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:


Not yet in HER

Cheshire, Derbyshire, West Riding of Yorkshire

Melandra (Glossop Heritage Trust) (Pastscape)

Highstones  Fortlet (Derbyshire HER) (Pastscape)


The existence of a Roman road heading north east up Longdendale has been speculated upon since the recognition of the Roman Fortlet at Highstones in 1978 (Hart 1981, p.90), but never found. However, recent work by the Glossop and Longdendale Archaeological Society (Hargreaves 2017) has put the existence of the road beyond doubt. The crucial discovery was made by Steve Whiteley, who identified a straight alignment on lidar (fig. 1) just south of Bottoms Reservoir, Tintwistle, and a second straight alignment on Google Earth (fig. 4). Much of the information for the following account was provided by Roger Hargreaves.

Longdendale, known to motorists as the A628 Woodhead Pass, has for millennia been an important cross-Pennine route and salt road leading into Yorkshire from northern Derbyshire and Cheshire, and was the first such route to be turnpiked in 1732 (Manchester to Saltersbrook Turnpike Trust). Towards its western end, on an elevated site overlooking the River Etherow, is the Roman fort known as Melandra. The origin of the modern name is probably an Antiquarian creation of the 18th century (Bidwell & Hodgson 2009, p95), it being generally accepted these days that the Roman name was Ardotalia (Rivet & Smith 1979, pp. 256-7), probably borrowed from the name of the river, a common Roman practice. The fort was originally a timber structure built in the 70s, later rebuilt in stone, then abandoned c. AD140 and never re-occupied (Bidwell & Hodgson 2009, p95).

The road presumably left Ardotalia at the north gate but, unfortunately,  landslips since the Roman period have probably removed all trace of it as it descended to the Glossop Brook. One possibility is that it may have followed a direct alignment to what is now Hadfield (Hargreaves 2016), part of Hadfield Road being aligned with the north gate in what was the medieval heart of the village, but this could be no more than chance (route A, fig. 2).  Another possibility is that after crossing the Glossop Brook, the modern Green Lane approximates to it, running into Hadfield Road, through the village  of Hadfield and then on to Kiln Lane (route B, fig. 2). The bend in Kiln Lane possibly preserves an alignment change on the northern side of a slight ridge. Certainly, the northern end of Kiln Lane, along with Lambgates and then the hollow-way past Roughfield (the medieval saltway heading up the dale), are all in approximate alignment with the straight length of Roman road visible on lidar imagery, just south of Bottoms reservoir.

When the railway converges onto the likely course of the road, the road’s potential course becomes less certain. There is certainly a straight feature on lidar imagery as far as Deepclough Farm which may represent the road, but nothing shows to the naked eye on the ground and this could just be a temporary road created during railway construction (Hargreaves 2017 pers comm). East of Deepclough, the course is probably approximated by the public footpath, which the first edition Ordnance Survey map shows was straight between SK 045 979 and SK 051 981 prior to the construction of the railway and of the reservoir in 1865-9 (Hargreaves 2016). Projecting this across the valley, now beneath the Rhodeswood and Torside Reservoirs, the path aligns very closely with a short straight length of the first Manchester and Saltersbrook Turnpike road of 1732, just south of the Highstones fortlet (fig. 3).

Whilst there is no archaeological evidence (yet) that the Turnpike road follows a Roman line, this does seem highly likely. There is no sign of the medieval saltway, or any other early trackway or road and, despite some quite difficult ground in places, no braiding.  Braiding is very common near moorland routes, but not here, suggesting that the earliest route remained in use.

The Roman crossing of the Crowden Brook is not clear, however just east of the brook an old road then runs for over half a mile heading ENE, absolutely straight across quite undulating ground, passing beneath the modern road and earlier turnpike road. It survives mainly as a holloway, but there are possible traces of agger and a Roman origin seems probable, but not yet proven. It is briefly rejoined by the turnpike, which soon drops away eastwards to head to Woodhead Bridge. The probable Roman line keeps to higher ground, as would be expected, although becomes obscured by works related to the disused Enterclough quarry at SK08559988. Once east of the quarry,  a very obvious old road contours around the hillside, and drops down on a straight alignment (fig. 5) to disappear beneath the Huddersfield and Woodhead Turnpike road at SK 095 005 (now the A6024), opened in 1768.

This is as far as the road has yet been traced, where it appears to be taking the unlikely heading up the valley of the Heyden Brook towards Holmfirth, although an equally possible route would be up the valley of Withens Brook and over Withens Moor (Hargreaves 2017), crossing the intriguingly named Bailie Causeway Moss. The most likely destination would seem to be Castleford in West Yorkshire. The problem with this is to understand why such a road should have been constructed, since it would run almost parallel to the main Chester to York road (RR712). The roads would be about 7 miles apart, and the duplication does not serve any obvious purpose. It had previously been assumed however that, rather than head up the Heyden Brook, the Roman route would have continued up Longdendale along the route followed by the later salt traffic and the A628 towards Penistone, and then on to a point somewhere between Doncaster and Castleford, as yet not known. If the road contouring around the hill to head up the valley of Heyden Brook is Roman, then this latter alternative seems highly unlikely, unless of course there is a second road forking off at Woodhead.

Whichever alternative turns out to be correct, the distance to the road’s eventual destination is simply too far without intermediate forts or fortlets. There are none yet known that would serve to protect any of the putative routes. Hopefully, all these unknowns will be eventually be resolved.

Fig. 1. Lidar image showing agger of Roman road and possible road features, south of Bottoms Reservoir, Tintwistle, Derbyshire

Fig. 4  Google Earth aerial photograph from 2005 showing a linear feature east of Crowden apparently beneath the modern and turnpike roads

Click Images to enlarge

Fig. 5 Possible Roman road aligned with later Huddersfield and Woodhead turnpike road at Woodhead.

Fig. 2  Map showing possible routes from Ardotalia through Hadfield


Fig. 3  Aerial photo of Highstones fortlet, with the turnpike road bottom right

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

Hargreaves, R. (2016); Draft Report on Melandra-Woodhead road, Glossop and Longdendale Archaeological Society

Hargreaves, R. (2017); A Roman Road in Longdendale., In RRRA Newsletter Vol. 5, 2017., RRRA, Northallerton

Hart, C. R. (1981); The North Derbyshire Archaeological Survey. A. Wigley & Sons: Leeds

Rivet, A.L.F. & Smith, Colin. (1979); The Place Names of Roman Britain, B.T.Batsford, London

Reproduced with the permission of the

National Library of Scotland

© Roger Hargreaves 2016

6.2 miles identified

Historic Counties:

Roman Sites on Route:

Historic Environment Records, HE Pastscape and other records

Entry compiled by Mike Haken. Last updated, 25 January 2018