In 1882, improvement of the navigation of the Trent by dredging revealed two bridge piers 5 miles downstream of Newark, Nottinghamshire, near the village of Cromwell (SK 80886122, Figure 1). Before blowing up the piers with eight charges of dynamite, employees of the Trent Navigation Company measured and planned them. After this drastic method of excavation, the pier timbers were re-assembled and photographed. This information was later published as a record of a Roman bridge (Compton & Brooke 1885) with a drawing by J.H. Wheldon of the Trent Navigation Company (Figure 2).
The piers, 29 feet (8.8 m) apart, were thought, on the basis of the width of the modern Trent, to be the only ones left out of seven. Although no scale is published on the drawing, the text says that the piers were about 29 feet long. Each pier base consisted of a lozenge-shaped timber frame, fastened to a horizontal central balk by 4 tie-pieces. The frame was filled with rubble to form the foundation of the pier, and the central balk was held in place by four piles driven through oblique mortices, 11 feet (3.4 m) into the river bed. One of these piles was found in situ.
The only type of joint that can be identified from the drawing is a `tusk-tenon' formed by inserting the end of one beam, shaped into a long tenon or `tusk', into a mortice, cut through the entire thickness of a second beam; the tenon protrudes far enough to be penetrated in turn by a through mortice, which holds a peg or wedge. The frame and the tie-pieces were joined by simple square-headed tusk-tenons, but the joint between the central baulk and the tie-pieces used a unique octagonal-headed type. This detail is confirmed by a photograph published a few years later (Figure 3; Page 1906: 25).
Doubts were expressed about the Roman date even at the time of discovery, especially as Ad Pontem (Thorpe, Nottinghamshire), the expected site of a Roman bridge is well south and west of Newark. Scafforld Lane, a spur from the Fosse Way north of the Roman fort of Brough Crococalana, which appeared to lead towards the site of the Cromwell bridge (Figure 1), was not recognized as a Roman road, a designation given by the modern Ordnance Survey. Trial trenching by Chris Taylor of the Trent and Peak Archaeological Trust has shown no evidence for a road or causeway on the flood plain east of the bridge (1991 pers. comm.)
Of the many timbers salvaged the only known survivor is a fragment of a boxed heart oak balk, 111 cm x 37 cm x 34 cm with a through mortice at right-angles to the surface, now in Newark Museum (catalogue number 879/1.1). The original accession label called it the `base beam of the Roman Bridge' and it looks very similar to the central balk shown on the photograph (Figure 3). This is not a timber that could be replaced, unless the whole pier was washed away.
A slice sawn from the end of the fragment allowed Robert Howard of the Nottingham University Tree Ring Dating Laboratory to measure the 250 surviving rings along 3 radii. This sequence (reference number CMF-B01) was compared with a series of National Reference Chronologies in addition to East Midland, Baltic, Roman and prehistoric chronologies. The only cross-match was against the British Isles Chronology MGB E01 and its excellence is indicated by the exceptionally high value of 7.6 on the Student's t test (Baillie &, Pilcher 1973), which is usually considered significant if it reaches a value of 3.5 (Baillie & Pilcher 1973). The first …