First indicated on aerial photographs taken in 1979 (Britannia, 1980), a subsequent magnetometer survey undertaken by StrataScan in 2003 ( has revealed a sequence of forts, including perhaps the largest fort anywhere in Wales, also traces of an extramural civilian settlement and a possible military bath-house, all lying undisturbed beneath fields in the grounds of Dynevor Park, Llandeilo, close beside the Deserted Medieval Village (DMV) of Llandyfeisant.

The Vexillation Fortress – Dynevor A

OS National Grid Reference: SN621224
Dimensions: c.720 x 590 ft (c.220 x 180 m)
Area: c.12 acres (c.4.85 ha)

With reference to the StrataScan 2003 magnetometer survey; the original fort on the site is represented by a substantial eastern corner-angle and attached lengths of the adjacent NE and SE sides, a long length of the north-western defences (visible beneath the features of the secondary fort – Dynevor B), also the southern part of the south-western defences, the northern portion being obscured beneath the SW defences of the later fort, which lie on the same alignment. From these apparent features the primary encampment would have had an external dimension of around 590 feet (c.180 m) square, and therefore covered an area of about 8 acres (c.3.23 ha).

The defences identified on the south-west, however, appear as a pair of ditches on the GeoPhys plot, unlike those on the north-west and at the eastern corner-angle which are trivallate in nature. This leads one to conclude that the bivallate south-western ditches are perhaps associated with a contraction of the fort’s original defences which must therefore lie outside the SW edge of the 2003 survey. If this is the case, and assuming that the dimensions of the original fort were built to the standard Roman military ratio of 3:2, then the original SW-NE dimension may have been somewhere in the region of 885 feet (270m), which would give an area of around 12 acres (c.4.85 ha) before reduction of the fort to the square outline revealed on the survey.

The only feature within the original enclosure appears to be a roadway which roughly bisects the fortress along its main axis, recorded on the 1979 A.P.’s and also shown on the 2003 GeoPhys survey. Features in the southern half of the fortress would seem to be associated with the road leading from the south-east gateway of the secondary fort, Dynevor B, which was built across the primary encampments north-western defences. The 2003 plot also shows what appears to be an annexe attached to the southern part of the fortresses north-eastern defences, measuring perhaps 360 x 150 feet (c.110 x 45 m) and enclosing an area of about 1¼ acres (c.0.5 ha).

The hefty trivallate defences would have taken considerable effort and indicate that this camp was intended to be occupied for as long as it took to quell unrest in this part of Wales. The area originally enclosed would have been sufficient to hold perhaps half a legion and a unit of cavalry under field conditions, somewhere in the region of three thousand men. The lack of any features within the bounds of the camp shows that the occupation period was not long, perhaps only a couple of seasons. The annexe on the north-eastern side may indicate troop build-up in the area, possibly being occupied by a supplementary force attached to the original garrison, which was subsequently reduced by about a third before the encampment was eventually abandoned, the defences very likely being levelled at this time.

Some Speculative Thoughts on the Origins of Dynevor A

It is possible that the Dynevor Park fortress was established during the tenure of governor Quintus Veranius, who administrated Britain during A.D. 57/58 and is known to have campaigned in south-west Wales. Less likely in the author’s opinion would be a foundation date during the administration of Sextus Julius Frontinus, governor of Britain from around A.D. 73 to 77, who is also known to have campaigned in southern Wales, although his efforts seem to have been mainly directed towards the war-like Silures tribe of Monmouthshire and Glamorgan. It may well prove that the apparently short-lived occupation of the Dynevor A site is linked to the notably short tenure of Veranius, who died in unrecorded circumstances after only one campaign season.

The Auxiliary Fort – Dynevor B

OS National Grid Reference: SN622225
Dimensions: c.490 x 330 ft (c.150 x 100 m)
Area: c.3¾ acres (c.1.54 ha)

After an unknown period a smaller fort was built across the primary encampments north-western defences. This fort was protected by substantial multivallate defences with internal dimensions of about 490 feet NE-SW by 330 feet transversely (c.150 x 100m), and therefore had an internal occupation area of around 3¾ acres (c.1.54 ha). This is sufficient to have comfortably housed a Cohors quingenaria peditata or a regiment of 500 foot-soldiers, or possibly a part-mounted unit or Cohors quingenaria equitata under somewhat cramped conditions.

The 2003 GeoPhys plot clearly reveals the grid-like structure of the roadways within this fort, along with the outlines and post-holes of many of its internal buildings. There is also evidence of activity along the roads leading from this forts south-eastern and north-eastern gateways, the former possibly being military in nature, the latter probably an associated civilian roadside settlement or vicus.

More Speculation on the History of Dynevor B

It is possible that this fort was established during the tenure of governor Agricola during the first year of his administration (c. A.D. 77/78), in order to keep a watchful eye on the inhabitants of Dyfed, the Demetae, being situated as it is, close to the border between this apparently peaceful tribe and their truculent neighbours the Silures. If this proves to be the case, then it is most likely that some sort of agreement was reached between Agricola and the inhabitants of the Cardigan Peninsula which precluded the stationing of any great force in their territories, thus enabling the Roman general during the next campaign season to concentrate his forces in northern England, where it seems likely he drafted another similar agreement with the Carvetii tribe of Cumbria.

Other Roman Sites/Finds from Llandeilo

Possible Military Bath-House

Another area of interest revealed by the 2003 magnetometer survey lies in a separate field about 220 metres outside Dynevor B’s north-western defences, where a scatter of geophys readings have been tentatively interpreted as the remains of a bath-house associated with the auxiliary fort. This remains doubtful, however.

The 1891 O.S. map of the area ( indicates the site of a “Roman Temple” in the woods near St. Tyfei’s Church to the south of the fort. The author is inclined to believe that this represents a far better location for a Roman military bath-house. Perhaps the antiquaries which originally recorded this site were prompted to allocate their evidently Roman building a religious role merely because of the proximity of the old church, for certainly, to find a genuine Roman temple this far westwards would be most unusual.

The 1891 map, incidentally, also records the find-spot of an unspecified number of Roman coins in the field to the south-west of the fort around the turn of the 19th century.

RIB 2262 - Milestone of Tacitus

For the Emperor Caesar Marcus Claudius Tacitus Pius Felix Invictus Augustus.


Tacitus, A.D. 275-6.Its original position is unknown, but was presumably in Carmarthenshire.

A Roman milestone or honorific pillar was recorded in use as the cornerstone of a small farm building on the estate of Dynevor Castle in 1697, but this building was demolished and all its useable stone subsequently reclaimed for use in construction of the Dynvor Park boundary wall during ’emparkment’ in the late 17th century; the text of the stone, however, has survived.


“For Imperator Caesar Marcus Claudius Tacitus¹ Pius Felix Invictus Augustus.”

(RIB 2262; milestone; found 1697; now lost; dated 275/276AD)

  1. The emperor Tacitus donned the purple aged 75, having only just served as consul, following the death of emperor Aurelian on campaign in Persia in late-275AD. He immediately left Rome and advanced into Thrace and Asia Minor, earning himself the title Gothicus Maximus for his exploits. He was murdered at Tyana in Cappadocia after ruling for only 6 months.

Another stone from Llandeilo bearing Latin script was also reported by the antiquarian Edward Lluyd in 1697 lying “by ye ch’yard of this town”, and inscribed …IACET CVRCAGNVS …VRIVI FILIVS, which seems to be some sort of grave marker, but its non-Roman names suggest that the stone should be dated to sometime after the Romans had left the British Isles. This stone, like the Tacitus milestone, has since been lost and is thought to have been broken up shortly after discovery.

Dynevor Park Excavations 2005

The site was excavated between June 27th and July 18th 2005 in a joint venture run by the National Trust (who provided the land) and Cambria Archaeology (who provided the brains), assisted by a number of volunteers (who provided the brawn), yours truly included. The proceedings were featured on the British Channel-4 Television series Time Team, as part of their much publicised “Big Roman Dig”.

A series of trenches were dug at a number of strategic locations in order to clarify the evidence plotted on the Stratascan magnetometric geophysical survey, and to establish a dating sequence; eight trenches were planned but only seven were actually opened.

  1. Across the south-western defences of the vexillation fortress close by the central gateway. [Nikki Cook]
  2. In the central area (latera praetorii) of the auxiliary fort on the top of the hill. [Richard Ramsey]
  3. Within the eastern quadrant of the fortress to the east of the road leading from the south-east gateway of the fort. [Richard Jones]
  4. The area of the civilian settlement (vicus) outside the main (north-east) gateway of the fort. [Ken Murphy]
  5. A southward extension of the above trench.
  6. Across the apsidal building plotted in the field (“Cae Brics”) to the north of the military camps.
  7. Across the roadside ditches east of the civilian settlement.
  8. Alongside the Dynevor Park boundary wall beside the modern road. [NOT EXCAVATED!]

“Dig Diary” from Archaeoleg Cambria

Trench 1 – The South-West Gateway of the Fortress

This trench was positioned across the south-western defences of the Vexillation Fortress and located the rampart and two defensive ditches on this side, both with square-cut terminals, along with the post-holes of a gateway tower and the remains of a metalled road surface. The interior ditch was investigated by means of a “box-section” which revealed a typical Roman military V-shaped profile with a square cleaning slot in the bottom; the lower fill of this ditch was composed of a quantity of burnt grain which was bagged and sent off for environmental processing in the Lab. The outer ditch bore the same military profile but with lesser dimensions. Lying in a hollow just outside this ditch were found the broken remains of a large amphora, originally containing olive oil from southern Spain, which was possibly used by the gateway guard detail as a “pissoir”, the urine being collected for use in the processing of leather.

The Fortress Gateway Viewed from the Observation Platform Looking North-East
The Primary (Inner) Ditch of the Fortress Looking North-West

The rampart had been strengthened by an internal revetment of upright posts revealed by a series of small post-holes set roughly along its central axis. The gateway was represented by four substantial post-holes set in a pattern which suggested that the original arrangement was of two six-post towers set to either side of the metalled road which ran along the primary axis of the fortress. The width of the rampart remains suggest a height of around 10 feet (3m) for the level of the rampart-walk, which would have been fronted by a timber palisade, no trace of which survived, while the substantial gateway towers would have extended to perhaps 30 feet (9m) in height.

Trench 2 – The Central Area (Latera Praetorii) of the Auxiliary Fort

The Auxiliary Fort Viewed from the Observation Platform Looking South-East
Emily Excavates one of the Postholes in the Principia Looking North-East

This ditch lay at an angle slightly inclined from the roadway and ditches of the overlying layer, and had a typical Roman military V-shaped profile with a square-cut drainage slot in the bottom, measuring 8 feet across by about 4 feet deep (c. 2.4 x 1.2 metres). Excavation revealed that this ditch had been allowed to silt-up naturally over a substantial period of time before being deliberately backfilled with material. This suggests that the original fortress was abandoned, perhaps in a hurry, with no attempt being made to level the area until the site was re-used some years later.

Open Day Saturday 16th July Gwilym Conducts Another Tour Group While Emily and Andy Discuss Trowel Strategy Looking towards the South

Trench 3 – The Interior (Retentura) of the Fortress

Trenches 4 & 5 – The Civilian Settlement (Vicus) Outside the Fort

Trench 6 – The Suspected Military Bath-House

The Old Church at Llandyfeisant South of the Roman Encampments Possibly Occupies the Site of the Roman Military Bath House

The 2003 GeoPhys survey showed an apsidal plot in the field to the immediate north of the Roman forts which was thought to represent a Roman military bath-house, perhaps fed by one of the numerous springs which exist in this area. Upon excavation, however, the structure revealed on the magnetometric plot turned out to be a “post-medieval brick-kiln”, and was not associated with the Roman forts on the top of the hill. In hindsight, the name of the local area, “Cae Brics” or “Brick Field”, gave some clue as to the nature of the building uncovered here.

Trench 7 – The Ditches of the Roman Road to Llandovery

This trench was opened in the first week of the dig and soon struck bedrock only a foot beneath the modern ground surface. However, two rock-cut ditches with V-shaped outlines spaced several metres apart were uncovered. These represented the roadside ditches to either side of the main Roman road to/from the known Roman military establishment at Llandovery, some thirteeen miles to the north-east.

Trench 8 – Across the Line of the Llandovery Road

This trench, although planned and laid-out, was not excavated due to lack of resources. However, a casual surface survey revealed a number of modern finds including a set of coins (penny, halfpenny and farthing) issued during the reign of King George V.

Yours Truly Assisting in the Site Survey Using a Trimble 5600 Total Station Looking north; the break in ground level in the near background marks the line of the auxiliary fort’s north-western rampart, the NE rampart lies beyond the trees on the right.

Books and Articles in Togodumnus’ Library

Online Reference Material

Map References for Dynevor Park/dinefwr Parc, Llandeilo

NGRef: SN6222 OSMap: LR159

Roman Roads near Dynevor Park/dinefwr Parc, Llandeilo

W (13.5) to Moridvnvm NE (13.5) to Llandovery (Llandovery, Dyfed)

Sites near Dynevor Park Roman Fort