The church begun by Abbot Simeon in 1082 had a 13-bay aisled nave, four-bay aisled transepts, a crossing with a tower, and a four-bay aisled chancel terminating in an apse. At the W end was a second transept with E chapels and a second tower. A western Galilee porch was added in the 13thc. (1198–1215), and the chancel was extended to the E with a six-bay retrochoir, completed in 1252. In 1321 the Lady Chapel was added to the N of the choir, and a year later the crossing tower collapsed. The octagon, built to replace it, was completed by 1342, and in the same campaign the remaining bays of the 11thc. chancel were replaced. The only above-ground survivals of the original chancel are the two easternmost piers of its straight section. Elsewhere in the building, the N section of the W transept collapsed in the late 15thc., and the NW corner of the N transept in 1699. The former was merely consolidated, the latter rebuilt.

The generally accepted chronology for the standing Norman work dates the lower parts of the entire S transept and the E wall of the N transept to the period between 1082 and the start of the abbatial vacancy in 1093. Stylistic comparisons with work dated between 1118 and 1125 at nearby Peterborough Cathedral suggest that the remainder of the transepts and the nave were completed between the appointment of Abbot Richard in 1100, and the 1120s. The earliest work in the lowest levels of the W transept also belongs to the 1120s, but there is a marked stylistic break above gallery level, and thereafter such late-century features as pointed arches, keeled mouldings and crocket capitals begin to appear. These apparently belong to the campaign of Bishop Geoffrey Ridel (1174–89), who also completed the tower. Not included in this report is the new Galilee porch built, or posthumously funded, by Bishop Eustace (1197–1215). The following description of the various parts of the cathedral roughly follows the building history.

S transept:

This must begin inside the S transept, whose E and W walls have a three-storey, three-bay elevation of arcade, gallery with twin openings and clerestorey with a passage. The three arches of the E arcade were originally entrances to chapels, and the S one still is, but the other two have been blocked off with masonry and combined to form the Old Library, accessible from the S choir aisle. On the W side, the aisle was screened off in the 12thc., but the evidence of two blocked doorways in the E walk of the cloister indicates that entry to the transept was originally possible from the monastic buildings via this aisle. The arcade level to E and W (including the E chapels and W aisle and extending to the first respond of the S nave aisle wall) is the earliest standing fabric of the church. Arches are unmoulded, capitals are of the Norman volute type, some carved with foliage and animals, and the ashlar is coarsely tooled. The same features are noticeable in the lowest level of the S wall, but not in the arcade erected in front of it, which provides a support for the platform linking the E and W galleries. Above the arcade level on all three walls there is a marked change. Tooling is finer, volute capitals have been replaced by cushions, and the orders of arches have acquired angle and soffit rolls. It is normal to connect this disjunction with the abbatial vacancy that lasted from the death of Simeon in 1093 to the appointment of Abbot Richard de Clare in 1100, although Barlow has suggested that the presence of Ranulf Flambard, who managed the abbey's finances from five years before Simeon's death, and throughout the interregnum might have provided a stabilising influence.

Whatever the date of the change in design, it is clear enough that the upper levels of all three walls, the tall arcade on the S wall, and the blocking of the W aisle all postdate it. Much later comes the addition of the 15thc. Fourth storey window in the S gable. Some evidence for the original arrangement appears on the exterior, where remains of the springing of a wall arcade are visible. The E windows of the E chapels were replaced in the late 13thc., and those of the E and W galleries in the 15thc.

On the exterior the S transept facade reflects the nave and aisle structure within. The gabled, five-storey central section is divided from the two-storey aisles by broad, flat buttresses that transpose, at the bottom of the fifth storey, into two-storey octagonal turrets with pyramidal roofs. The two storeys of the turrets are decorated with blind arcading and there are cusped corbel tables under the eaves and sets of three corbels at the top of the lower storey.

N transept:

The interior of the N transept shows the same change in design as the S, but not in the same places. Only the E arcade and its two southernmost chapels have all the earlier forms, the N chapel has cushion capitals on the vault supports and a roll moulding in its N window. The N and W sides of the transept have the newer forms throughout (although roll-moulded arches do not appear on the W arcade). Examination of this area of the building is complicated by the restoration following the 1699 collapse of the NW corner. This affects the W half of the N wall, inside and out, and the N part of the W arcade and aisle wall. Other changes to the fabric in the N transept are the replacement of the windows of the E chapels, the W gallery and the N clerestorey in the later Middle Ages.

A comparison of the arcades of the two transepts reveals a rather curious disjunction. Both have an alternation of compound and circular piers, but the system has been reversed so that in the S those nearest the crossing are circular, whereas in the N they are compound. The original arrangement, with four-bay transepts and a smaller, square crossing, would have made this inconsistency even more obvious. The reason for the change surely has nothing to do with the balcony at gallery level on the end wall, but might relate to the original roofing arrangements. Both transepts have wooden roofs now, and did originally, but the present roofs date from the 15thc. In the S there is nothing to indicate the original form of the roof beyond the thin buttresses between the bays at clerestorey level only. In the N, however, a half-column respond runs the entire height of the wall between the two northernmost bays only. This would seem to imply a roofing system based on double bays, although it is curious to find the strong roof support here rather than a bay further S.

The N transept facade differs from the S in the treatment of the turrets. A square plan with chamfered and shafted angles is maintained from ground level up to their sixth storey, where they clear the gable. At this point the plan becomes circular, and decorated with a blind arcade syncopated with the cusping of the corbel table above. The E turret is original; the W is a copy.

Nave:

The original 13-bay nave had its easternmost bays communicating with the transept W aisles. This arrangement was lost with the introduction of the Octagon at the crossing, and the present 12 bays are all W of the transept. The elevation is similar to that of the transepts, with alternating round and compound piers, galleries with double openings and triple openings to the clerestorey passage, and it is remarkably regular. Unlike the transepts, however, every pier has a respond running the entire height of the wall as a roof support. Again, too, the original roof is no longer evident; the present ceiling dates from the 19thc. Pevsner finds the ambiguity between a double-bay and a single-bay system indecisive. Within this uniform system there are differences between the N and S elevations. On the N side the roof support responds have the form of half shafts against dosserets, but this system is used only for the first two piers on the S. Further W the main half shafts are flanked by a pair of smaller ones. The treatment of the circular piers also differs between the N and S elevations. In both cases the arch they carry is of three orders, and the piers are supplied with a broad cushion capital for the two inner orders linked to a smaller one for the outer order. On the N side these small capitals simply hang, with no member to support them, whereas on the S they have half-column supports coursed with the piers. These differences suggest that the N arcade was laid out before the S, but above the level of the capitals the two arcades apparently proceeded side by side. The key to this is a change of arch design at bay 5 on both sides, and a change of gallery opening design at bay 4. In the arcade arches the four E bays have angle rolls on their inner and outer orders only, whereas from bay 5 all three orders have angle rolls. In the gallery the change is the addition of a hollow moulding to the face of the inner order. Both changes seem slight, but they provide valuable evidence for the sequence of building.

Capitals are mostly cushions, but they are by no means entirely uniform. Common variables include the presence or absence of clearly defined shields, or keels or tucks at the angles, but scattered seemingly at random around the upper levels are a few scallop capitals, capitals decorated with fluting, and decorative imposts.

The interior nave aisle walls are decorated with simple blind arcading and a chevron string course at dado level. On the N side the aisle windows have been replaced, but on the S a change of design can be seen in bays 9–11, where the windows are taller, the chevron units of the string course are longer, and the blind arcade has five units per bay instead of four. The easternmost bay on this side (bay 12) has no window (because of the SW transept chapel alongside it), and is articulated with two rows of blind arcading, the upper with intersecting arches, and with opus reticulatum in the lunette below the vault. The arch into the W transept from bay 12 is richly decorated with chevron ornament. The elaboration of this bay is the only preparation for the riot of surface decoration that breaks out in the SW transept.

The gallery windows on both sides of the nave, and the N aisle windows were replaced in the later Middle Ages.

W transept:

In the SW transept, the E wall is the plainest, having two bays with single openings on the ground storey, twin openings at gallery level and triple openings to the clerestorey passage. The two lower levels of the S bay give onto a two-storey E chapel, dedicated to St Catherine, which was ruinous in the early 19thc. and was rebuilt in 1848. Only the S wall is original. The S and W walls of the transept are much more richly articulated. Each has six storeys of arcading, which include a gallery and a clerestorey with a passage. The S wall also has a triforium passage, and the remaining storeys, three on the S and four on the W, are of blind arcading of various designs. There is a stylistic progression from the lower storeys to the upper. On the lower levels capitals are predominantly simple cushions, there are scallops at gallery level and trefoil, waterleaf and crocket capitals above this. In the clerestorey arcade too the arches are pointed, all suggesting that the upper levels were not completed much before the end of the 12thc. What little remains of the NW transept suggests that it was organisationally a mirror image of its companion, but with some differences in detail, including fret decoration on the E wall gallery arches and cushion capitals where the S transept has crockets, implying that it was built first.

The largely rebuilt transept chapel is built right in the angle of nave and transept, so only two windows are possible. On the exterior, it is divided into eight bays by responds, the third and sixth bays from the S being wider and containing the ground floor and gallery windows. Bays 1, 2, 4 and 5 each contain a blind arch at both levels. The exterior of the transept displays perhaps the most spectacular and satisfying ensemble of superimposed arcading in the country: five storeys of arcading above a plain plinth storey on the main walls, rising by two more arcaded storeys on the decagonal angle turrets.

W Tower:

The cathedral is liberally supplied with corbel tables: at gallery and clerestorey levels on nave and transepts, on the main transept turrets, below the battlements of the west transept and on the turrets there, and at the eaves level of St Catherine's chapel (although these almost all date from the 1847 restoration). The quality is high, although the range of subjects is fairly restricted, including human, animal and grotesque heads and a few exhibitionists. String courses too are limited in type, the carved examples confined to billet, sawtooth, single chevron and chequer ornament.

The same system continues higher on the walls of the W tower, which, in its 12thc. lower part, rises six storeys above the nave battlements. It is square in plan with octagonal angle turrets rising no higher than the tower body. It was completed by around 1200, and provided with a spire some 30 years later. The spire was replaced by an octagon storey in the later 14thc. Presumably the tower arches inside were replaced at the same time, but the original chevron-decorated arches remain visible.

Cloister:

Of the Romanesque cloister nothing remains except sections of wall arcading on the outside of the S nave aisle, and the handsomely carved doorways for which Ely is famous. The Prior's doorway, at the W end of the cloister walk, is deservedly the most celebrated, but the remains of the Monks' and Vestry doorways at the E end of the walk are considerable works by the same carvers. All three must date from around 1130–35, and the remains of an earlier doorway into the S transept aisle also survive at the N end of the E walk.

Tombs:

Only one tomb is included in the survey; the beautiful Tournai marble slab depicting St Michael carrying a naked soul, assumed to be Bishop Nigel's (1131–69), in the N chancel aisle.

In the following descriptions, features such as piers and bays are always numbered starting at the crossing. Hence bay 1 of the N transept is the S bay, while bay 1 of the S transept is the N bay. The interior treatment of windows and doorways is only described when it includes some elaboration.

The author is grateful for the assistance of Stephen Wikner, the Cathedral Bursar, Susan Matthews, the Curator of the Stained Glass Museum, and for the cheerful and unfailing support of the clergy, vergers and cathedral guides.

Benedictine monastery (to 1109), Benedictine Cathedral Priory (to 1539), Cathedral (from 1541).

An account of the history of the abbey from its foundation by St Aethelthryth in the early 670s to 1109 when it gained cathedral status is given in Keynes (2003), and this includes the evidence for the early years of the building of the present church. According to the account in the Liber Eliensis, Abbot Simeon laid the foundations, probably in the mid-1080s, and completed the domestic buildings for the monks and laying the foundations of the church before his death in 1093. Thereafter Ranulf Flambard was put in charge, although no appointment was made to the abbacy in the reign of William Rufus. When Henry I became king in 1100, Richard was appointed to the vacancy, and the Liber Eliensis declares that he tried to finish what Simeon had begun, "in as fitting a shape and on as fitting a scale as possible", so that it would deserve to be regarded as the finest church in the land. By 1106 it was necessary to move the founder's body into the new church, where it was placed 'in a special chamber behind the high altar'. Evidence for the work at the west comes from Wharton's collection of transcripts (Anglia Sacra, I, 631). Bishop Ridel (1173–89) 'completed the new work towards the west with the tower even to the top'.

The earliest sculpture is found on the volute capitals of the entire lowest storey of the S transept and the E side of the N transept, the former carved with animals, birds and foliage in low relief. The fire-breathing dragon on the first capital of the S nave aisle also belongs to this campaign, which Zarnecki and Fernie have dated to the abbacy of Simeon (1081–93). Zarnecki's closest parallel for the animal and bird sculpture is the border of the Bayeux Tapestry, but comparisons with works at Jumieges point to a Norman origin for the workshop. The start of the second phase of work may have been delayed until the appointment of Abbot Richard in 1100, and certainly the change in capital forms from volutes to cushions produces a marked visual disjunction between the lower and upper storeys of the S transept in particular. The inference is that the Norman workshop had left, to be replaced by a large group of carvers accustomed to producing cushion capitals on a production line basis, but to slightly different designs. This would explain the variety in shield forms and angle treatment found in the upper parts of the transepts and throughout the nave, which follows no discernable pattern.

Work proceeded rapidly, so that at least some of the carvers of nave capitals remained active while the W transept was being erected. It is certainly the case that the sculptural ornament here, both inside and out, is more stylistically advanced at higher levels, but the present author is inclined to the view that this process is less gradual than has sometimes been suggested. On the interior of the SW transept, for example, there is a distinct stylistic rupture at the fifth storey of the S and W elevations, where trefoil-headed arches with cusp rolls and Gothic crocket capitals first make their appearance. Only in the sixth storey do we find pointed arches and keeled mouldings. These stylistic observations agree with the study of Fearn, Marshall and Simpson into the building sequence, in which it is concluded that work was interrupted in about 1140 at this level and not taken up again until Bishop Geoffrey Ridel (1174–89) provided the funds.

On the design sources (or the lack of them) for the Ely W transept in Britain, the Holy Roman Empire and the Mosan area, see McAleer (1963), 120–87 and Ferguson (1986), 225–73. The motif of the trefoil headed arch with rolls on the cusps is peculiarly ubiquitous at Ely. It appears on the Monks' doorway, where the present rolls are modern additions. That these were not originally present is suggested by Bentham's 1770 drawing, which shows none, although Zarnecki (1989) has argued that the presence of rolls in a doorway at East Dereham (Norfolk) is evidence for their original inclusion in the Monks' doorway as well. The little doorway to the N transept turret, has the rolls, but is not trefoil-headed, although it would be if its tympanum were removed. As it is an insertion of unknown date, speculations about its significance are of limited value, but a doorway at Twywell (Northants) has the same form, differing only in the pattern of diapering on the tympanum. Twywell is close to the Barnack quarries where the stone for the cathedral was mined, and also nearby is St Kyneburgha's church at Castor (Huntingdon). A chancel window here has a trefoil head but no rolls, and the church is dated by an inscription to 1124. Whatever the precise genesis of the trefoil headed arch with cusp rolls, it was taken up with enthusiasm in the blind arcading of the W transept and tower of Ely Cathedral.

Turning to the great doorways, their dependence on Italy has long been recognised (Kingsley Porter 1917, 235). The design of the jambs and arches of the Monks' and Prior's doorways, with alternating round and square orders covered with foliage scrolls and motifs in roundels can be paralleled in Corrente Comasco work at Pavia (S Michele, S Pietro in Ciel d'Oro), and the lion supports on the Prior's doorway, commonly found in Lombardy and Emilia, provide more compelling evidence. A further international link is provided by the composition of the tympanum of the Prior's doorway. This kind of Christ in Majesty flanked by angels is Burgundian in origin, the closest parallels for the angles' twisted posture occurring in such Brionnais churches as Anzy-le-Duc (Saone-et-Loire), which also provides a model for the corbel heads supporting the tympanum. That said, there is no doubt that the carving of figures and foliage is English in style. Since the doorways have been redated to the 1120s, Zarnecki's comparisons with Ely manuscripts of the 1130s and '40s can no longer be taken as indications of sources, but his parallels with Anglo-Saxon drawing styles, as in the Christ in Majesty in the British Library Psalter Cotton Tiberius C.VI, f.18v. or the figures of monks in Arundel 155, f.9v. remain relevant.

Iconographically the main feature of discussion is the meaning of the sequence of figures in roundels on the outer pilasters of the Prior's doorway. For Zarnecki (1958) the sculptor was not interested in symbolism, but copied such scenes as attracted him, mainly from manuscript cycles of the Zodiac and Labours of the Months, and in this he is surely correct, although whether a sculptor would have had access to painted manuscripts must be questioned. It is worth noting from this perspective the portal of Sagra di San Michele near Turin, dateable circa 1120, where the sculptor Nicholaus also carved zodiac scenes on doorjambs. In his 1989 revisitation of the doorways, Zarnecki accepted Meredith's comparison between the corbel heads on the Prior's doorway and heads on the Bari Throne. The visual comparison is certainly compelling, but it is hard to know what to make of it. More immediately he also suggested a possible workshop connection with a tympanum at Wordwell (Suffolk) where Byzantine blossoms similar to those on all three of the Ely doorways appear. The older view of the chronology of the doorways was given by Zarnecki (1958), who dated the chip-carved arch in the S transept to c.1090, and the other three doorways to c.1135. Meredith, on the other hand, has argued for a progression in style on the basis of Italian comparisons, notably with the sculpture of Niccolo at Sagra di San Michele and elsewhere. In this view the Prior's doorway is the earliest, dating from c.1125, with the Vestry and Monks' doorways dating from ten years later. To account for the surprising fact that the earliest doorway is the furthest W, she ingeniously suggested that the Prior's doorway was originally sited where the Vestry doorway is now, providing a temporary grand W entrance; the chip-carved arch that remains there being the original label of the Prior's doorway rather than the remnant of a much earlier doorway. Unfortunately the theory depends on the measurements matching. The outer diameter of the Prior's doorway, at 2.49m, is 0.29m wider than the estimated inner diameter of the chip-carved arch, so the hypothesis seems untenable. In general, however, a date in the 1120s rather than the `30s for the three great doorways fits much more comfortably into the known chronology of the building, an observation of Franklin's that Zarnecki had accepted by 1989.

Prior to the present survey, details of the nave and transept corbels have not been widely available. It is now clear that much of this work is of very high quality, and that workshops with distinctive styles were involved. This is not the place for a detailed analysis, but the S nave clerestorey corbel table includes a toothache head (bay 1, corbel 3), and humans and beasts of great charm and vitality (bay 10, corbel 4; bay 3, corbel 3) that rank among the best of this date in the country. Emphasis is often given to the mouth, deeply carved to cast a black shadow, and taking a wide variety of forms from the lolling-tongued dog (bay 2, corbel 4), to the mask of tragedy (bay 11, corbel 5). Further west, in the E clerestorey of the SW transept, we find a workshop carving caricature heads of great expressiveness, marked by pointed noses, long upper lips, and distinct nasolabial folds (S bay, corbel 1).

Within Cambridgeshire the Prior's doorway was copied in a simplified form at Kirtling. The N doorway of Wentworth church also shows a marked dependence on the cathedral, in the form of roll corbels and cushion capitals with fluted shields. The chancel windows at Coton follow the typical Ely pattern with nook shafts supporting cushion or scallop capitals and arches with angle rolls and face hollows. At Grantchester two reset corbels belong to the same world as those on the cathedral nave. The extensive use of billet in Ely string courses left its mark at St John's in Duxford, Swaffham Prior, Willingham, and St Mary Magdalene's in Cambridge. Sawtooth string courses (or remains of them) are found at Great Shelford, Haslingfield, Stretham and again St Mary Magdalene's in Cambridge. The only example I have noted of the lateral chevron string course found on the cathedral's nave aisle walls is inside the church of the Holy Sepulchre in Cambridge, but the execution is not identical and it might not even be 12thc. work.