Except where stated otherwise (e.g. when the copyright of photographs is retained), you are free to reproduce any of this work for non-commercial purposes only, provided proper and appropriate attribution, credit and citation is given to the author, any original contributor or source, and the RRRA, in accordance with the terms of this licence.

Creative Commons Licence
This work is licensed under a Creative Commons Attribution-NonCommercial 2.0 UK: England & Wales License

© Mike Haken & RRRA, 2018

Home Gazetteer of Roads Margary's Numbering Itineraries & Sources Glossary/Biography RRRA Website

Margary Number:

Other Numbering System:


South Yorks. HER 04915

South Yorks. HER 05016

Historic Environment Records, Pastscape and other online records

32 Miles

Roman Sites on Route:

Lincolnshire, Nottinghamshire, West Riding of Yorkshire


Marton Roman Fort

Gate Burton Roman Fort

Segelocum (Littleborough) settlement

Scaftworth Fortlet

Rossington Bridge Potteries

Rossington Fort

Danum (Doncaster)

It would appear that the main road from Lincoln to York was originally intended to be RR2d and RR2e, crossing the Humber by ferry and then heading north from Petuaria (Brough) to Derventio (Stamford Bridge) and Eboracum (York). However, to avoid the Humber ferry which would have been problematic for large troop movements, a second road was also built that skirted around the west of the Humberhead levels. As a route to York, RR28 was circuitous, being ten miles further, but as a route north via Newton Kyme and the Rudgate (RR280) to Aldborough and beyond (by-passing York), it was actually 2 miles shorter, and this was probably its original function.

Rather than run direct from Lincoln, RR28a branched off RR2d (modern day A15)  along what is now the access road to the Lincolnshire Showground, presumably to avoid wet low lying ground. It then followed a single straight alignment, nearly ten miles long, through North Cliff Farm, then along what is now the A1500 (Till Bridge Lane, followed by Tillbridge Road and Stow Park Road) as far as Marton, after which the alignment is followed by Littleborough Lane to the crossing of the R. Trent. The road bisects two Roman fort sites on the east bank of the R. Trent, one possible (Gate Burton), one probable (Marton), neither having been investigated. Some time later, a short cut was constructed (the recently discovered RR28aa) which cut the corner at the Showground by heading direct from Lincoln to a point just south of Scampton, cutting about ¾ mile off the journey.

The Roman crossing of the R. Trent is traditionally by a ford, presumably the one that was removed in 1820 (Riley, Buckland & Wade, 1995 p. 263).  In general, it seems highly improbable that the Roman army would expend all the necessary resources to build roads, only to risk them being unusable when rivers were in spate by building fords. Bridge building was straightforward given their labour resource, so that was almost certainly what they did . In this instance however, there is not yet any archaeological evidence for a bridge, and it is always possible that this crossing of a tidal river may prove to be the exception. Immediately across the Trent and north of the road is Segelocum (Littleborough) accurately recorded in both Iters V and VIII of the Antonine Itinerary and a milestone found in the centre of the Upper Colonia in Lincoln as being 14 miles from Lincoln, so there can be no doubt about identification. It is a substantial Roman settlement where occupation probably began in the 50s (Hartley & Dickinson 1995, p.267) or at the very latest the early 60s A.D. (Riley, Buckland & Wade, 1995 p. 260), so roughly in accordance with the likely establishment of a Legionary fortress at Lincoln. The street layout, which has been mapped from aerial photographs, does not  relate to the only known road past the site, RR28a (fig. 2). Its orientation may have been remodelled following a devastating flood event (ibid. p. 263), or simply determined more by topography and the location of river wharves than by the road. Changes forced by flooding and alluviation might also explain why the road does a slight dogleg as it passes the town.

It is clear that there was a change in the alignment of the road at Segelocum, with the road now heading along a line through Sturton le Steeple, South Wheatley and North Wheatley to the top of Haughgate Hill some 5.3 miles WNW. The change of alignment seems to have been at about SK 8240 8255 on the western bank of the R. Trent. If this is correct, then according to John Poulter’s ideas about “best view” it is highly likely that the alignments were set out from west to east, the opposite of what might have been expected (Poulter, 2009, p.27). The line to Haughgate Hill appears to be preserved fairly well by the modern road stretches of which fall perfectly along a straight line, although it must be said that at the time of writing the road has not been located with certainty at any point between Segelocum and Haughgate Hill, and geophysical survey commissioned by Nottinghamshire County Council in 2012 to attempt to find it near Sturton Le Steeple was unsuccessful (Horsley, L., Pers. Comms. 9 Oct 2017).

From Haughley Hill, the surveyed alignment changes again. A line can be projected from here that falls precisely along the one and a half mile straight stretch of the road past Scaftworth fortlet to Bawtry, and this was clearly the principal surveyed alignment. To avoid un-necessary crossings of the R. Idle, and some low-lying marshy ground, the road appears to have been laid out along two additional alignments slightly to the north, first from Haughgate Hill through Clayworth to Drakeholes (SK 7059 9046), and then from Drakeholes past Everton, on a line just north of and parallel to the modern A631, returning to the principal surveyed line at SK 6745 9171, east of Scaftworth. The evidence for these two alignments comes mainly from modern roads and a few hints in lidar, but there can be little doubt that the alignments are correct. The actual point where the survey alignment changes is on the eastern side of Haughgate Hill, from where Bawtry cannot be seen, so again it would seem that the alignment was set out from west to east (Poulter 2009, pp. 26-7). This is further corroborated by there being no clear visual sighting point at Bawtry.

As the road descends to cross the R. Idle near Bawtry, it shows very clearly on lidar (fig. 4), although curiously the Scaftworth fortlet does not. The road was excavated in 1995 as part of the English Heritage commissioned Humber Wetlands Survey, the work revealing two distinct phases of construction. The first was a causeway using a corduroy construction of tree trunks laid in three parallel lines, with smaller branches laid at right angles on top. A further layer of brushwood was applied, to pack the uneven surface, and then finally a layer of turves placed on top (Van der Noort & Lillie, 1997). The excavators did not find evidence of any metalling above the turf, but in such a wet environment it is hard to see how metalling could be avoided. The second phase, which partly overlay the first, was a gravel agger between rows of oak pegs some 6m apart and solved the problem of crossing the marsh in a different way. Rows of large oak piles (dated by a single carbon date to the third century AD) were sunk at right angles to the line of the road, between two rows of oak pegs, forming a series of “compartments” to stabilise the agger of the new road (ibid.). The fortlet at Scaftworth appears to post date the second phase road as all the recovered pottery is of 4th century date. However aerial photos clearly suggest at least three phases to the site, so a 1st century fort cannot at this point be ruled out (fig. 5).

From Bawtry northwards, the line of RR28e from Lincoln to York is marked today by the long straight stretch of the Great North Road (A638). In point of fact the road is not quite straight, and deviates to the east of the surveyed Roman alignment by 60m close to the entrance to the Northern Racing College, before returning to the alignment near Rossington Bridge. When projected northwards, this alignment runs only just west of the Roman fort at Roall Hall by the R. Aire, where it appears to meet another surveyed alignment northwards that runs through Tadcaster and was utilised both by Rudgate (RR280) and by Dere Street (RR8) (Poulter 2010, pp.45 & 48). These survey lines appear to be part of a larger system of surveying that eventually formed the basis, further north,  of what we know as Dere Street and was set out from south to north. The question then arises, as the surveyed alignments followed by RR28a between Scampton and Bawtry appear to be set out west to east, are we looking at two different surveys and potentially two different timescales?

The road planning around Rossington Bridge is clearly not straightforward (see fig. 6). At Rossington, the modern road (A638) bears away to the north-west but the Roman line carried straight on, marked by a footpath along the western edge of Hurst plantation, the agger still being just about discernable in a few places. In Margary’s day it was still visible (Margary 1973, p. 411) as it approaches the corner of the modern housing estate on the other side of the R. Torne but much of this area has now been destroyed by the construction of a new road, the “Great Yorkshire Way”. The Roman road passed between the sites of Roman pottery kilns on either side of the river, before entering a cutting which eases the slight ascent, from which the road emerges into the trees that line the edge of the estate, presumably all planted in the agger of the road. The line of the road is now taken up by the curiously named Warning Tongue Lane, which borders the estate, before RR28a bears north west beneath the estate, although exactly where has never been clear. Warning Tongue Lane, however, keeps straight on along the Roman line, now a branch road known as the Cantley Spur (RR281), heading to the fort on the R. Don at Long Sandall.

Margary described RR28a as marked by a line of hedgerows past Bessacar Grange, but this line is sinuous and unlikely (fig. 7), and he then assumed that Cantley Lane (previously known as Sand Lane) marked the line before heading along the edge of Doncaster racecourse, an unnecessarily indirect route (ibid.), as was pointed out by White as long ago as 1963 (White, 1963). Lidar has, however, revealed a short length of road approaching Warning Tongue Lane from the east, which might suggest a crossroads, rather than RR28a simply bearing north west. This is a little to the north of where Margary suggested RR2a bears to the North west, but when projected does align reasonably well with a length of road just east of Doncaster marked as Roman on the 1854 OS map, one that has inexplicably become disregarded in favour of Margary’s supposed line along Cantley Lane then across the racecourse. The line of the road shown in 1854 is partly preserved today as “Roman Road” in Doncaster, and by the old part of South Parade, and the Roman road on approximately this line has been confirmed by excavation in Hall Gate and High Street (South Yorks HER Monument no. 05016) passing to the south of the Roman fort at Danum, and then in Trenchgate (Buckland & Magilton, 1986, pp 43 - 49) as it headed for a bridge across the R. Don.   

Archaeological Services WYAS. (2004) 10-14A Hallgate Doncaster South Yorkshire: Interim Report.

Archaeological Services WYAS. (2008) 10-14A Hall Gate Doncaster South Yorkshire: Archaeological Evaluation.

Bidwell, P. & Hodgson, N. (2009); The Roman Army in Northern England ; The Arbeia Society

Buckland, P C and Magilton, J R. (1986); The Archaeology of Doncaster, The Roman Civil Settlement BAR British Series no.148, Oxford

Buglass, J. (2002); Archaeological watching brief on Aughton Water Main Rehabilitation, near Morthen/Report No 02/19. Northern Archaeological Associates [assessment & evaluation reports].

Codrington, Thomas (1903); Roman Roads in Britain, Society for Promoting Christian Knowledge, London

Hanson, W.S. & Campbell, D.B. (1986); The Brigantes: From Clientage to Conquest;  Britannia, vol 17 pp. 73-89; Society for the Promotion of Roman Studies, London

Hartley, B.R. & Dickinson, B. (1995); The Samian in  Riley, D.N. , Buckland, P.C., Wade, J.S. (1995); Aerial Reconnaissance and Excavation at Littleborough-on-Trent, Notts; Brittania vol. 26, pp. 267-269

Margary, Ivan D. (1973); Roman Roads in Britain, John Baker, London

Poulter, J., (2010). The Planning of Roman Roads and Walls in Northern Britain; Amberley Publishing, Stroud

Riley, D.N. , Buckland, P.C., Wade, J.S. (1995); Aerial Reconnaissance and Excavation at Littleborough-on-Trent, Notts; Brittania vol. 26, pp. 253-284

Van der Noort, Robert & Lillie, Malcolm (1997); Scaftworth: a timber and turf Roman road ; Current Archaeology, vol. 151; London

Warburton, J., 1720. A New and Correct Map of the County of York in All its Divisions, London.

White, D. A. (1963); Roman Road from Bawtry to Doncaster; Yorkshire Archaeological Journal, Part 161, Vol XLI, pp. 20 -24

Click Images to enlarge

RRRA Forum for RR28a


Fig. 3. 2005 aerial photo from Google with Segelocum (Littleborough) clearly showing as cropmarks, and suggested line of RR28a.

Fig. 6. Lidar image showing the road arrangement at Rossington, Doncaster with RR28, RR281 and RR282, along with two access roads to the Fortress and a spur along the line of RR28a running down to the R. Torne

Fig. 1 The junction of two roads still in use today, RR2d (Ermine Street) and RR28a at the Lincolnshire Showground

Fig. 2 The junction of RR2d (Ermine Street) and RR28a, together with recently discovered RR28aa which cuts the corner.

Fig. 4. Lidar image showing RR28a descending to cross the R. Idle east of Bawtry, past the fortlet at Scaftworth which cannot be seen.

Fig. 5. Aerial photo taken c. 1970 looking north, showing the triple ditched fortlet at Scaftworth - RR28a is just visible. Taken by Eric Holder, on one of Derrick Riley’s many aerial sorties. © Eric Holder LRPS 2017.

Fig. 7. 1st Edition Ordnance Survey 6inch to the mile published in 1854, surveyed 1849-50. A length of Roman road is marked east of Doncaster, which was inexplicably ignored by Margary and others. The length of agger showing on lidar in fig.5 aligns reasonably well, and certainly better than some known lengths of Roman road, but it is far from perfect.

The study of LiDAR imagery in the vicinity of Cantley and Rossington  has also revealed a road aligned along the SW side of the mid 1st century fort or fortress at Rossington which appears to link the fortress to RR28a as it heads to WNW to Doncaster. It runs across what is now Doncaster Golf Club, an area known as The Warren, and is no longer visible on the ground, although it was marked on the 1854 OS map as “Site of Old Road” (fig. 6), leading to an area known as Castle Hills. Extant earthworks have been associated with a supposed medieval moated site for which there is no evidence, and the name seems more likely to relate to the Roman fortress on the other side of the R. Torne. If the line of the road is projected south east, it aligns perfectly with the Roman crossing of the R. Idle at Bawtry, although there is no sign that any road was actually built south east of the fortress. Another short road was built, however,  just north of the fortress, to link with RR28a near Hurst Plantation. Finally, there is a road known only from aerial photographs heading due south from the fortress, destination unknown, which has been given the number RR282.

Understanding the sequence of events responsible for this arrangement is far from easy, however one thing seems clear - the road we now think of as the main Lincoln to Doncaster road (RR28a) was not conceived as one entity, rather it seems to have evolved from at least two and possibly three different roads, each planned with its own function. The layout suggests that the WNW - ESE aligned stretch of RR28a to Doncaster from the R.Torne, existed before the branch road to the fortress, along which the fortress was aligned. If this is correct, the obvious implication (given the probable mid 1st century date for the fortress) is that there was a fort at Doncaster in the mid 50s or earlier, roughly contemporary with Templeborough further up the R. Don, despite the usual thinking that the fort at Doncaster was founded under Petilius Cerialis in c.AD71 (Bidwell & Hodgson 2009, p.131). Presumably the early fort is on a site a little to the south but which has not yet been identified.   The length of access road to the fortress was clearly intended to be much more when it was laid out, as it is aligned precisely on the west bank of the R. Idle below Bawtry where RR28a crosses. In other words, it seems that the original plan was to build a road along this line, and not along the route of RR28a that was actually constructed. This suggests that the the short newly discovered spur east of RR28a, as far as the R. Torne, might be part of an earlier scheme to cross the R.Trent near Gainsborough, as the modern road does, but this scheme was later abandoned in favour of a route over slightly higher ground via Segelocum. Finally, the surveyed alignment beneath RR28a from Bawtry northwards goes to the fort at Roall Hall, with a second alignment from there underlying parts of the Rudgate (RR280) and Dere Street near Aldborough (RR8). This seems to suggest a plan to construct a road across the Humberhead levels northwards (but not to York, as is often claimed), and the Cantley spur (RR281) north from Rossington to the fort at Long Sandall was probably part of that scheme, that was never completed.

This route, branching as it did from RR2d four miles north of Lincoln, was not the most direct. A combination of lidar and aerial photographs now suggest that a shortcut road was constructed some time later, cutting about ¾ Roman mile off the distance to York. This newly discovered route, which we suggest should be numbered 28aa, diverges from RR2d 150m after leaving the north gate, keeping just to the east of the steep scarp slope that runs from Lincoln to North Carlton.

Fig. 8.  The Roman road network around Rossington and Cantley, showing the probably directions of setting out. Roads are marked in red, planning alignments in blue. The Roman potteries that were extensive between Doncaster and Auckley have been left off this figure as they were mainly 2nd and 3rd century, well after the roads were all constructed.

Margary's Roman Roads in Britain, road 28a

Entry prepared by Mike Haken, last updated, 2 June 2017

Historic Counties: