The excavation site during November 2009 floods.

After the storm: new Roman remains at Cockermouth

Devastating floods in 2009 wreaked havoc in Cumbria; Lynda Howard, however, found a silver lining amongst the clouds in the discovery of enigmatic Roman remains, exposed by the deluge.


Mention Cockermouth, and most people will think of the November 2009 floods which caused so much devastation, or perhaps more recently, the resulting £276 million repair bill that the county has undertaken during the past year. The floods had a positive element, however: in a field just outside the town, beneath the Roman fort of Papcastle and stretching across the floodplain, there appeared an enigmatic series of finds, including walls to the base of a possible altar.

The town of Cockermouth is a thriving commercial centre at the confluence of the rivers Cocker and Derwent, and has a well recorded history stretching to the Middle Ages. The origins of Papcastle, close to Cockermouth but elevated above the floodplain, go even further back to Roman times, and perhaps earlier. Much is known about the Roman fort (Derventio), which was originally a crossing point of the River Derwent, although little remains of the final 3rd-century structure. The majority of the fort’s stone was removed to build Cockermouth castle, and what remains is buried beneath modern housing. Occasional glimpses appear in gardens and during construction work. The surrounding vicus was discovered, and publicised, by the visit in 1998 of Channel 4’s Time Team, who made the significant find of a bronze mirror. The surrounding floodplain, however, was thought to be devoid of major archaeology.

The excavation site during November 2009 floods.
The excavation site during November 2009 floods. Photo: Greg Greenhalgh.

The archaeological clues revealed by the floods of November 2009 raised sufficient interest for a local project, Unlocking Hidden Heritage (run by Grampus Heritage and Training Ltd), to begin an investigation. A geophysical survey of the fields surrounding the newly uncovered remains was carried out and extended to areas around the Roman fort, which had never before been surveyed. The survey was carried out by a profusion of local volunteers, who had already done similar surveys in the Bassenthwaite catchment, and a geophysical survey in the vicinity of Castlerigg Stone Circle, which had uncovered a Roman marching camp.

LEFT David Jackson, from North Pennines Archaeology, examines a Samian ware bowl. BELOW The Geophysics plot of Broomlands field, with the position of the main features outlined: 1. Possible Roman bath house. 2. Roman road. 3. Large circular structure. 4. Roman Mill.
David Jackson, from North Pennines Archaeology, examines a Samian ware bowl. Photo: Mark Graham.

The magnetometer survey results were remarkable. A clear image of ditches, possible roads, a circular structure, and buildings emerged. An area of ironworking was recognisable and confirmed by the presence of fragments of slag. The hunt was on for a possible Roman river crossing point.

In August 2010, Grampus Heritage and Training Ltd (funded by the Heritage Lottery Fund), organised a community-based four week dig, overseen by North Pennines Archaeology. Eight trenches were opened, initially by machine, as much of the surface topsoil had been disturbed by the previous flooding. The entire field had been scoured by a flood channel, and a large amount of new cobbles and other debris had been deposited. Some of this had already been ‘reinstated’, but the majority of the site was undisturbed at the lower levels.

Initially, it was difficult to distinguish well- layered floodplain gravels, silts, and sand deposits from materials deposited by human action, but occasional finds of Black Burnished ware and even Samian ware fragments spurred on the enthusiastic team. Several clear ditches emerged, as well as post holes and burnt timbers, and many rusty nails. What became immediately obvious was that the erosive power of regular flooding, and possibly previous restorations, had removed everything but the very bases of the structures that had been so clearly visible on the geophysics.

The Geophysics plot of Broomlands field, with the position of the main features outlined: 1. Possible Roman bath house. 2. Roman road. 3. Large circular structure. 4. Roman Mill. IMAGE: Grampus Heritage.

Work progressed unhindered by rain, sunshine, and the occasional visit from the locals (equine variety!). Teams with metal detectors covered the entire 20m2 of the field, finding everything from coins to tractor bolts.

The near miss…

The trench closest to the present river channel was just about to be closed, having revealed little more than what appeared to be an iron-rich, in-filled ancient riverbed, when the last minute discovery of a further nail required investigation. A few inches further down, a piece of wood appeared in the section wall, along with a dressed stone a few metres further along the trench. We had obviously just caught the corner of something quite interesting. Speculation on this ran from a boat to bridge structures, or a Roman mill race.

ABOVE Map to show excavation location at Broomlands field near Cockermouth. BELOW Sherds from a pottery colander.
Map to show excavation location at Broomlands field near Cockermouth. Map: Ordnance Survey and Ordnance Survey of Northern Ireland.

The digger was brought back in to laterally extend the trench, and a series of pumps installed to reduce the ever-encroaching water table. More wood emerged, cut into planks and clearly jointed, together with some well-preserved dating evidence in the form of a mortarium (Roman kitchen vessel) and small beaker. It became apparent that this was, indeed, the mill race of a Roman mill, with 8m of wooden lining and a substantial stone retaining wall still present. This also explained the abundance of slate fragments that we had already recovered, which would have formed part of the roof. The iron-rich in-fill may have derived from a later iron working site, which has shown up on the magnetometer survey.

Sherds from a pottery colander. Photo: North Pennines Archaeology

Roman mills are unusual in the north of England, and this one is one of only a handful, with another known at Haltwhistle Burn, and another at Chesters fort on Hadrian’s Wall. The small building (around 6.3m long) may possibly have had two coaxial undershot wheels, as there was evidence of a second, parallel channel. Inside the building, we found the remains of iron fixings, possibly for the grinding wheels, and mechanism – but no grindstones were found. The mill may have been used for grinding corn or powering another industrial process, such as metal smelting. Environmental samples are still being processed, but hopefully will reveal more details that may include grain, weevils, or metallic slag. Set beside the Roman mill were the shallow remains of a substantial building, approximately 20m by 8.5m, with a possible tower at one end – suggested by the presence of a central support base, with cobbles in clay foundations. Whilst the structure is Roman in age, dated from a coin found in the foundations, the function of this building is unknown.

ABOVE Amphora rim and handles. RIGHT Inscribed stone.
Amphora rim and handles. Photo: North Pennines
Inscribed stone.

Modern Cockermouth has several historically recorded mills and was always a textile area; the town had a fulling (a process also known as felting) mill by 1156, and probably had earlier structures. By 1829, the number of industrial sites had significantly increased, including corn millers, woollen firms, cotton manufacturers, tanners, and flax and linen manufacturers. Fitz Mill, a flax mill listed on the 1867 OS map, still has visible remains just upstream on the River Derwent, some of which were washed away in the 2009 flood.

Domestic or industrial?

Inland of the mill structures, where there was less damage, there was evidence of a Roman road, and domestic and industrial structures, including floor levels. Trenches here contained an abundance of pottery fragments, including Samian ware, mortaria, Black Burnished ware jars and flagons, a large amphora (handles and rim), a colander, and even a small oil lamp. A range of imported pottery types, mainly from the 2nd and 3rd centuries, included Severn Valley ware and a single shard of grey Crambeck ware (4th century). Lumps of unfired red, white, and yellow clay were excavated, together with what appeared to be areas of burning.

BELOW The planking in the mill race.
The planking in the mill race. Photo: Grampus Heritage.

Other domestic items recovered included fragments of hand quern-stones, a trumpet brooch, and Roman coins ranging from the 1st to the 4th centuries. Metal objects included a large amount of later coins, mostly Medieval, and other scattered artefacts such as buttons and buckles, dating to modern times. Unique amongst the finds were three stones with tantalising glimpses of possible inscriptions, and other part-carved stones.

A few enigmas presented themselves here, also: a small, circular stone structure capped by a large boulder, contained not only partly dressed stones, but also a small pile of material with the texture of pumice, and a coin at the base. Although there were abundant small fragments of burnt bone and charcoal scattered throughout the site, there were no intact burials. Several areas contained concentrated grain deposits, some with coins, which may represent ritual deposits.

ABOVE The Mill during excavation. Above RIGHT An enigmatic circular structure with inscribed stones and coin at base.
The Mill during excavation. Photo: Mick Fairfield.
An enigmatic circular structure with inscribed stones and coin at base. Photo: Grampus Heritage.

Amphitheatre, gyrus, or circus tent?

The large, 60m diameter circular structure was the subject of much speculation. Opinions as to its original function varied from amphitheatre or gyrus (a horse training arena, similar to the one at Lunt Roman Fort at Baginton, Warwickshire) to the imprint of a circus tent. Evidence for any definitive structure proved elusive. The remains of the base belonging to a substantially built wall were clear, with clay and cobble construction; but the interior appeared to have been reworked by previous flood events, and contained little dating evidence or clues as to the use of the structure.

The river crossing-point also remains an enigma, as no substantial remains of a bridge were discovered at any point. The river, however, has been mobile in the past (as it is today), and may even have been multi-channel and shallow enough to be forded in Roman times. The lack of defined features in the geophysical plot between the mill and the present river certainly suggests that this area has been reworked since Roman times, with dominant river migration towards the north.

Roman Mills

Roman Mills were first described by Vitruvius around 40 BC, with earlier mills used by the Greeks in the 3rd century BC. Mills have been recorded as having a variety of uses, from grinding grain and bread-kneading, to sawmilling and pounding ore.

The 3rd century grain mill at Haltwhistle Burn represents the best described example of a Roman mill in the North of England. The site, located on Hadrian’s Wall close to milecastle 42, was excavated in 1907-1908 by F G Simpson. The use of water for both civilian and military purposes was the main means of supplying a continuous source of power. The most common means of achieving this was via an artificial channel diverted from the main river, usually at a bend; however, there are examples where a culvert is built into a bridge, such as at Willowford – also on the Wall, near Birdoswald Roman Fort. The power derived from the flow of the race would be controlled via a sluice or diversion channel. The vertical wheel would turn by the pressure of the water at the base (undershot). There was often little gearing involved, as in more modern Medieval mills and the attached structure, such as a grindstone for example, which would turn at the same rate as the water wheel. Wheels and gears would be constructed of wood with wooden supports, as the only feasible construction, although metal and stone spindles may have been used.

Image: Mick Fairfield.

The remains of millstones at Haltwhistle suggest it was used as a grain mill; along with other sites in the north (such as Chesters and Birdoswald) it would have supplied the army with food. Other British sites have evidence of metalworking (Ickham) and may have been used in the manufacture of military equipment. The mill at Cockermouth is similar in construction to that at Haltwhistle, complete with a wood-lined mill race, but the lack of a millstone and evidence of grain leave its ultimate function elusive.

And so to the baths?

After the four intensive weeks of excavation were over, the trenches were refilled so that the field could be reinstated for arable use. A geophysical survey on the opposite bank suggests the presence of a substantial building (35m x 30m) at the base of the hill (Sibby Brow), beneath the fort – possibly a bath house? This, hopefully, will be the subject of future excavations in 2011 and 2012, funds permitting. The work of conservation and cataloguing of all the finds is ongoing.

The discovery of so much Roman activity based around the river was very surprising, and the well-preserved remains are testament to the fact not all rivers rework the archaeology that has been invested in the floodplain – a point to ponder whilst considering further sites for future archaeological investigation.

J Collingwood Bruce (1966) Handbook to the Roman Wall, Harold & Hill, Newcastle.
Guy de la Bedoyere (2001) The Buildings of Roman Britain, The History Press.
J Bernard Bradbury (1994).Cockermouth and District in Old Photographs, Alan Sutton.
Thanks to the landowners, Robert Jackson, David Robinson, and Eldred Curwen for their permission to carry out the fieldwork; Mark Graham and Joanne Wilkinson of Grampus Heritage, Frank Giecco of North Pennines Archaeology and Roger Asquith of Cockermouth for supplying some of the photographs and information.
Dr Lynda Howard