Both roundels are still in situ, placed high up on the south face of the nave south wall, and visible from within the south aisle.

The roundels appear to be in their original position.

St Michael

Good. Both appear somewhat weathered.

Two roundels set above a horizontal string-course of plain rectangular section. 2a is connected to the string by a fillet as broad as the roundel itself, whilst 2b, which is set somewhat higher above the string, is joined to it by means of a taller stem. Both roundels are framed by a double roll moulding within which are carvings in relief.

Roundel 2a (west): The decoration in the complete roundel is composed of four strands of foliage disposed so as to form a cross. At the point where the four stems join, four small pellets in the interstices emphasise the intersection. Each stem spirals away from the centre and terminates in a single leaf. These elongated, heart-shaped leaves have pointed, curling tips and their bodies appear to be hatched with diagonal strokes (although it is not easy to see this detail through the whitewash). At least two of the leaves are also decorated with a single lobe at their junction with the stem. The leaves pass alternately over and under their own stems. There appears to be a pellet attached to the outside of the rim at the top left, although it is unclear whether this is an original part of the design.

Roundel 2b (east): This roundel is less complete than 2a. The eastern third is invisible (and is possibly destroyed) behind the junction between the south nave wall and the east wall of the south aisle, whilst the uppermost third has been hacked away to make room for the first joist in the south aisle roof. The roundel has also had an electrician's conduit and junction box placed over it. The original design was an equal-armed cross with large circular bosses decorated with concentric incisions at the terminals (Bailey and Cramp 1988, 34, fig. 7, type 3i or 3ii). The arms of the cross, also decorated with incised lines, are curved and centre on a small incised lozenge. In the single surviving interstice a single leaflike shape curls inwards towards the centre of the design. This shape lacks definition partly because of the overlying whitewash and partly because it is obscured by the junction box. The tip of a second 'leaf' in the corresponding position in the eastern part of the roundel is just visible, emerging from behind the east wall of the south aisle.

Although earlier accounts had considered these two roundels to be Romanesque, Dr Taylor (1963) identified them as in situ Anglo-Saxon work. He observed that the wall in which they sit, which is clearly earlier than the thirteenth century (the date of the south arcade which is evidently punched through the string on which the roundels sit), is of the limited thickness (only 2 ft 9 ins) characteristic of other pre-Conquest examples, but very rarely found in Norman masonry. Dr Taylor felt that the roundels were in situ in this masonry, and that roundel 2a, at least, is decorated with sculpture in an undoubted Anglo-Saxon style. We accept Dr Taylor's reasoning and agree that the wall, and the building it represents, should be dated by the sculpture in the roundels to the Anglo-Saxon period.

Fortunately, it is relatively straightforward to date the sculpture more precisely. Roundel 2a is decorated with leaves so similar in style to those on faces A and C on the Edenham 1 shaft that, given their occurrence in the same location, they must be related. As Taylor cautioned, it is just possible that the carvers of the roundels were merely copying an older monument at a much later date, but in such cases, it is usually the overall designs which are copied whilst minor details, such as leaf types, tend to vary. Given the close similarities with the Edenham shaft, it is suggested here that the Edenham round els are probably of a similar mid ninth-century date to the shaft itself.

If this is correct, then the building of which fragments still stand at Edenham is remarkable indeed. It must predate St Peter's, Barton-upon-Humber, by at least a century, making it easily the earliest standing building in the county. Unfortunately little is known of its plan – it is uncertain whether the nave north wall is also part of this building, although Taylor reported notes in the church which recalled that the chancel arch prior to 1808 was round-headed (1963, 6). Some further information may be hinted at by the fact that the stems attaching the two roundels to the string below are quite different – in shape and length – and this may well suggest that the two were not part of a uninterrupted sequence of panels. Instead this discrepancy may be explained by a structural division in between the two roundels, such as a projecting porticus. Although we have found no additional evidence for such a porticus, we have included such a feature on the reconstruction in Fig. 17.

This building's decoration is exceptional. Solid strings of rectangular section, like the one supporting the Edenham roundels are the norm in English architecture before the Conquest, but external decoration in the form of carved panels is quite unusual – though not unparalleled. There may have been something similar decorating the church at Abingdon, Berkshire (Tweddle et al. 1995, 249, fig. 36). It may be that the exterior of the monastic church at Saint-Riquier in north-east France was decorated with roundels of this general sort (Parsons 1989, 4, fig. 1), but the engraving of 1610, which provides our only evidence for its appearance, is probably not reliable in such details. External decorative panels, however, occur at Barnack church, Northamptonshire (eleven miles south), where rectangular panels carved with foliage designs are set above string-courses of rectangular section on the upper stage of the tower (Ill. 489). As at Edenham these panels are placed for mere decorative effect and serve no architectural function. Various dates have been allocated to the Barnack tower from the early ninth to the late eleventh century (Fernie 1983, 139–41, fn. 5). Considering the apparently more secure date for the Edenham roundels, however, a date earlier in this bracket may be indicated for the Barnack panels – as suggested by Cramp (1972; 1977) and Taylor (1966).

Further evidence for external decorative panels on major churches of the late eighth or early ninth century is found at Breedon-on-the-Hill, Leicestershire (forty miles west), where lengths of frieze (and perhaps sculpted panels as well) were clearly intended to be disposed in a similar way (Cramp 1977; Jewell 1986). If the decorated frieze sections above the arcade at Breedon are still in situ, as Taylor argued (1966, 30), then they provide a good parallel for the placing of theEdenham roundels high up along the flanks of the building. There are no real similarities of sculptural style between the decorative external panels at Edenham, Barnack and Breedon, but they can nevertheless be seen as examples of a tradition of architectural decoration in eastern Mercia. If we accept the current dating of these three groups of panels then this tradition appears to last a century or so; from the late eighth century (Breedon) through the mid ninth century (Edenham) towards the late ninth or tenth century at Barnack.

Mid ninth century

(—) 1889–90a, x, and fig.; (—) 1913–14b, 319–22; Cox 1924, 122; Taylor and Taylor 1963, figs. 2–3, pl. III, 1–2; Pevsner and Harris 1964, 26, 522; Taylor and Taylor 1965, 227, pl. 459; Taylor and Taylor 1966, 35, 50; (—) 1971, 8; Taylor 1974a, 298, fig. 4; Bailey and Cramp 1988, 164; Pevsner et al. 1989, 43, 268; Webster and Backhouse 1991, 242, fig. 28; Gem 1993a, 47, pl. VIII, A; Tweddle et al. 1995, 249