The Conquest of the Silures: A Scenario of the Middle Usk Valley Seamus Hamill-Keays MA No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any other information storage and retrieval system, without prior written permission from the author. © SH-K 2011 2 The Conquest of the Silures: A Scenario of the Middle Usk Valley Our knowledge of the history of peoples who lived in Wales in the middle of the 1st Century AD, is almost entirely due to the Roman historian Tacitus. He recorded the military successes and failures of a succession of Roman commanders in Britain in the years AD 43 to AD 76 against peoples he identified as the Decangi or Deceangli, the Ordivices and the Silures. The boundaries of their territories is a matter of scholarly debate. Tacitus’s relation of the campaigns in what would centuries later become Wales are geographically vague. Ptolemy identifies Bullaeum/Burrium, i.e. Usk, as a town of the Silures and so places their territory east of the territorial town of the Demetae, Maridunum i.e. Carmarthen (Geographia: 2.3.12) To the east of Siluria lies the territory of the Dobuni with their town of Corinium i.e. Cirencester. It can be deduced from other place-names in Ptolemy that the territory of the Ordovices lay to the north of Siluria. It can be further deduced that there were other peoples in Wales and the Marches not mentioned by Tacitus (Davies & Jones 204: 46). Much of the debate about Siluria is directed as to where the eastern boundary of Silurian tribal territory lay. Webster (1981: 17) considers that their territory was bounded to the north and east by the Wye. Salway (1985: 45) comments that tribal boundaries in Wales are even more uncertain than in England and identifies the Glamorgans, Gwent and southern Powys as the land of the Silures. Jones and Mattingly (1990: 61) suggest that it may have been east of the Wye towards the Severn before the Roman advance, citing signs of conflict at the Sutton Wall hill-fort. The OS Map of Roman Britain 5th Edition (2001) has the legend ‘SILVRES’ stretching across the Glamorgan valleys from the Neath to the Usk suggestive of Silurian territory lying between these two rivers. After the total conquest of the Silures, c.AD 75, the Romans established Venta Silurum i.e. Caerwent, as the market town of the Silures, close to the Severn estuary, east of the Usk and west of the Wye. Wherever the boundaries were, it is generally agreed that the river valley of the Usk was an important part of Siluria. It was Publius Ostorius Scapula who, in dealing with internal disruptions and thus advancing the frontier of the Roman province to the Trent and the Severn in AD 47, established a series of encampments to protect the expanded province (Annals 12.31). When he was bringing the Deceangli in north-east Wales to heel, he had to march against, subjugate and disarm the Brigantes in northern England. However it was the Silures were the most obdurate in resisting the power of Rome and who continued to threaten the Romans and their allies across the Severn. According to Tacitus, the situation was such that it was necessary to establish legionary fortresses, castra legionum, as bases from which to attack and suppress this recalcitrant people (Annales 12.32). The creation of a new colony of veterans at Colchester released the garrison there, Legio XX, for service against the Silures (Webster 1981: 24, Manning 2004: 62). These legionary fortresses may include the vexillation fortress at Clyro in the middle Wye valley, large enough for half a legion, and one at Kingsholm on the left bank of the lower Severn near Gloucester; opinions differ on the locations © SH-K 2011 3 (Webster 1981: 42, Salway 1985: 105, Manning 2004: 63, Davies & Jones 2006: 47)1. Manning observes that if the one fortress was Kingsholm, the border of Siluria was the Severn but suggests Weston-under-Penyard on the left bank of the Wye as its more probable location. Davies and Jones (2004: 49) give the generally accepted date of this event as AD 49. The Silurian resistance to the Romans was led by Caratacus who had been fighting the invaders of Britain since AD 43. He had been pushed relentlessly westward and by AD 50 was encouraging the Silures to attack peoples who had agreed to collaborate. However, according to Tacitus, he decided that it would be strategically advantageous to move his theatre of operations to the land of the Ordovices. There he was defeated and in AD 51 was later betrayed into Roman hands by the queen of the Brigantes. Taken to Rome he was famously pardoned by Emperor Claudius and so vanishes from history (Annals 12.37). The route of the Scapulan attack on the Silures during the three campaigning seasons AD 49 to AD 51 is unknown. However the marching camp at Clifford, and the later campaign base at Clyro (Davies & Jones 2006: 63), together with the support structure evidenced at Leighton (St Joseph 1973: 145) originally aimed at the Decangli, and at Leintwardine (St Joseph 1961: 124) may indicate the direction of the main Scapulan advance into Siluria, crossing the low watershed between the Wye and Usk valleys and interdicting cooperation between the Silures and the Ordovices thus forcing Caratacus to shift northward. On the other hand, Manning (2004: 64) focuses on the main route from Gloucester to Weston-under-Penyard and then southwards in the direction of the Usk valley to a fort at Monmouth; it is not certain if there was a fort at Usk predating the construction of the Neronian fortress (see below). A branch off this axis led through the Golden Valley and thence to a marching camp at Clifford on the right bank of the Wye and the later campaign base at Clyro on the left bank. Manning also draws attention to an alternative route from the Severn valley to the Wye valley north-west from Gloucester to Stretton Grandison twelve miles east of Hereford. A Roman road connects this place to Clyro, with the settlement at Magnis i.e. Kentchester, mid-way between the two. It could well have been a combined operation that penetrated Siluran territory from both main routes. However there seems to be no evidence that at this time, c.AD 51, the Romans had penetrated the middle Usk valley from either north or south. Clyro, blocking the middle Wye valley, attracts many scholars as a place pivotal in the Roman conquest of the Silures. After the capture of Caratacus, the Silures continued a guerrilla war, attacking legionary cohorts who had been directed to build forts in their territory (Annals 12.38). In one instance, a camp-prefect, eight centurions and many legionnaires were slain; losses even more severe would have been incurred if neighbouring forts had not quickly sent help. In another action, cavalry squadrons on a foraging expedition were forced to flee. This action escalated; Tacitus tells us Ostorious then threw his lightinfantry into the battle but they were unable to stabilise the situation. Only when the legionary troops joined in were the Silures forced to flee, but with little loss. 1 Manning (2004: 63) translates castrisque legionum as ‘and a legionary fortress’; arguably this would be castrisque legionis. The problem here is that castra is a plural noun used for a single object. See also Manning (2004 : 35 ) for a discussion. © SH-K 2011 4 Skirmishes continued, in forests and morasses, with the Silures incensed at the report that it was now Roman policy for them to be transplanted, maybe to Gaul, or even to be utterly destroyed. Their anger was taken out on two auxiliary cohorts that were ambushed when on a raid for booty; some of the survivors were given to or sold as slaves to other tribes. It was with glee that the Silures learnt of the death of Ostorious, worn out by the unrelenting stress of this guerrilla war (Annals 12.39). They had a further notable success before the arrival of the new governor by defeating the legion commanded by Manlius Valens. They continued to rampage widely. The new governor, Aulus Didius Gallus, on his expedited arrival, quickly regained control. Unfortunately Tacitus does not tell us a great deal more about his campaign against the Silures in his Annals and treats the period of Didius’s governorship, AD 52-57, as a period of masterly inactivity on his part. In the Agricola he does credit him with consolidating what had been conquered in Britain and advancing the frontier line by the construction of a few forts ...paucis...castellis (Agricolae 14). An eyebrow can be raised here at Tacitus’s disingenuousness because there now seems little doubt that a new legionary fortress was constructed at Usk during Didius’s tour of duty. Manning, despite detailed consideration of all the evidence, has not given an exact date for the foundation of the fortress but believes its construction to have begun in the early years of the reign of Nero (AD 54 – 69) and certainly before AD 60. He further suggests that its construction was part of a reorganisation of legionary dispositions at Usk, Wroxeter and Exeter (Manning 1981: 32-34). Davies and Jones postulate that operations directed against the Silures at this time were more concerned with consolidation and containment than with penetration (Davies & Jones 2006: 51). The attitude of the new emperor had on military operations can be surmised through Gaius Suetonius Tranquillus’s report that he was considered withdrawing but not wanting to diminish the glory obtained by Claudius in AD 43, kept the legions there (Lives: Nero 18). The fortress at Usk was strategically placed in a region where the many Iron Age hill-forts are indicative of a centre of population. Situated on the left bank of the Usk at the confluence of that river and the Olway Brook, a stream that issues from the valley route to Monmouth, the fortress dominated movement along the coastal plain and the Usk valley itself. Manning sees it as part of a pre-Flavian ‘carefully constructed and ingenious system intended to provide the answer to the problem of Silurian raids while allowing for future advance’ (Manning 1981: 43). The first Flavian emperor was Vespasian (AD 69 – 79). Forts established in and around Siliuria before his victory in the Year of the Four Emperors, AD 69, and part of the system described by Manning included Cardiff, Walton, Jay Lane, Clyro and an anomalous fort in the Twyi valley at Llandovery.2 A fort that blocked the gateway to the valleys of the Brecon Beacons and the Black Mountains was established at Abergavenny. It seems reasonable to assume that the Usk valley to the north-west of this fort was still controlled by the Silures in this period. The governorship of Didius came to an end in AD 57. He was replaced by Quintus Veranius, an officer who had served with distinction elsewhere in the empire (Annals 3.19, 12.5). Tacitus describes his efforts against the Silures as ‘trifling raids’ 2 Excavation of a Flavian fort at Llandovery revealed the remains of an earlier defence tentatively identified as pre-Flavian (Nash-Williams & Jarrett 1969: 95). © SH-K 2011 5 (Annals 14.29). Veranius did not have a real opportunity to show what he could do because he died the next year. Manning considers that, by this time, the basic preparations had been made for further conquests by the construction of the preFlavian fort system (Manning 1981: 31). An obvious route was up the valley of the Usk using the fortress at Usk as a logistical base; Manning’s excavation there revealed many more granaries than normal in a legionary fortress (Manning 1981: 177). The Wye/Usk watershed from Clyro did present an easier entry into the middle Usk valley but one that was much further away from the Usk fortress. However events occurred that delayed for a decade any projected further conquest of Siluria. A new and accomplished soldier, Suetonius Paullinus, was the next governor of Britain arriving in AD 58. He had conducted a successful campaign against the mountain tribes in the High Atlas in Morocco and was one of the leading generals of Nero’s time. Upon his arrival in Britain it can be surmised that he was content with the containment policy directed against the Silures for the time being for he commenced operations in north Wales with the ultimate objective of conquering Angelsey and destroying the Druidic culture that was based in that island. (Dudley & Webster 1962: 52). Tacitus describes in graphic terms how Suetonius’s soldiers stormed the foreshore of the island, AD 60. The sacred groves of the Druids were destroyed (Annals 14.30). Grave news then arrived. The Iceni, led by Boudicca, and joined by other tribes, had risen in bloody rebellion. Colchester, had been sacked by an immense host and its population put to the sword. The infantry of Legio IX had been annihilated as it came to the rescue (Annals 14.32). Suetonius hurried to London, seeking to concentrate all forces under his command. He took the hard but strategically correct decision to abandon London to the rebels who then sacked the city amid great slaughter and terrible atrocities. The sack of St. Albans followed. Suetonius had summoned part of Legio II Augustus, probably the garrison at Usk, to join him as he prepared to confront the forces of Boudicca that greatly outnumbered his ten thousand men that included Legio XIV and Legio XX. The officer in charge at that time, praefectus castrorum Poenius Postumus, disobeyed. Suetonius obtained a smashing victory with little loss. According to Tacitus, Boudicca committed suicide by poison (Annals 14.37). Tacitus gives no information on why Poenius Postumus did not march towards the forthcoming battle. There are no reports of trouble with the Silures at this time but that might be entirely due to the decision of this professional soldier to maintain the policy of containment with the forces in and around Usk. Distraught at having denied his troops the glory of participating in a great victory, he threw himself on his sword (Annals 14.37). The troops who had been under his command then marched to join Suetonius. Suetonius received reinforcements from Germany and continued to restore order by ravaging igni atque ferro, with fire and sword, until at least the spring of AD 61 (Dudley & Webster 1962: 78). He was joined by a new civilian administrator, procurator Julius Classicanius, who disagreed with this ruthless approach. Political machinations now intervened and a pretext was found to recall Suetonius to Rome before his period as governor had expired, probably in the winter of AD 61 (Dudley & Webster 1962: 83). He was replaced by Petronius Turpilianus. During the aftermath © SH-K 2011 6 of the rebellion, as indeed during it, there were no reports of trouble with the Silures; it was seemingly all quite on the western front (Davies & Jones 2006: 53). The next period of eight to ten years was a vital one for the history of Roman Britain (Salway 1981: 125). Salway attributes economic and social recovery of the province from war and famine to the policy of winning over the general support of the native aristocracy. Tacitus refers to the way in which the next governor, Trebellius Maximus, never ventured on a campaign and through courteous behaviour encouraged the barbarians to indulge themselves in seductive vices in order to appreciate the pleasures of peace and civilisation (Agricola 16). However it seems that in Wales the status quo was maintained with the Silures left alone in their valleys and mountains west of the middle and upper Usk. Nero committed suicide in AD 68 after a life of debauchery, depravity and perversions. There was no clear successor and the next year became known as the Year of the Four Emperors as powerful men fought a civil war for the purple when it was realised that emperors need not been born to it, but could use the power of the legions under their command to gain it. Remarkably the civil war did not spread to Britain and the legions that remained there, although possibly favouring Vitellius, did not participate in it (Histories 59, 60).3 On the other hand, there was an acrimony between the commander of the Legio XX Valeria Victrix, Roscius Cælius and the governor Trebellius, so much so that the governor had to flee to join Vitellius in his march on Rome. The acrimony many have been because of the restraining influence of the governor on the inclination of Roscius to despoil the natives (Salway 1981: 129). The headquarters of this legion was at Wroxeter (Nash-Williams 1969: 13) and units of its battle group were in a position to penetrate Wales and attack amongst others, the Silures. A new governor, Vettius Bolanus, arrived in AD 69 and was joined by Julius Agricola who had been given command of the troublesome Legio XX Valeria Victrix by Licinius Mucianus, a major supporter of the Flavian bid for power (Agricola 7).4 Agricola was no stranger to Britain, having served under Suetonius Paulinus during the Boudiccan revolt. He brought the legion under proper discipline once more. The accession of Vespasian brought back the previous commander of Legio IX that had been severely mauled during the Boudiccan revolt, Petillius Cerialis, as governor of Britain (Agricola 17) A dynamic commander and a survivor of military disasters, he was a clear sign that there would be a new forward policy in the province (Salway 1981: 134). He had Tacitus’s father-in-law, Julius Agricola, under his command as the legate of Legio XX Valeria Victrix (Agricola 7). It was the Brigantes in the north who felt the onslaught of the new policy of subjugating the whole island; a new legionary fortress was established at York. At the end of his tour of duty Cerialis was replaced by Sextus Julius Frontinus possibly as early as AD 74 (Jarrett 1965: 35). The Silures were about to experience the forward policy; it was time for a last stand. 3 Legio II Augusta at Usk had previously been commanded in Britain by Vespasian AD 43, and although supplied units to Vitellius might have favoured its old commander (Salway 1981: 132). 4 The honourific Valeria Victrix was awarded after the defeat of Boudicca, likewise Martia Victrix for Legio XIV Gemina (Nash-Williams 1969: 13). © SH-K 2011 7 A conglomeration of hill-forts was situated along the Usk valley on the escarpment of the Beacons and Black Mountains and the uplands of Myndd Epynt (Manning 1981: Fig. 9). A large and complex multivallate earthwork, Castell Dinas, is on a prominence above the saddle between the valley of Afon Llynfi and the Rhiangoll valley (RCAHMW 1986: 96). This fortification at one time controlled access to the latter valley. It has a southern counterpart in the univallate fort at Myarth high above the left bank of the Usk near Cwm Du, the largest in Breconshire (RCAHMW 1986: 83). North of the close narrowing of the Usk valley between a spur of Myndd Llangors and Tor-y-Foel, another very large multivallate fort surmounts the long crest of Allt-yr-Esgair (RCAHMW 1986: 91) dominating the Usk valley to the west and the valley of the Llynfi and Llangorse Lake to the east. Further up the Usk valley another large multivallate fort crowns the summit of Pen-y-Crug, 2km northwest of the confluence of the Usk and the Honddu (RCAHMW 1986: 68). This fort was visible from the summit of Allt-yr-Esgair. Equidistant from Pen-y-Crug and Alltyr-Esgair is another large univallate fort, Hillis Fort near the Alltfilo, that commanded the watershed between the Wye and the Usk (RCAHMW 1986: 104). Many of these forts were visible from one other and together with the restricted topography, might have presented Roman forces with difficulties if re-occupied by the Silures in defence of their remaining territory. The population of this area could put themselves into a significant defensive posture when awaiting a further Roman advance either up the valley of the Usk or over the watershed, or both. Jarrett estimates the maximum population of the Silures c.AD 60 as 40,000 with a fighting strength of less that half that (Jarrett 1965: 36). Tacitus’s does honour to both sides in his short description of Frontinus’s campaign that may have spanned AD 74 to AD 77, three, possibly four campaigning seasons (Jarrett 1965: 37). He wrote: ‘He conquered the strong and militant tribe of the Silures, triumphing over the valour of the enemy and the difficulties of the terrain’ (Agricola 17). Traces of marching camps that were routinely constructed at the end of a day’s movement on campaign by legions and auxiliaries might give an indication of the route of a campaign by indicating key assembly grounds (Webster 1981: 41). The dating evidence that is found in permanent fortifications, particularly ceramic sherds, is absent from marching camps because pottery was not normally part of a soldier’s equipment on campaign; the evidence is fragmentary and piecemeal. The marching camps that have been found so far in Silurian territory to the west of the Usk vary greatly in size.5 They are numbered in Figure 1: 1. The camp at Blaen Cwm Bach of 26.69ha (66.19 acres) that lies on a ridge 5km ENE from the fort at Neath (RCAHMW 1976: I.2, 99 no.738). Davies and Jones view it as probably indicating a force moving westward through the Vale of Glamorgan then changing direction to ascend the eastern flank of the Neath valley or taking to the high ground before attempting the crossing of the River Neath in a westerly direction (Davies & Jones 2006: 59). 5 The density of men/acre of Flavian forts has been computed by Maxwell, cited in by Davies and Jones, as lying between 194 and 255 men/acre i.e. 480 and 630 men/ha (Davies and Jones 2006: 39). © SH-K 2011 8 2. A small 4.7ha camp, the alignment of which Davies and Jones hold is suggestive of a force moving north-west towards the middle Vale of Neath, is Carn Caca at Melin Court on the south side of the Vale (Davies & Jones 2006: 112). 3. A marching camp of about 14ha (34.5 acres) is situated close to the later auxiliary fort at Coelbren at the confluence of the River Pyrddin and the Camnant stream. It is also close to the alignment of Sarn Helen. It has excellent views to the north-west (Davies & Jones 2006: 114) 4. Another site that lies on the Neath valley axis is the Plas-y-Gors marching camp of 8.7ha (21 acres) that lies east-facing on a slope north of Ystradfellte, adjacent to the presumed route of Sarn Helen (Davies & Jones 2006: 91).6 5. A large marching camp of 17.8ha (44 acres) is located at Arosfa Garreg north of the Black Mountain area. It commands good views to the west and also to the north down the ravine of the River Clydach. The camp is aligned WNW-ESE (Davies & Jones 2006: 101). 6. Two superimposed marching camps are located at Y Pigwn on Trecastle mountain with commanding views in all directions. The larger, older camp of 15ha is aligned NE-SW. The smaller of 10ha is aligned ENE-WSW. 5.5km to the south of these camps is the large Arosfa Garreg camp of 17.8ha with a possible alignment WNW (David & Jones: 103). 7. On a ridge to the west of the Taff Vale is the 15.3ha (38 acres) camp at Peny-Coedcae, south of Pontypridd, carefully sited with excellent views to the north and south-east; its alignment may be to the east (Davies & Jones 2006: 15, 116). 1. Between the Rhondda Fach and Cynon valleys, above Ferndale, with commanding views to the north and east is an irregularly-shaped fort, probably designed to take advantage of the ground. Twyn-y-Briddallt is about 6.5ha (16 acres) (Davies & Jones 2006: 118). The locations of the marching camps above are reasonable evidence of routes but only those penetrating Silurian territory up the Vale of Neath, Taff Vale and the valley of the Cynon. The large camp at Blaen-y-Bach is unusual not only because of it size but also its shape A typical ratio of length to breadth for a Roman camp of classic ‘playing card’ shape with four gates is 3:2; this camp has a ratio of 3:1. Vegetius Renatus, specifying camp construction and writing in the late fourth or early fifth century AD, but drawing on many earlier sources states that: 6 It was standard practice for Roman camps to be aligned in a specific direction with the front gate, porta praetoria, facing the enemy and the intended direction of march (Pseudo-Hyginus 56). Provided the front and rear gates are distinguishable one from the other, the orientation of the camp can be determined. © SH-K 2011 9 Figure 1. Marching camps in Siluria c. 75 AD (after Davies & Jones 2006) The dimensions must be exactly computed by the engineers so that the size of the camp may be proportioned to the numbers of troops. A camp that is too confined will not permit the troops to perform their movements with freedom, and one that is too extensive divides them too much (Epitoma rei militaris 3.8). Jarrett (1965: 35) estimated Blaen-y-Bach as being of sufficient size to accommodate three legions plus auxiliaries. Davies and Jones, citing Maxwell, give a range between 13,280 and 16,880 troops (Davies & Jones 2006: 42).7 Even taking the lower figure, this represent the activities of one of the largest Roman armies assembled on the Welsh front (Davies & Jones 2006: 59). The unusual shape of the camp at Blaen-y-Bach of a 3:1 profile, is certainly due to its position on a ridge. It may be that the topography prevented the construction of a cluster of separate forts in a ‘gathering ground’ in the style of the five camps in the Walford Basin (Davies & Jones 2006: 35) and so it was then necessary to depart from the classic 3:2 profile. The likely disposition of legions in Britain on Frontius’s arrival, c.AD 74, was Legio II Augusta with headquarters at Usk but shortly to be involved in constructing a new fortress at Caerleon, Legio XX Valeria Victrix with headquarters at Wroxeter, Legio II Adiutrix was probably at the site of another new fortress planned or under construction at Chester (Jarrett 1965: 35) and Legio IX Hispana at York. The number of auxiliary units is unknown; there was no fixed ratio of legionary troops to auxiliaries (Goldsworthy 2003: 57). Vegetius, writing about campaigns in the days of the Roman Republic, is unequivocal that ‘ it was an invariable rule ... that the number of allies or auxiliaries should never exceed that of the Roman citizens’ (Epitoma Re 7 This a close to the size of the force consisting of three legions, six cohorts of auxiliary infantry and three alae of cavalry under Quinctilius Varus, destroyed in the Teutoburger Wald in AD 4 by renegade German auxiliaries (Keppie 1984: 168). © SH-K 2011 10 Militari 3.1). Gilliver contends that Vegetius was expressing an ideal to enhance the prowess of the citizen soldiers. She calls attention to the claim by Velleius Paterculus that there were two socii for every citizen soldier in the first century BC but remarks that ‘the histories of the early Imperial period show little concern for this issue’(Gilliver 2001: 28). The strategy adopted to finally conquer the Silures is a matter of conjecture but it seems reasonable to suppose that there were two task forces, one providing the hammer that, striking up the Vale of Neath, crushed the Silures against the anvil provided by a task force in the middle and upper Usk valley. The make-up of the armies required to carry out this supposed strategy is itself speculative. Inclusion of a vexillation of Legio II Augusta from Usk in one task force is probable together with its associated auxiliaries. Legio II Adiutrix may have been brought from Chester with its auxiliaries. Some of the ground enclosed by the rampart at Blaen-Cwm-Bach may have been too steep or marshy for use by troops. If this was the case, then the task force numbers occupying the fort may be seen to have been legionary troops with auxiliary infantry and cavalry totalling in the region of 14,000 men. The second task force, possibly operating in the middle Usk valley, might have been provided by Legio XX Valeria Victrix and its auxiliaries, operating out of Wroxeter and very familiar with this ground. The army commander, Frontinus, in his Strategemata, whilst listing successful battle strategies that led to a successful outcomes for many ancient generals, regrettably does not include any of those that led to his own success. However it can be concluded that he was a wily soldier who would be aware of opportunities to crush his enemies. In the case of the Silures in their mountain stronghold, the speculative hammer against the anvil scenario is apposite. Turning now to the middle Usk valley, the words of Davies and Jones on the lack of discovered marching camps have a resonance: ...true of the Usk valley in general, with not a single temporary camp being known until Arosfa Garreg and Y Pigwn are reached on the very uppermost margins of the same. Its flanks from the vicinity of the Neronian legionary fortress at Usk to the W. of Brecon Gaer must hold clues as to links between the above-named camps and the starting points for these campaigns somewhere to the E (Davies & Jones 2006: 64).8 In a search for clues as to the posture of a task force in the middle Usk valley it seems reasonable that the hammer advancing from Blaen Cwm Bach in the vale of Neath would seek to drive any of the Silures who resisted, against a military anvil positioned in the flat open ground of the Usk valley where legionary power would be invincible if the enemy was foolish enough to enter a pitched battle. Such a deployment would necessitate a campaign base, not unlike the pre-Flavian fort at Clyro, and logistically supported by the fortress at Usk. Clues that are popular in seeking to determine the routes of the Roman advance in Siluria are the Flavian forts built after the conquest of this territory, typically to 8 This author’s italics. © SH-K 2011 11 accommodate a cavalry cohort of approximately 500 men. As previously mentioned, Webster considers that although campaign camps, i.e. marching camps, should provide valuable evidence of the direction and scale of campaigns ‘in practice they are of little help’ (Webster 1981: 41). Associated with forts in general are the roads that they were designed to guard and maintain. In the case of Siluria three post-conquest routes can be identified each with a line of forts, some pre-Flavian (Jarrett 1969: Fig. 4): 1. Usk fortress – Abergavenny – Pen-y-Gaer – Brecon Gaer. 2. Cardiff – Caerphilly – Gelligaer – Penydarren – Brecon Gaer. 3. Neath – Coelbren – Brecon Gaer. The importance of the Usk valley in the Brecon area is emphasied by these routes, and others from the west, north and east, yet no marching camp or campaign base has yet been identified here that might have been used in Frontinus’s conquest of the Silures. On the assumption that any such fortification would be close to a road later constructed by the Roman army, it is necessary to consider the modern interpretations of Roman routes in detail in this area to seek the clues that Davies and Jones assert must exist. An antiquarian who showed great interest in Roman roads was Richard Fenton who in his Tours in Wales 1804 – 1813 relates how he often went out of his way to view reported remains of such. He visited Brecon Gaer and noted ‘the Roman road from Gobannium’. As he travelled to Crickhowell he much admired the vale of Usk and wrote ‘A very peculiar feature of it is the endless openings into Smaller Vallies on each side’, a feature that daresay was not lost on Frontinus (Fenton 1917: 24) . Theophilus Jones, in his The History of the County of Brecknock originally published in 1805, described a Roman road from Caerphilly descending into the Usk valley along the length of the parish of Llandetty, crossing the Caerfanell and reaching the Usk at Llansantffraed (Jones 1898: 428). Thomas Codrington in his Roman Roads in Britain, published in 1903, noted possible traces of a Roman road from Clyro to Brecon. He wrote further: This road was joined near Brecon by a Roman road from Abergavenny, of which however there is little trace. To the west of Crickhowell the course seems to be by Tretower to Pen-y-gaer, where there are the remains of a Roman camp, and along a lane to Ty-maur, about half-a-mile west of which at Bwlch a parish boundary joins the present main road and follows it for half-a-mile, and continues along a lane for a mile, to the southward of Allt-yr-Yscrin. A stone pillar, probably a milliary, dedicated to Victorinus (A.D. 265-7) formerly tood by the side of the road at Scethrog, and seems to show that the present road follows the course of the Roman road to the west of the Brecon and Merthyr Railway (Codrington 1903: Chap. 10).9 9 The stone has been dated by V. E Nash-Williams to the 6th century AD. It is in fact dedicated to Nemnius, son of a Victorinus (Nash-Williams 1950: Pl.VII & Fig. 54) and was erected next to Cwm Gelanedd (Vale of Corpses). © SH-K 2011 12 Codrington is much more definite in describing the Roman roads leading from Brecon Gaer to Neath i.e. Sarn Helen and the road to Llandovery passing close to the marching camps at Y Pigwn but omits any mention of a road from Brecon Gaer to Penydarren, Gelligaer, Caerphilly and Cardiff. Unlike Codrington, O’Dwyer seems to have seen Roman roads in most valleys of what might be considered to be Siluria (O’Dwyer 1937). A reviewer of his work comments that ‘in his determination to fill his maps with Roman roads has led him to use all matters of facts and fancies without discrimination’(P.C. 1939: 57). O’Dwyer did nominate a road from Cardiff to Talybont-on-Usk from the fort at Penydarren that went along the Cefn Ystrad ridge north of Dowlais to a junction with a road from Doly-Gaer to Pen Rhiw Calch and Bryn Melyn, then along the western flank of Tor-yFoel to pass through Maes Mawr Farm to a crossing of the Usk at Llansantfraed. O’Dwyer is puzzled by an OS map of his time; it depicts a Roman road going right over the hump of Allt-yr-Esgair and not taking the easier route closer to the river to join his postulated route through Maes Mawr Farm, at Llansantffraed. This low level route below Allt-yr-Esgair is undoubtedly the one described by Codrington lying to the west of the railway during his time. The OS Map of Roman Britain 3rd Edition (1956) shows a Roman road from Cardiff branching at Dol-y-Gaer, one branch crossing the Beacons through Bwlch ar y Fan, ‘The Gap’, and along the eastern flank of Bryn Teg to cross the Usk at Llanfaes. The other branch is essentially O’Dwyer’s route through Maes Mawr Farm to Llansantffraed from Pen Bwlch Glascwm. This OS map does not show this road joining the road from Abergavenny to Brecon Gaer; that is shown as going along the ridge of Allt-yr-Esgair. Margary’s exhaustive examination of the evidence for Roman roads in Britain created a useful numbering system for the roads (Margary 1967). He identifies the road in the OS Map of Roman Britain 3rd Edition from Cardiff to Brecon Gaer as RR621. The Talybont-on-Usk branch is numbered RR620 but the route he chooses is past Maes Mawr Farm through Pencelli and past Llanfrynach to cross the Usk presumably at the ford near Llanhamlach (Margary 1967: 336). He supposes that RR621 is an alternative route to Brecon Gaer that avoids ‘very high ground’. The height of Bwlch ar y Fan on RR620 is 604m; the highest point on RR621 is 510m but on the other hand this route has a very steep ascent from Maes Mawr Farm to Bryn Melyn. RR621 has easier gradients in both directions that would have been much better for horse, mule or oxen-drawn transport. Margary does not specifically describe his RR63a as cresting the ridge of Allt-yr-Esgair but refers to it as ‘being along an elevation’ (Margary 1967: RCAHMW considers O’Dwyer and Margary in respect of the road connecting Penydarren to Brecon Gaer, RR620, and RR621 past Maes Mawr Farm to Llanfrynach (RCAHMW 1986: 166) but states that no satisfactory evidence of a Roman origin for this branch has been presented. As far as the ascent of Allt-yr-Esgair is concerned: ...there is little good reason why a laborious ascent to the crest of the ridge need have been made when there are viable routes at several elevations along the W. side (RCAHMW 1986: 168). © SH-K 2011 13 A report by CPAT investigates the fieldwork evidence for RR620 from Taff Fechan to Llanfrynach and although acknowledging that the track across the watershed has a long history, has to conclude that there is no convincing trace of the Roman road, so concurring with the opinion of the RACHMW above (CPAT 2004: 29). However the same report includes a map entitled ‘All known and predicted Roman roads in mid and north-east Wales’ (CPAT 2004: Fig.3). This shows not only RR620 but also shows the route over Allt-yr-Esgair, the route at a lower level next to the River Usk and the road through Maes Mawr Farm that appears in O’Dwyer and the OS Map of Roman Britain 3rd Edition. The latter two roads appear to join at Llansantffraed where there was a ford across the River Usk. The road through Maes Mawr Farm is shown in detail in the 1884 OS 1:2500 map and again in the Ordnance Survey 1904, 2nd edition, Sheet XXXIV.12, Talybont and Llansantffraed. This sheet also shows a Roman road along the south-west flank of Allt-yr-Esgair passing through Tal-y-Bryn-Uchaf. The Ordnance Survey no longer holds details of the written details of the survey for the 1904 map sheet above. Normally the Object Name Books of this era are available to view at the National Archives at Kew. Unfortunately the relevant record series OS35 covering Talybont-on-Usk and Llansantffraed is reported by the National Archives as believed to have been destroyed in the bombing of Southampton during the Second World War so the provenance of the Llansantffraed to Maes Mawr Roman road has been lost.10 This might be the reason why later OS maps do not show it. Returning now to Frontinus’s conquest of the Silures, having reviewed the possible post-campaign road network in the middle Usk valley, particularly in the vicinity of Talybont-on-Usk, it is necessary to distinguish campaigning from road building. As observed by Hugh Davies, the army in the field would not have waited for properly-engineered roads before advancing into enemy-held territory. If suitable tracks were available, they would have been used. If track-ways were not available, the army would have to manage as best it could choosing the most appropriate ground to advance over (Davies 2002: 113). In the case of a battle group advancing up the Usk valley, possibly cooperating with a battle group advancing in the opposite direction from Clyro, and having defeated the populations associated with the Myarth, Allt-yr-Esgair and Pen-y-Crug hill forts, a suitable place for an encampment could have been close to the confluence of the Usk and the Caerfanell rivers such as the site of Maes Mawr Farm. An advantage of a site there would have been the blocking of enemy forces in Glyn Collyn and its many re-entrant valleys together with prevention of access to the Usk valley from the Taff valleys and connected valleys further west. This strategy is seen in many locations in Agricola’s campaigns in Scotland as where forts are often found at the entrance to glens, as observed by Anne Johnson (Johnson 1983: 256). If such an encampment was constructed at Maes Mawr by a task force advancing from Usk or Clyro it would be necessary to build it away from the risk of flooding so a location close to the river junction would have been ruled out. A location at Maes Mawr Farm 10 Letter from the National Archives in response to a request for permission to view Series OS35. © SH-K 2011 14 Figure 2. Speculative outline of 12ha ( 29 acres) fort at Maes Mawr Farm based on Ordnance Survey 1904, 2nd edition, Sheet XXXIV.12, Talybont and Llansantffraed. (Crown copyright expired) would probably have avoided the risk. Although the hydrology of the Usk and the Caerfanell may have been altered by man and nature over the centuries, it is relevant to note that Maes Mawr Farm lies outside the present 1000 year flood plain (EA 2010). It is suggested here that the Roman road shown on the OS 1904 map sheet could well be a causeway built from the ford at Llansantffraed to a campaign base at Maes Mawr Farm. Therefore RR620 would not be the line of a penetration route to this location but possibly an useful ancient track leading from a camp at Maes Mawr Farm, to the high ground dominating Glyn Collyn and Dyffryn Crawnon © SH-K 2011 15 Sites where an alignment of streets within a Roman fortification camp have been approximately preserved despite later building can be seen at Usk (Frere et al. 1987: Fig. 5), Chester (Mason 1987: Fig.1), Leintwardine (Nash-Williams 1969 : Fig. 46), Llandovery (Nash-Williams 1969: Fig. 49) and many others. The lane between farm buildings at Maes Mawr Farm, and taking into account the causeway from the ford over the Usk, is suggestive of the via praetoria of a fort. The porta praetoria would be to the south, facing any enemy in Glyn Collyn beyond the Wenallt spur. This would give fort dimensions of approximately 380m x 300m (Figure 2) and an enclosed area of about 12ha (29 acres). This is approximately 15% larger that the campaign base at Clyro given in Nash-Williams (1969: 77). Factors that ancient treatises give regarding the siting of a fort include ready access to water for men and animals, fodder for animals and wood for fuel. PseudoHyginus, unknown writer of the work De munitionibus castrorum dating possibly from the 3rd century, previously attributed to the surveyor Hyginus Gromaticus, listed desirable topographical characteristics in order of preference: ...in the field, they have as the first choice a rise with a gentle slope in which the porta decumana constitutes the highest point so that the camp is subject to view from there; the porta praetoria always looks in direction of the enemy. The place they have as second choice is one stationed in a plain, the third on a hill, the fourth on a mountain, the fifth in any place unavoidable but designated necessary (De munitionibus castrorum.: 56, trans. SH-K). Given the above extract, the site at Fig. 2 is a second choice as the site is essentially flat. High ground from which a camp could be reached by missile weapons is advised against by Vegetius (Epitoma rei militaris. 3.8). The site is overlooked by the 337m Wenallt that, ignoring the 19th century canal embankment, rises to 15m above the site at a distance of 100m.11 It is not possible to determine how close the forest would have been that might have enabled hostile forces to gather unseen. Johnson suggests a distance of 35 to 45 metres beyond the ditch was necessary to give a clear view from the rampart of a fort and, if necessary, would have been obtained by tree-felling (Johnson 1983: 39). Its overall characteristics are sufficient to encourage consideration of this site as suitable for the location of a fort. Although the ancient sources do not specifically mention the confluence of two rivers as a desirable location for a fort, modern writers have noticed it. John Strange, cited in Manning (1981: 4), visiting Usk in 1779, noted that its position at the junction of two rivers ‘...was in accordance with Roman military practice’. Coxley commenting on Roman connections in 1801, also cited in Manning (1981: 5) noted Usk’s situation at the confluence of two rivers. Caerleon fortress is likewise situated near the junction of the Usk and Afon Lwyd (Nash-Williams 1969: 29). Brecon Gaer overlooks the junction of the Usk and the Yscir (Nash-Williams 1969: 48). However, it is not only the river junction that offers one of the sought-for clues, but the name of the Caerfanell itself that is of interest. 11 Elevation profile data from Google Earth. © SH-K 2011 16 ‘Caer’ is well-known as a Welsh name prefix indicating a fortification. As a suffix ‘manell’, subject to a soft mutation following a feminine noun giving ‘fanell’, is unknown. Samuel Lewis, in his 1849 work A Topographical Dictionary of Wales refers to the river as the ‘Carvenell’, an anglicised version of the name, similar in effect to ‘Cardiff’ for ‘Caerdydd’ (Lewis 1849). Theophilus Jones, historian and etymologist remarks: ... so that unless history or tradition has placed a Roman or British fortress or large carn upon its banks, Carfanell may imply the swift stream from the cliff...(Jones 1898: 428). Jones goes on to poetically warn against reading too much into this name, and into others. In a Welsh poem he calls the river the Afon Anell. Dr Christine Jones points out that ‘Caerfa’ also signifies a fortification but the suffix in this case ‘nell’ is unknown.12 The exact etymology of the river’s name remains hidden but maybe additional research will at last place a Roman fort upon its banks. Frontinus’s tenure as governor of Britain was from 73/74 AD to 77/78 AD (Salway 1981: 746) but how the campaign against the Silures fitted into this period can only be a matter of conjecture, like so much else. However it seems that the fortresses at Caerleon and Chester were completed during these years (Nash-Williams 1969: 29, 35). Caerleon was situated on the navigable estuary of the Usk and was a more suitable base than Usk from which to ship supplies to a task-force penetrating the vale of Neath, through a harbour at Neath (Johnson 1983: 3). It might be that the final campaign against the Silures commenced after a legionary encampment was established at Caerleon. If this is a reasonable argument, it may have a parallel in the construction of the fortress at Chester that may have predated Agricola’s final subjugation of the remainder of Wales; Chester also gave access to the sea. Agricola returned to Britain as governor after the departure of Frontinus remaining until 84 AD (Salway 1981: 746). Under Agricola the forward policy continued. The Ordovices had destroyed nearly a squadron of auxiliary cavalry within their territory just before his arrival. He conducted a campaign against this tribe and cut to pieces the whole of their force (Agricolae: 18). Tacitus then sings the praises of his father-in-law as he compelled multae civitates, previously belligerent, to treat for terms by bold military operations combined with clemency and inducements to peace. The location of these many polities is not stated. Tacitus relates how a ring of garrisons and forts was placed around them: praesidiis castellisque circumdatae (Agricolae 20). The time frame of this encircling deployment is not known but it seems to come within the period in which the Flavian forts at Pen-y-Gaer, Brecon Gaer, Llandovery, Coelbren, Penydarren and Gelligaer are thought to have been built. An earth and timber cavalry fort at Brecon Gaer was originally dated to 75 – 80 AD (Nash-Williams 1969: 51). RCAHMW (1986: 144) reports that Dr. Grace Simpson prefers to date the earliest occupation to c.80 AD. This date is in the 12 Dr Christine Jones; Head of Department of Welsh, University of Wales Trinity St David, Lampeter.; by email. © SH-K 2011 17 Agricolan period and might be supportive of this fort and the others above being a praesidiis castellisque circumdatae of Agricolae 20. If so, then the speculated fort at Maes Mawr may have been in occupation from c.75 AD to c.80 AD. Conclusion This paper has presented a scenario for the conquest of the Silurian territory by the Romans in the years c.47 to c.80 drawing much upon the remarkable work of Jeffrey Jones and Rebecca Davies in relation to marching camps. A possible sequence of events has traced the Roman intrusion into and final conquest of Siluria. In particular, a search for clues that might reveal Roman disposition in the middle Usk valley has highlighted the need for further consideration of the Maes Mawr Farm site as a campaign base during Frontinus’s successful campaign. It seems fairly certain that there are fortification sites that remain hidden. It was only in the 1960s that a fullsized legionary fortress was revealed to lie under the town of Usk so it is not implausible that this river valley may contain further surprises. As far as can be ascertained, no archaeological interest has previously been shown in this site, possibly because of literally earth-moving events of the early and midnineteenth century: the construction of a canal and the building of a railway and a road. In addition, modern development has taken place on the site, a development that is likely to continue. Confirmation or denial of the conjecture here presented might be obtained by the use of a non-invasive investigative technique such as magnetrometry. © SH-K 2011 18 Bibliography Primary sources Claudius Ptolemy. Geographia I-IV (trans.) Müller C. and C. T. Fischer 1901 (Paris) Cornelius Tacitus. The Annals (trans.) Church A. J. and W. J. Brodribb (Perseus Digital Library, Tufts University: Boston http://www.perseus.tufts.edu/) Cornelius Tacitus. Annales (ed.) Fisher C. D. (Perseus Digital Library: http://www.perseus.tufts.edu/) Cornelius Tacitus. The Life of Cnæus Julius Agricola (trans.) Church A. J. and W. J. Brodribb (Perseus Digital Library, Tufts University: Boston http://www.perseus.tufts.edu/) Cornelius Tacitus. De vita Iulii Agricolae (ed.) Furneaux H. (Perseus Digital Library, Tufts University: Boston http://www.perseus.tufts.edu/) Sextus Julius Frontinus. Strategmata (trans.) Bennett C. (University of Chicago: online: http://penelope.uchicago.edu/Thayer/E/Roman/Texts/Frontinus/Strategemata/home.html) Gaius Suetonius Tranquillus The Lives of the Caesars (trans.) Thomson A. (Perseus Digital Library, Tufts University: Boston http://www.perseus.tufts.edu/) Publius Flavius Vegetius Renatus. Epitoma rei militaris (trans.) Lang C. (1885) (The Latin Library: http://www.thelatinlibrary.com/vegetius.html ) Pseudo-Hyginus. De munitionibus castrorum (The Latin Library: http://www.thelatinlibrary.com/hyginus/hyginus6.shtml ) Secondary sources Codrington, T. 1903. Roman Roads in Britain (Society for the Propagation of the Christian Gospel: London) CPAT 2004. The Roman Roads of East and North-East Wales, Report 621. (ClywdPowys Archaeological Trust: Welshpool) Davies, J. and Jones, R. 2006. Roman Camps in Wales and the Marches (University of Wales Press: Cardiff) Davies, H. 2002. Roads in Roman Britain (Tempus: Stroud) Dudley, D.R and G.Webster. 1962. The Rebellion of Boudicca (Routledge & Keegan Paul: London) EA 2010. Flood Modelling On-line: Post-code LD3 7YX (Environment Agency: © SH-K 2011 19 Cardiff) http://www.floodrisk.co.uk/flood_consequence_assessment_wales.htm Fenton, R. 1917. Tours in Wales 1804 – 1815 (ed.) Fisher J. (The Cambrian Archaeological Association: Cardiff) http://www.archive.org/details/toursinwales180400fentrich Frere, S. S. 1987. Roman Britain in 1987 in Britannia Vol. 19 (Society for the Promotion of Roman Knowledge: London) 415-508 Gilliver, C. M. 2001. The Roman Art of War (Tempus: Stroud) Jarrett, M. G. 1965. Early Roman Campaigns in Wales reprinted from The Archaeological Journal Vol. CXXI. (The Royal Archaeological Institute: London). Johnson, A. 1983. Roman Forts of the 1st and 2nd Centuries AD in Britain and the German Provinces (Adam & Charles Black: London) Jones, B. and D. Mattingly 1990. An Atlas of Roman Britain (Guild Publishing: London) Jones, T. 1805. The History of the County of Brecknock (reprint 1898) (Davies: Brecon) Keppie, L. 2004. The Making of the Roman Army: From Republic to Empire (BT Batsford: London) Lewis, S. 1849. A Topographical Dictionary of Wales: Llanthetty (Institute of Historical Research: online) http://www.british-history.ac.uk/report.aspx?compid=47857#s7 Manning, W.H. 1981. Report on the Excavations at Usk 1965-1976: The Fortress Excavations 1968-1971 (University of Wales Press: Cardiff) ____________2004. The Conquest of Wales in Companion to Roman Britain (ed.) Todd M. (Blackwell Publishing :Malden) Mason, D. J. P. 1987. The Canabae Legionis in Britannia, Vol. 18 (Society for the Promotion of Roman Studies: London) 143-168 Nash-Williams, V.E. 1950. The Early Christian Monuments of Wales (University of Wales Press: Cardiff) ________________ 1969. The Roman Frontier in Wales (2nd Edition revised by M. G. Jarrett) (University of Wales Press: Cardiff) O’Dwyer, S. The Roman Roads of Wales VI, Brecknock & Glamorgan with maps (Montgomery Printing Company: Newton) © SH-K 2011 20 P. C. 1939. Review of S. O’Dwyer, The Roman Roads of Wales VI, Brecknock & Glamorgan with maps in The Geographical Journal Vol. 94 No. 1 57-58 RCAHMW 1986. Brecknock: Hill-forts and Roman Remains (The Royal Commission on Ancient and Historical Monuments in Wales:HMSO) Salway, P. 1981. Roman Britain (Clarendon Press: Oxford) St Joseph, J. K. 1961. Air Reconnaissance in Roman Britain, 1958 –1960 in The Journal of Roman Studies, Vol. 51, Parts 1 and 2 (1961) (Society for the Promotion of Roman Studies: London) 119-135 ____________1973. Air Reconnaissance in Roman Britain, 1973-1976 in The Journal of Roman Studies, Vol. 67, 91977) (Society for the Promotion of Roman Studies: London) 125-161 Webster, G. 1981. Rome Against Caratacus (Routledge: London) © SH-K 2011 21