Type: Fort, Possible Minor Settlement
|Probable Tactical Road: SE (4) to Brompton (Shropshire)
WSW (13) to Caersws (Caersws, Powys)
Possible road: ENE (30) to Viroconivm
WSW (13) to Mediomanvm
Road: NE (12) to Westbvry
|SO207989||c.500 x 460 ft
(c.152 x 140 m)
Along the Severn Valley into the Welsh mountains, the Roman fort of Levobrinta stands at the conflux of the River Severn and the Afon Rhiw flowing from the hills of Dyfnant Forest. Another road, known as the Hen Ffordd or the Old Road, linked Forden Gaer with the Greensforge complex a days march to the south-east via a fort and marching camp at Craven Arms, closely following the route of an older, native trackway which linked a number of hillforts in the area. The only classical geographical reference which names the Forden Gaer fort is the Ravenna Cosmology of the seventh century, wherein the item Lavobrinta (R&C#80) is listed between the entry for Utriconion Cornoviorum (Wroxeter, Shropshire) and the unknown station Mediomanum. This permanent fort probably stands just outside the western border of the Cornovian canton (Webster fig.4 p.8), and there are a number of other military sites in the area; the fort of Brompton, Church Stoke lies 4½ miles to the south-east (SO2493) together with three marching camps, and another temporary camp lies 7 miles to the south-west across the Welsh border at Glanmiheli near Newtown in Powys (SO1590). There is also a small fortlet some 8 miles to the north-west at Llanfair Caereinion in Powys (SJ1004) (O.S. Maps).
The Forden Gaer site was occupied by a large early-period fort which was replaced by a smaller Flavian establishment, both forts being timber built. Samian ware has been recovered from the site of the Trajanic, Hadrianic and Antonine periods, which suggests continued occupation throughout this time, and Severan pottery shows that the fort was even then still considered a crucial strongpoint for the military control of mid-Wales. Destruction debris around 296AD, found in the fort and in the well of the principia may indicate a planned withdrawal of the garrison or destruction of the fort by an unknown enemy. Neither hypothesis can be proven. There are other occupation layers dated to around 343 (Period IVa) and 383 (Period IVb), separated by a burnt (destruction?) layer dated around 350-360. The latest coin from the site is that of Valens (Imp. 364-378). Measured between the ramparts the dimensions of the Flavian fort are about 500 ft from north-north-east to south-south-west by about 460 ft transversely (c.152 x 140 m), which gives an occupation area of about 5¼ acres (c.2.14 ha). There are three ditches on all sides and gateways in the centre of the NNE and SSW defences. There is evidence of a civilian settlement or vicus extending along the road leading from the fort's north-east gate, including a rectangular ditched enclosure aligned north-west to south-east and measuring 240 by 160 feet (c.73 x 49 m), an area of just under an acre (c.0.35 ha); it is possible that this feature may have been a temple precinct. The Flavian fort is certainly large enough to have housed any number of different unit types. (JRS 1953 p.85; 1973 pp.235/6; Simpson.)
Visibly, the fort is still quite impressive, and the full circuit of the defenses can still be made out in the clay soil of the Severn valley, rising to the height of about 2m above the surroundings. The NE corner, running alongside the farm track to the fields on the banks of the Severn, is covered by a small plantation of fir-trees. The field is very muddy, with a sticky, sand-coloured clay, which retains the surface water in opaque puddles. This must have been a picturesque, though incredibly dirty assignment (Pers. obs.).