevant fabric is that of the eighth century or earlier. " Willis, op. cit., in n. 1, ch. 3, 47. 18 P. M. Johnston, 'An Architectural Itinerary of Canterbury Cathedral', in Kent, 6th Edition, London, 1934. 29 0. EDWARD GILBERT t*.k i. •: 4SK MSI » 3& «. Al- *» SSS I ¥• I :>^ai I fc<;i sii3 ?, ;* I I $w* 4* £i M, V...-.7 < • »» t ^ ii %.?**i.p. #** ** ••Ji A* Fio. 4. The North-west Tower before 1834. evidently meant the internal walling under the aisle windows, for what we caU the plinths ofthis walling are not specificaUy Norman work. Clapham shows Norman work only at the core of the existing nave waUs. There seems nevertheless to be confirmation of the view of Willis and Johnston in the existing triforia. Both of them have embedded in their west walls what are remains of early Norman arches about 8 ft. 6 in. wide. The outer orders of the arches show with voussoirs of the smaU square type, and traces of a nook-roll. The impost of the arch in the north triforium (Fig. 5) is twofold—a late Saxon habit, as at Hadstock, Essex, in about A.D. 1000 (Fig. 6).20 The jambs of the 20 For Hadstook, G. Baldwin Brown gives A.D. 1000-1040, Taylor A.D. 960- 1066.1 am informed by Mr. S. E, Rigold that later examples of the double impost exist in Germany, which was, however, notably inolined to arohaism. 30 THE FIRST NORMAN CATHEDRAL AT CANTERBURY v \ \ r—' -"-i FIG. 5. Imposts of West Arch, North Triforium. (Note the double character.) same arch are also ordered, and show traces of a nook-shaft, of which the capital appears to be a hoUow-moulded frustrum of a cone, and hence unlike Lanfranc's typical capitals which are simple cushioncapitals. Except for this capital and the double impost, the arch is very Hke those on the bhnd arcade of the west waU of the ancient dorter. Some of this waU survives and is always attributed to Lanfranc. The corresponding arch in the south triforium lacks the double impost, but is otherwise simUar. The height of the arches is uncertain, as their bases are below the floor level. The doorways covered by these arches appear to have communicated between the west towers and the triforia. On the tower side the arches are not visible, being covered by a skin of walling. In the north triforium the archway is not centrally placed. It is on the extreme outside of the triforium, and is stiU north of central to the north-west tower, which projects beyond the triforium. If there was no Norman triforium at this level, as is usuaUy assumed, the archways would cover external doors to the west towers. From the height above groimd, the non-central position, the lack of any apparent purpose for such doors and their lack of agreement with general Norman practice, such doors are improbable. I t is more Hkely 31 EDWARD GILBERT C- "—J I Fio. 6. Hadstock. Impost of North Door, c. 1020. that the Norman triforia were at the same level as the existing ones, and that these doors formed entrances to them from the towers. The position of that to the north triforium is pecuhar, suggesting that it opened to a chapel with an altar, as happened with Saxon porticus. The triforia are apparently referred to in the texts by the term 'porticus'.21 Hardly less interesting than the west waU of the north triforium, is its south waU. There is no triforium arcade. Instead, it is divided into bays, corresponding to those of the nave below, by perfectly plain square-cut buttresses 2 ft. 6 in. wide and of 1 ft. 6 in. projection. They look Hke external buttresses. In each bay is a large opening of about 8 ft. x 3 ft. with a flat head. It is buUt of the early, squarish stonework, here so finely jointed that the joints are only visible with difficulty. The openings are not dressed at the angles, and have mono- Hthic Hntels and sUls. They are very shghtly splayed, the narrow end towards the nave. One thing is certain about these strange openings: 21 Willis, op. cit., in n. 1, 128. Prior Goldston built the south oampanile 'ab altidudine porticus ecclesiae'. He in fact rebuilt from above the triforium level, so that 'porticus' seems to refer to the triforium. 32 THE FIRST NORMAN CATHEDRAL AT CANTERBURY they are not the same date as the exterior quatrefoil windows of the triforia (Plate I). These have the typical ChiUenden megahthic ashlar in the splayed jambs, and are doubtless his work. At the south-east angle and partly at the south-west angle of the north triforium are a few inches of thicker walling on each face which has apparently been cut away on the east, south and west waUs. In the south-east angle, the adjacent thicker waUing contains the spring of a perfectly plain arch-head, which Henderson,22 foUowing Willis, thought must be the remains of a clerestory arch, but looks now more like the remains of a triforium arch. A simUar fragment of archspringing exists externaUy at the north end of the west waU of the transept. This is the one that WilHs saw and supposed to be a clerestory arch. It would seem natural to suppose that the thicker triforium waUing was Lanfranc's and that it was cut away to make ChUlenden's walling, especiaUy on the south waU of the north triforium. A difficulty in this view is that the clearly early Norman arch in the west wall of the triforium seems to be embedded in the thinner waUing. Moreover, the flat-headed triforium openings do not look like ChUlenden's work. It is perhaps best to regard their date as uncertain. In the thinneddown east waU of the triforium there is no blocked arch to the transept. The triforium seems always to have been entered from a waU-passage, as today, and to have been a kind of secret room, not having an entry visible from below. Some Ottoman triforia are arranged like this, as for example at St. Michael's, Hildesheim, and no doubt at later churches also. It is stiU difficult to say exactly how much of Lanfranc's nave and west end today remain, but somewhat more than WilHs thought. There are, moreover, various technical detaUs which seem to reflect pre-Norman practice, as though older traditions and habits had not yet died out in Lanfranc's day. Concerning the main body of the nave of Lanfranc's day we have Httle information. Gervase tells us it had eight 'pfllars', which may, of course, mean piers. This, no doubt, includes the piers of the west towers, and gives the same number of bays as today. WiUis beheved that ChiUenden entirely replaced, rather than encased, the Norman piers, but there does not reaUy seem any means today of being certain on this point. GUbert Scott, and foUowing him Woodruff and Danks, placed the Norman choir in the nave entirely,23 on the grounds that there would be no room for it in the Norman sanctuary, nor in the transept, because of the organ in the south transept. The same conclusion is suggested by the fact that the great candelabra was in the 22 A. E. Henderson, Canterbury Cathedral: Then and Now, London, 1956. 23 C. E. Woodruff and William Danks, Memorials of Canterbury Cathedral, London, 1912, 20, and plan facing. 33 EDWARD GILBERT nave. Moreover, the choir could hardly have occupied any part of the crossing with its flights of steps. According to Gervase, the north nave-aisle contained a chapel to St. Mary, which WiUis and others place at the angle between nave and transept. In this area a skeleton was found, believed to be that of Archbishop Theobald (A.D. 1139-1161). Here again we find the survival of ancient customs, for such a chapel or porticus existed in Saxon churches in the same position, for example at North Elmham,24 while the archbishops in early Saxon days were buried in a porticus in a simUar position at St. Austin's, Canterbury. SIGILLEM ECCLIE. XKI CATTTUABIE. PEIME SEDES BRITANNIAE The Canterbury Cathedral Seal of about A.D. 1130 gives us considerable information about the form of Lanfranc's cathedral (Plate II). An interesting fact is that it had a west porch, of two storeys projecting between the flanking west towers. The rest of the representation is so accurate that the reahty of this feature need not be doubted. I t is included in my reconstructed plan. The flanking towers of the nave facade are both there, and are of four storeys, thus confirming the suggestion above that the fourth stage of 1834 was Lanfranc's, despite the drawing figured. The towers finished with cockerel weather-vanes on conical roofs. The nave itself has the usual round-headed clerestory windows below which is apparently an enriched band of waUing. There are no visible triforium windows externaUy, which would explain why the flat-headed triforium windows are splayed so as to Hght the triforium internaUy. The range of windows below the clerestory windows, and rendered much Hke them, is apparently intended to be the aisle windows. It is always a question of how much rehance can be put on old representations of architectural features. No simple answer can be given. SmaU detaUs can never be regarded as certain; nevertheless, this seal seems a particularly good example, and carefuUy done. LANERANO'S TRANSEPT Continuing the examination of the Canterbury seal, we find a transept of which the south facade has one of those fuU-height arched recesses, as at the west front at Tewkesbury. The gable has three plain Romanesque windows, the centre being the largest. The central tower has two stages above the crossing, either very heavfly fenestrated or treated with Carolingian pierced arcades. I t finishes with another conical roof and, as a finial, an enormous angel, which gave the 21 S. E. Rigold, 'The Anglian Cathedral of North Elmham, Norfolk', Med. Arch., vi-vii (1962-3), 68. 34 THE FIRST NORMAN CATHEDRAL AT CANTERBURY tower the name 'the angel steeple'. I t looks like some outsize butterfly. That the structure is archaic by Anglo-Norman standards is selfevident. Even more remarkable is the fact that the base of the tower, as it rises above the crossing, is much above the waU-plate of the transept fa9ade, so that the transept either sloped down from the central tower, or dropped in a broken line. The former is Hke St. Riquier (A.D. 796), and the latter like many of the archaic Auvergne churches. So archaic is this feature that, if we had not Eadmer's words to the contrary, we should suspect the transept of being Wulfred's of A.D. 813. The sanctuary shown on the seal is Ernulf's and does not concern us, except that it is tolerably weU represented. Lanfranc's church is also represented in some drawings. One of these was printed by WiUis, and dated to A.D. 1160. Here the west porch was gone, and external triforium windows appeared. It is never very safe to rely for detaU on drawings, but assuming this to be accurate, it foUows that work was done on Lanfranc's cathedral between A.D. 1130 and 1160. The 'angel steeple' is weU represented in this drawing, and here are given three stages above the crossing rather than two. And it seems to be round. The transept shown on the seal is partly identical with the present west transept. This is externaUy almost entirely of the smaU square stonework, though much modified internaUy and in parts externaUy. The present central tower, except for its top stage, is probably Lanfranc's 'angel steeple'. None of his fabric shows, it is true, but in 1895 the Norman shafts and their cubic capitals were found under the piers of the crossing.25 Here then, at least, the thinner waUs were Norman, the thicker being of about A.D. 1400. The present top stage is the work of Prior Goldston. The tower piers are curiously placed. In WilHs' plan (Fig. 1) the western Norman piers, whose exact position is obscured by the fourteenth- century casing, are shown in Hne with the transept west waU, but the eastern piers are appreciably west of the Hne of the transept east waU. This would make the tower narrower than the transept, and that is how it appears today, though exact measurements are impossible. The transept itself measures about 124x34 ft., i.e. the same width as the central aUey of the nave, a common Romanesque arrangement. I t has, at its angles, clasping buttresses about 4 ft. wide, with the peculiarity of triple square-cut recessing at their corners. Ernulf used the same detaU in his equivalent buttresses, a fact which warns of the danger of assuming that technical identity necessarUy means identity of date. At the north-west and south-west angles of the transept are internal newel staircases. That to the north-west is contained behind 25 Woodruff and Danks, op. cit., in n. 23, 29. 35 EDWARD GILBERT the buttress, in waUing always accepted as being Lanfranc's. The staircase windows are flat-beaded sHts. That at the south-west angle is simUar, but the external walling has been renewed extensively. InternaUy, in both newels, the walling is renewed for some reason. An interesting fact about the transept externaUy is that its west waU, at the north end, including that part covering the newel, was in a technique which cannot at present be paraUeled anywhere else in the cathedral. Here, the smaU cubic stones, much weathered, have occasional Roman tiles set verticaUy and at random in thick mortar jointing. This technique is a Carolingian one, and was used in the Loire vaUey, as at the church of St. Solenne, Blois, probably of the tenth century.26 InternaUy, the great curiosity of the transept is the arrangement of the levels. The wings of the transept are at the nave level, whfle under the crossing is the great podium rising 14 steps from the nave to the sanctuary (Plate III). This is rendered only formaUy in WiUis' plan (Fig. 1). In reahty, it rises by two steps to a platform 3 ft. wide between the west tower piers. Then come three more steps to a platform of passage 6 ft. wide running between the wings of the transept. Then comes, at present, a flight of 9 steps to another platform 6 ft. wide which runs without a break into the choir. Since, as we shaU see, Ernulf raised the crypt by about 2 ft. 6 in., he presumably added four or five steps to the top flight of steps, and in Lanfranc's day the top platform must have projected some 9-10 ft. into the crossing. The width of the podium was in Lanfranc's day the same as that of the nave, but it was later, and apparently by Ernulf, extended northwards, but not southwards. The containing-waUs on both sides are predominantly of the smaU square stonework. The wings of the transept are reached by descents of stone steps from the central passageway of the podium. This complex of levels is virtuaUy unique in Norman or Anglo- Norman cathedrals, only that at Rochester even approximating to it, and this fact alone would strongly suggest that the podium was conditioned by its Saxon predecessor. When, in addition, we know from Eadmer's description that the Saxon cathedral had such a podium, the fact becomes too remarkable to be merely coincidence. It foUows that here, at least, the Norman cathedral is on the Saxon foundations, and possibly re-used some part of the Saxon cathedral. There is other evidence to support this view. The lowest part of the east waU of the existing tower appears to be a Saxon survival (see p. 47), and the relative shortness of the transept is some indication that the wings too may be on Saxon foundations. That the podium must represent the Saxon podium reclothed is 26 For St. Solenne, Blois, see Dr. Lesueur's article, Bulletin Monumental, 1930, 435. 36 THE FIRST NORMAN CATHEDRAL AT CANTERBURY suggested by a crucial passage in Eadmer's description.27 He says that the Saxon altars were reached by a few steps from the choir, i.e. the nave. The word he uses is 'altaria', which could mean the sanctuary, the area where the altars were kept. In that case, it would include the transept, and Eadmer's remark would mean that the transept was reached by a few steps up. He then goes on to say that because the vaults of the crypt were so high, the parts above them, i.e. the apse, could only be reached by 'phires gradus'. In classical Latin 'plures' means 'many' in this context. In SUver Age Latin, and presumably a fortiori in Medieval Latin, 'plures' can also mean 'more'. It makes perfectly good sense to say that the transept was reached by a few steps, and apse by 'many' or by 'more'. It is not good sense to say almost in the same breath that the altars were reached by a few steps and also by 'many' or 'more' steps, unless they were on different levels. The altars concerned would be those of St. Wilfrid, against the east waU of the apse, and that of St. Saviour, to the west of St. Wilfrid's. What seems unsatisfactory, as criticism, is simply to disregard the apparent contradiction between the statement that 'few' steps were needed to reach the altars, and 'many' to reach the apse where at least one of the altars is known to have been. Whether one takes 'altaria' to mean the sanctuary as a whole, or 'the altars', Eadmer's words seem to imply that the Saxon podium had more than one flight of steps, and, therefore, a platform in the middle, presumably for the altar of St. Saviour. Since this same platform clearly existed in Lanfranc's church, and stUl exists today, there seems here a strong reason for supposing the Norman podium to have reproduced the Saxon one, as indeed does the existing podium, with certain modifications. Another curious fact about the existing podium is that there is beneath it towards the east end a passageway, with a barrel-vault, connecting the transept wings. In its present form, its entries are of fourteenth century, and it was no doubt used for Hturgical requirements at that period. As shown in 1970, however, it or something like it would be required for the form of Saxon crypt envisaged by WUlis,28 a form supported by some evidence from this Norman church (see p. 47). This passageway seems to back on to the waUing there discussed, and rated as probably Saxon. These facts are quite compatible with the podium having been a reclothing of the Saxon original. 27 'ad haeo altaria nonnullis gradibus ascendebatur a choro cantorum, quoniam cripta, quam confessionem Romani vocant, subtus erat, ad instar confessionis sancti Petri fabricata, suius fornix eo in altum tendebatur, ut superiora ejus non nisi per plures gradus possent adiri'. Cf. Willis, op. cit., in n. 1, 10. 28 Willis himself did not realize this. He made the entrance to the 'via una' from the west. But this is contrary to what Eadmer says, since he says a thick wall separated it from Dunstan's tomb. Since Dunstan's tomb is on the west, the passage must here have turned north and south. 37 EDWARD GILBERT The transept wings of Lanfranc's church had east chapels, placed at their outer extremities to leave room for the flights of steps leading up to the choir-aisle and down to the crypt. These chapels were twostoreyed, as is the south chapel today. The lower chapel was dedicated to St. Michael on the south and St. Benedict on the north, whUe the upper chapels were dedicated to AU Saints on the south and St. Blaise on the north. WilHs represents these chapels as being apsed. The Latin word used in describing them by Gervase is 'porticus'. These early porticus, certainly in Saxon days, were more commonly rectangular, and since the existmg chapels on their site are not apsed, there seems to be Httle reason for supposing that Lanfranc's chapels were apsed. More hkely they were flat-ended and, in any case, did not project so deeply as the present chapels. There is reason to beheve that the upper chapels were of special importance and sanctity in Lanfranc's day, and even that they were organized Hke Httle churches, with nave and sanctuary, for the transepts had upper gaUeries, presumably at the level of the upper chapels, and it seems that the area of these, behind the chapels, may have been incorporated with the chapels. Thus, on the south, there is still a large blocked arch of early Norman type, perfectly plain, without imposts or orders, embedded in the waUing at the entry to the upper chapel. This has aU the appearance of a sanctuary-arch, and indeed the entrance to the upper chapel was by a staircase, which stfll survives, from the choir-aisle waU. On the north the transept gaUery carried aU the tombs of the Saxon archbishops from Cuthbert onwards, except for Jaenbert: it must therefore have been a mausoleum, and St. Blaise's chapel to the east of it seems to have acted as a chapel for this mausoleum. According to WilHs, the opening from the upper gaUery to St. Blaise's chapel was only through a door in a waU or screen, an arrangement quite compatible with the gaUery and chapel being together regarded as a mausoleum. The vaulting under the upper gaUeries of the transept wings was supported, according to Gervase, on the side waUs and on a single pUlar, in the centre of the 'cross', which was his word for the transept wings. Existing reconstructions confine the vault and gaUery to the parts of the transept outside the nave aisles, but Gervase's 'cross' seems to mean the whole of the transept outside the crossing. In that case, the gaUeries reached the crossing, an arrangement which is Saxon in origin, as at Deerhurst, Glos., where there are superimposed arches to the crossing, and which lasted weU into the eleventh century, occurring, according to Clapham, at Jumi&ges and Bayeux in Normandy.29 Gervase also teUs us that the crossing in Lanfranc's church was cut off from the nave by a screen which presumably ran from pier to the 29 A. W. Clapham, op. cit., in n. 14, I I , fig. 2, 13. 38 THE FIRST NORMAN CATHEDRAL AT CANTERBURY pier of the central tower, and had the altar of St. Cross to its west, though Gilbert Scott places the latter at the west end of the Norman choir, far down the nave. In any case, the screen carried a rood, with figures of Mary and John, presumably in the lower compartments, and cherubim, presumably sitting on the crossbar, as in the Saxon rood at Romsey, Hants. One would expect also screens separating the transept wings from the nave aisles. Gervase does not mention these, but as late as the eighteenth century iron railings cut off the transept from the nave aisles,30 and the way from one transept wing to the other was through the passage under the podium. RELIGIOUS MEANINGS OE LANFRANC'S CHXTROH The arrangements of Lanfranc's transept have considerable reUgious interest. It is plain that the transept was regarded as an integral part of the sanctuary. This is shown not only by the existence of the rood at its west entrance, which alone would not be decisive, but by the unparaUeled coUection of sacred tombs in the north transept. The idea could, as we have seen, have derived from the Saxon cathedral, but it goes back at least to Merovingian times, for example, as in St. Martin's at Tours, of which the total east end, probably apse and transept, is described by Gregory of Tours about A.D. 600 as the 'altaria' in opposition to the nave which he caUs the 'capsus', 'ox waggon'.31 'Altaria' is the same word as is used by Eadmer with possibly the same reference. Lanfranc's altars were mainly in the transept whereas almost aU Ernulf's altars and most sacred tombs were concentrated in the new choir, while the transepts, even if still east of the rood screen, were nevertheless deprived of honour. Indeed, it seems likely that the purpose of Ernulf's vast new choir was to contain the main coUection of tombs and altars. I t would, therefore, seem logical to move the rood to the entry of this choir, when it was buUt. No text pinpoints this move, but it was apparently in existence by A.D. 1305, when texts place the pulpitum at the entrance to the choir.32 Presumably, Ernulf moved it there when resiting the altars. The transept would then begin to lose sanctity and tend to be seen as the transition from the world, i.e. the lay church in the nave, to the City of God, i.e. the monks' church embodied in Ernulf's glorious choir. If it is true, as Woodruff and Scott beheved, that Lanfranc's choir was in the nave, then the clergy were also upgraded by Ernulf, and it would even seem as if the original sanctuary of Lanfranc was stiU regarded as a kind of Holy of Holies, not to be entered by human creatures except in pursuance of some special function. 30 WiUis, op. cit., in n. 1, 112. 31 Gregory of Tours, Historia Francorum, ed. Ruinart, Paris, 1609, II, 14. 32 Willis, op. cit. in n. 1, 97. 39 EDWARD GILBERT LANERANO'S EAST END This is not described either by Gervase or by Eadmer and has, therefore, always been, and stiU is, a mystery. WiUis considered it to be a three-aisled structure, the central aUey on arcades and aU three finishing in apses. He assumed that the central aUey would be ahgned with the tower piers. This is not the present arrangement. Today, the central aUey is wider than the tower, extraordinary in a Norman church. WiUis therefore argued that Ernulf must have widened Lanfranc's central aUey, as he certainly widened the sanctuary as a whole.33 Such a proceeding would seriously threaten the stability of the central tower, however, and it does not explain why the choir begins with sohd waUing before breaking into arcades. Willis regarded the sohd waUing as the work of Ernulf, and as being designed to buttress the now unstayed tower. Such a method of buttressing would be highly irregular in Norman practice, and the gain in width to the choir, about 4 ft. on each side, hardly seems worth the enormous expense of pulling down Lanfranc's waUs and buUding anew, or the risk to the central tower. The alternate thesis is that Lanfranc buUt the wide choir waUing; but this also would be most abnormal for Norman work, or indeed for late-Saxon work, and the influence behind it, if Saxon, would have to be earher Saxon. GUbert Scott, foUowed by Woodruff and Danks,84 conjectured that Lanfranc's choir would have had sohd waUs, and this seems reasonable from the comparison with the sister church at Rochester,36 and from the common occurrence of this feature in late-Saxon and early-Norman times, as for instance at St. Albans, and Old Sarum.36 Clearly, moreover, there must be a possibihty that the length of sohd waUing at the west end of the present choir is a fragment of Lanfranc's choir. The fabric would certainly be satisfactory for such a thesis. WiUis supposed an apsed finish to the choir and its side aisles (see Fig. 1); and the excavators of 1895 supposed they had proved this, the aUeged line of the north apse being inserted in the crypt floor. Canon Livett, however, in 1889, had already laid bare the east finish of the associated crypt,37 and was doubtful if it was apsed, tantamount to saying that an apse could not be assumed. Clapham considered the 1895 excavation unsatisfactory,38 and we need to bear in mind that the sister church at Rochester, traditionaUy built under the influence of Lanfranc, had a sohd east waU with no apses. The difficulty in accepting a similar finish for Canterbury is that this form of east end was abnormal 33 Ibid., 66. 84 Woodruff and Danks, loc. cit. in n. 23. 86 W. St. John Hope, Arch. Cant., xxiii (1898), 194. 80 For Old Sarum see Clapham, op. cit. in n. 14, 23. For St. Albans, ibid., 24. 87 Canon G. M. Livett, Arch. Oant., xviii (1889), 253. 88 Clapham, op. cit. in n. 14, 21. 40 THE FIRST NORMAN CATHEDRAL AT CANTERBURY for the early Norman period. This is true, but it is not unusual in late- Saxon work, as at St. Mary's, Dover,38 and it would not be unusual if we had here another Saxonism in Lanfranc's church. Certainly, the Saxon cathedral had an apsed east end, so that if Lanfranc's cathedral was flat-ended, the influence was not from the preceding church. But it could have come from general Saxon practice at the time. On the whole, we must conclude with Clapham that the form of Lanfranc's east end has never been satisfactorfly determined. THE CHOIR AISLES If there is uncertainty over the position of the waUs of Lanfranc's choir, there is not so much over those of his choir aisles. These were V M / /* xy 7 o 1f . / 7 V X// / // / / / \ / / / / t FIG. 7. North Wall of South Entrance to Crypt. Part of Network Diaper. on the same foundations as those of the present choir aisle waUs at their narrower western end. Lanfranc's choir aisles were reached from the transept by a flight of steps which presumably had his choir wall as their inner waU, and outside them, as today, were the steps down to his crypt. On the south, such a flight of steps to the choir aisles stiU exists, with the present choir waU as their inner waU, their outer wall bounding the present steps and passage to the crypt.40 This outer waU has on it enrichment consisting of a network ornament in stones about 1\ in. square set lozenge-wise (Fig. 7). This looks like Lanfranc's 80 For St. Mary's Dover, see H. M. and Joan Taylor, Anglo-Saxon Architecture, Cambridge, 1965,214. 40 See Willis' plan, Fig. 1. 41 EDWARD GILBERT work. There is a similar ornament set in larger stonework at the east end of the reredorter of Westminster Abbey which appears to be of the Confessor's date.41 The top of the network at Canterbury slopes upward at the same angle as the existing choir aisle steps, but finishes about a foot below them (Fig. 8). Therefore structuraUy it looks Hke Lanfranc's waU, heightened by Ernulf when he raised the level of the choir aisle, to aUow for his heightened crypt. If this waU is Lanfranc's, Fio. 8. North Wall of South Entrance to Crypt. The Steps up to the South Choir Aisle. we have certain evidence that the existing choir waUs are buUt on Lanfranc's foundations. The dating of the network pattern is aUimportant (Fig. 9). It was on the supposed date of this that Willis assumed that Ernulf rebuUt the choir aisle steps in toto. On the north, the choir aisle steps no longer exist today owing to the extension northwards of the podium. On its north waU, facing the crypt passage, is also a diaper, this time of a grUle pattern, much more sophisticated than the network pattern. It, too, does not slope like the network diaper, but reflects the flat top of the extended podium. StructuraUy and sculpturaUy, the diapers have the appearance of different dates, in which case the grUle pattern would be Ernulf's and 11 The dating of this particular piece of walling is much disputed, most authorities seeing it as 'Norman'. Baldwin Brown does so, but it must be remembered that for him the Confessor's abbey was a 'Norman' building! (op. cit. in n. 13, 244). In fact, such decoration is very uncommon in the external walls of Anglo-Norman buildings. 42 THE FIRST NORMAN CATHEDRAL AT CANTERBURY \ Fio. 9. Detail of Pattern of Network Diaper. the network pattern Lanfranc's. WUhs considered them both Ernulf's because both occur in Ernulf's later work at Rochester. It does not foUow, however, that at Rochester Ernulf copied only his own previous work, even supposing the network ornament at Rochester reaUy is Ernulf's. He could have copied earlier work either from Canterbury or even from Westminster. Clapham teUs us that the Anglo-Normans used no enrichment before about A.D. 1090 on their architecture.42 He excepts the network, but not the grUle pattern. In any case, the Saxons used enrichment on their architecture during the eleventh century, as at, for example, Stow and Barholme in Lincolnshire. Even the network pattern exists on an external stair turret at Mflborne Port, Somerset,43 attached to a tower generaUy accepted as late-Saxon (Plate V). The network pattern is also common in Gaul in the late eleventh century, as at Selommes near Blois,44 or the facade of the cathedral at Le Mans.45 Such a feature could easUy indicate Saxon influence at Canterbury. 42 Clapham, op. cit. in n. 14, 125, claims that the 'plain diaper' is in existence in England by at least A.D. 1066. I take it this must include the plain network pattern here disoussed. 43 Taylor, op. cit. in n. 6, thinks the tower Saxon, but the stair turret added by the Normans. But the external stair turret is, as he admits, a Saxon rather than a Norman feature. 44 For Selommes, near Blois, see Gabriel Plat, L'Art de b&tir, Paris, 1939. 45 For Le Mans, see Francois Salet, Congris Archeologique, 1961, 18. 43 EDWARD GILBERT The outer waUs of the narrower western part of the existing choir aisles are admitted by WiUis to have been on the Hne of Lanfranc's choir-aisle waUs, but he supposed the actual waUs to have been rebuilt by Ernulf.46 This ignores the sudden change in the width of the aisles Had Ernulf buUt the whole of the choir-aisle waUs, he would surely have bunt them in a continuous line. The natural interpretation of the present waUs is that Ernulf left Lanfranc's narrower choir aisles, so far as they stood, and widened them when he came to buUd anew. That there is no good reason for setting aside the natural interpretation is suggested by other evidence. In each of the choir-aisle waUs, at the western end, there is contained a narrow staircase leading, on the south, to the upper chapel of the south transept wing, which replaced the Norman chapel to AU Saints. Such staircases certainly existed in Lanfranc's church, since Eadmer refers to them, describing them as cochleae, newel staircases. TechnicaUy, the present waU-staircase is not a cochlea: it has sharp turns at the bottom (right-handed) and near the top (left-handed). I t is doubtful, however, whether Eadmer would have had avaUable any better word than cochlea to describe it. The existing staircase is probably that described by Eadmer, and if so, the choir-aisle waUs at this point are Lanfranc's waUs, and not Ernulf's. This is the more hkely, in that the entrance-door to the present upper chapel of AU Saints is a few steps below the present floor-level of the chapel and appears to have been made for a lower floor. The outer waU of the staircase is not in the smaU square ashlar of early days, but the inner waU is; and what is more there is in the inner waU what appears to be a double-splayed porthole window, a typical Saxon feature. It is now blind but apparently opens over the vault of St. Michael's chapel, presumably originaUy Hghting it. This window confirms that this waU must be Lanfranc's or earher, and, in addition, it explains why WUhs thought that the staircase had originally no outer retaining waU. I t foUows also that the existing east chapel of the south wing of the transept retains its early Norman waUs both to the west and north, ChiUenden rebuUding only the east and south waUs. THE CRYPT The crypt of Canterbury cathedral is one of the finest in existence. The eastern part of the crypt is not Romanesque and does not concern us. The western part was built by Ernulf, but he clearly re-used the spring of an earHer and narrower crypt at the west end. The wider part begins just where the choir aisle above it widens. The central aUey, 39 ft. wide, exactly underhes his own choir above, while the crypt aisles underlie his own choir aisles. The central aUey rests on 40 Willis, op. cit. in n. 1, fig. 1. 44 THE FIRST NORMAN CATHEDRAL AT CANTERBURY carved columns in double file. At the west end, however, for the first bays, the central aUey is only 25 ft. wide, and there are no side aisles, instead of which there are great blocks of masonry (Fig. 10) supposed to constitute underpinning for the central tower. This narrower aUey is itself divided into three aUeys by a double line of carved columns PASSAGE TO N. TRANSEPT TOWER FAST WALL N 0 o o 0 o o o o PASSAGE TO S. TRANSEPT "W 0 o o 0 o Fio. 10. Plan of the West End of Ernulf's Crypt. just like the central aUey of Ernulf's crypt further east; and, moreover, these columns are in file with Ernulf's columns. Each of the three aUeys so formed has its own plain quadripartite vaulting. The marks on the west waU of the crypt (Plate IV) show that the narrower western crypt had originaUy lower vaulting than that existing today, and this was heightened to the level of the vaulting of Ernulf's crypt further east, without raising the spring of the vault. Obviously, Ernulf did this. Above the line of the old vaulting, as it shows on the west wall, the fabric is the smaU squarish ashlar, here clearly Ernulf's, whfle below is a fabric 45 EDWARD GILBERT which is neither Ernulf's nor Lanfranc's, consisting of very rough rubble in Kentish rag and an irregular use of Roman tUe, the whole now whitewashed and covered at its centre by a large monument. In this narrower crypt the bays were only 6 ft. from east to west, and only one such bay remains, whereas in Ernulf's work further east the bays are 12 ft. from east to west. In the narrow crypt, against the west waU, are whole columns; and, presumably, the side responds were treated similarly, though none remain. Ernulf's waU-responds are engaged half-columns in the normal Norman manner. One result of this is that, whereas the crypt of Ernulf uses 24 whole columns in aU, including the re-used part of the older crypt at the west end, the older crypt would have had 24 whole columns in only five of its six-foot bays, a space about equal to that needed by the two bays and apse postulated by WiUis for Lanfranc's choir. This is the ground for WUhs' idea that all the whole columns in Ernulf's crypt may be Lanfranc's columns re-used, but carved in situ. Further support for this idea is given by the fact that the western columns, against the west waU, are placed to support not Ernulf's vaulting, but that of the previous crypt. That this previous crypt was Lanfranc's cannot seriously be doubted.47 The narrower crypt at the west end opened lateraUy to blind chambers inside the masonry blocks already described. That on the south stUl survives, and presumably that on the north also. Such chambers could be, technicaUy, porticus, and therefore Saxonisms. It is true that simUar chambers exist in the crypt at Gloucester Cathedral, but there is some doubt whether they represent the design of Serlo (A.D. 1088) or Aldred (A.D. 1058).48 It is not impossible that these chambers were the basements of towers at the angle between choir and transept. The masonry blocks containing them would more easUy support such towers than buttress the central tower, as they are aUeged to do. Any doubt that the narrower crypt was Lanfranc's should yield to a study of the surviving column bases (Fig. 11). They are either denatured, not giving any signs of a large hoUow mould, or else of the double-roU type, very close to those of the sister-crypt at Rochester. Ernulf's bases are built round a large hoUow mould. However, the west waU of this earHer crypt seems to be earHer still. As Canon Livett argued in 1889,49 it is in neither Lanfranc's Caen stone, nor in his technique. Lanfranc moreover would not of his own accord have used detached columns as responds, a praotice whioh is typioal of Merovingian days. He would be expected to use attached half- 47 Clapham, op. cit. in n. 14, 21. 48 Ibid., fig. 22. Clapham regards the ohambers as Norman. 49 For a comparison of Willis' and Brown's plans for the Saxon orypt at Canterbury cathedral see Gilbert, op. cit. in n. 16, especially figs. 1 and 2. 46 THE FIRST NORMAN CATHEDRAL AT CANTERBURY FIG. 11. Bases. (Left) Lanfranc. Re-used in Ernulf's Crypt, c. 1070. (Middle) Ernulf. Exterior Wall-arcade to Choir, c. 1100. (Right) Gundulf. Roohester Cathedral Crypt, c. 1080. columns, like his friend Gundulph at Rochester. The inference is that the waU was already there when Lanfranc buUt the crypt. This waU is covered with a hard white plaster which Livett, a good judge of things Saxon, thought was one of the numerous hard white Saxon plasters. This plaster runs behind the columns, further suggesting that the waU was built first and the columns placed in position later. The gap between the two is very sHght, and some of the plans are wrong in this respect. The weight of argument seems here to be on Livett's side, and probably we should assume that the waU concerned is a rehc of the Saxon crypt. The effects of doing so are far-reaching. They justify WilHs' theory of the form of the Saxon crypt, as against Baldwin Brown's. Further it goes far to suggest, as already discussed (p. 37) that the podium in the crossing is on the site of the Saxon one. If it could be determined how far this waU extended, we might have direct evidence for a Saxon transept. ORIGINS OE THE PLAN WiUis caUed attention to remarkable similarities between Lanfrano's cathedral and the church of St. Etienne, Caen, of which Lanfranc had been Abbot. He mentions the width of the naves, which he gives as 47 EDWARD GILBERT 73 ft. at St. Etienne and 72 ft. at Canterbury; the similar western towers, aisle-less transept, gaUeried and with chapels to the east; the length of the naves, which he makes 187 ft. at both churches, and of the transept, which he gives as 127 ft. in each case. Some of these features were fairly general at the time, but it seems reasonable to suppose either that St. Etienne heavfly influenced the cathedral, or that St. Etienne was itself strongly influenced by the preceding Saxon cathedral of Canterbury, or both were influenced by the same model. The relation between Norman and Enghsh architecture in the eleventh century has hitherto proceeded entirely on the basis of assumptions reflecting ideas as to the supposedly superior culture of the Normans— assumptions which do not, at first sight, seem very obvious. That Lanfranc's cathedral was on the site of the Saxon cathedral is absolutely certain. If not, it would not have been necessary to move the tombs of the saints at the east end of the Saxon cathedral whUe Lanfranc buflt his new east end, nor to move them again from the west end of the Saxon cathedral when Lanfranc came to bufld his own west end.50 The relation of the two podia further suggests that the Saxon nave, and the central tower and transept (if there were such), were the same width as those of Lanfranc. It is noteworthy that WiUis does not include relative measurements on the width of tower and transept in his comparison of the two churches, whfle he notes that the central aUeys of the nave were of a different width. Probably, the most likely influence from St. Etienne would have been the twin flanking towers of the nave fa§ade. CONCLUSION SO far as can be judged, none of the waUing of Lanfranc's cathedral above ground is re-used Saxon work. Even the fabric is apparently not re-used Saxon work, and it seems that the Saxon cathedral must have been of rubble with some Roman tUe. The numerous archaisms of Lanfranc's cathedral, most of which have been noted seriatim, do suggest Saxon influence in the cathedral. Not the least extraordinary are the flat-headed triforium windows. Something similar can be seen externaUy in the drawing found by Clapham and Peers in the tenthcentury manuscript in the Hbrary of St. Austins, Canterbury.51 This should represent a Carolingian church, not improbably the cathedral 60 These facts are given by Eadmer, cf. Taylor, op. cit. in n. 5, 128. The saints were taken from the 'orientali parte' of the Saxon, presumably St. John's Churoh, by now simply a mausoleum, and put temporarily in the western part of the Saxon churoh where the oratory of St. Mary was, and, presumably, in the poroh at ground floor level. Later they had to be moved from here when Lanfrano came to his own west end, and were temporarily put in the refectory. 61 A. W. Clapham and C. R. Peers, Archmologia, Ixxvii (1927), 201. 48 THE FIRST NORMAN CATHEDRAL AT CANTERBURY itself. On the other hand, flat-headed triforium openings occur in Romanesque bufldings, as for example, at Brioude and Chamaheres-sur- Loire. The archaic features are espeeiaUy interesting in that, whfle some may reflect the influence of the preceding church, others may represent only general late-Saxon practice. They are clues to the obscure subject of Enghsh second Romanesque architecture. Without more knowledge of this subject, comparisons of Enghsh and Norman architecture are almost futile. There is a possible explanation of some at least of the archaic features at Canterbury in the fact that whfle the Saxon cathedral was f-+* *4^+4 # • * # • m 4 « * * # # # 1 * • -o- 10 o 10 20 so 40 eo -At,it. Fio. 12. Reconstruction of Lanfranc's Churoh, modified from Willis. (East Wall based on Rochester Cathedral.) burnt in A.D. 1067, Lanfranc did not appear tiU A.D. 1070. In the meanwhile, the Saxon archbishop was fighting for his ecclesiastical rights against king and pope combined. An able and astute man, he was fuUy capable of reahzing that neglect of his cathedral would not help his cause. It is difficult to behove he did not at least appoint an architect and begin plans for rebuilding. Lanfranc may have taken over both architect and plans, modifying them to please himseh*. Hence could arise the curious mixture of Norman plan and Saxon detail apparent here. With regard to the plan (Fig. 12), I have tentatively preferred the flat east end, flat-ended transeptal chapels, tower piers not ahgned with the transept, and transept galleries right up to the crossing. All these decisions are debatable, and could eventuaUy go the other way, but they seem to me to have a shght preference. The form of the nave piers is, of course, not known. Lanfranc might have used columns; there are 49 EDWARD GILBERT some mighty enough for the purpose re-used by Ernulf in his choir. The plan aUows for simple two-ordered arches, but Gundulf's original work at Rochester was equaUy simple.62 62 My thanks are very muoh due both to Dr. 0. A. R. Radford and to Mr. S. E. Rigold for reading my text and making many valuable suggestions which will have helped to minimize errors in what has not proved a very easy piece of work. I also express my gratitude to the Leverhulme Trust for assistance with the expense of the illustrations. 50