1 WARFARE IN EARLY ISLAM SUBMITTED TO THE FACULTY OF ARTS AT THE UNIVERSITY OF GLASGOW IN FULFILMENT OF THE REQUIREMENTS OF THE DEGREE OF DOCTOR OF PH...
1 ýý17 94 EARLY MEDIEVAL EUROPE Edited by T. S. Brown Edward James Rosamond David McKitterick Rollason Alan Thacker VOLUME LONGMAN2 Theo...
1 2 THE MIDDLE AGES The Middle Ages is a historical period that began with the fall of the Roman empire in the 5th century and ended with the start of...
1 at Zone E, Beaulieu Chelmsford, Essex Archaeological Evaluation Report December 2016 Client: Countryside Zest (Beaulieu Park) LLP OA East Report No:...
1 2 CHINA RESEARCH MONOGRAPH 71 CENTER FOR CHINESE STUDIES Early Medieval Chinese Texts A Bibliographical Guide Edited by Cynthia L. Chennault Keith N...
1 2 zvi The Routledge Companion to Medieval Warfare This comprehensive volume provides easily accessible factual material on all major areas of warfar...
1 Zurich Open Repository and Archive University of Zurich Main Library Strickhofstrasse 39 CH-8057 Zurich Year: 2016 French studies: early medieval ...
1 Elisabeth Davis Dr. James Bono Late Medieval, Early Modern List Late Medieval Cohn, Norman. The Pursuit of the Millennium: Revolutionary Millenarian...
1 Early Christian and Medieval Martyrdom Let s start by reminding ourselves of the terrain we re attempting to delineate Review: martyr vs. saint At t...
1 Routledge Early Modern and Medieval History New Titles and Key Backlist2 Cover image: Taken from the cover of The European World , detail from Simon...
Early Medieval warfare: 411-1065 There is a total of 232 early medieval records on the database for England. Many cannot be classified as to type of action, or have been assessed as spurious. A further 29 appeared to be sieges and at least two were assassinations. This left 154 battles or possible battles – ‘possible’ because they are attended by uncertainties, and the paucity of contemporary primary written sources. Of the 154 at least 20 actions have more than one possible location, including Brunanburh 7, Ethandun 6, Ashdown 5 and Ashingdon 3. This leaves a total of 114 individual actions.
Figure 20: Distribution and some associations of early medieval battles
Thanks largely to the battle names, in most cases the information in the database has sufficient detail to enable analysis at least on a national scale. However, the bias that is introduced by the level of detail for Wessex from the AngloSaxon Chronicle is immediately apparent. Of those actions with more than one possible site some, like Maldon, offer a reasonable degree of confidence about location, but a small number, including some of the most important of battles, have alternative sites which are widely separated. Thus, for example, Brunanburh has alternative sites right across northern England and into southern Scotland, though place-name analysis has demonstrated a high probability that site lay in the Wirral.1 This is not to suggest that battles recorded with a single location are accurately located, for many are traceable only to a general area. For many, the evidence comes from just one reference in one source – the Anglo-Saxon Chronicle – and though the combatants are often known even the outcome is frequently uncertain and the nature of the action even less clear. Thus for example Bedcanford in 571 is now unlocated because the place-name evidence can no longer support identification with Bedford. In the case of Luton the place name association is accepted but there is no evidence as to where the site may be in the surrounding territory. The lack of topographical information from almost all the documentary records of the battles of the period means that there is rarely any evidence by which to pinpoint the site. One exceptional type of evidence which may assist in the resolution of problems for a few battles is where memorial chapels were established to commemorate them, as for example at Assandun. However, as is seen below with Heavenfield, the chapel does not always stand on the battlefield. There are also problems with earlier identifications of finds and burials as coming from early medieval battlefields. Cemeteries of various date have been misinterpreted as battlefield burials, such discoveries in the 19th century or before sometimes prompting assumptions about battles and giving rise to local names that have themselves become part of a spurious evidential base. In the case of Bloodmoor Hill in Gisleham, Suffolk, for instance, the name is now known to have been bestowed following the opening of an Anglo-Saxon barrow in 1758.2 On the other hand, in a few cases the attribution of a battle name may be genuine, perhaps for example with Battle Wood in Stoke Lyne, Oxfordshire, of which Stenton points out that the 1
Higham, 1997 Newman, 1996, suggests that the description of the area in past records as a battlefield probably resulted from the discovery of early Anglo-Saxon burials with weapons. 2
presence of a 12th-century name of a wood there as Fethelee can be linked to the battle of Fethanleag in 584.3 Until the 9th century the distribution, and some cases historicity, of early medieval battles must be viewed with the further caution that compilation of the Anglo-Saxon Chronicle itself did not begin until the 890s, and that although the Chronicle drew from other sources and annals that in some cases have not survived, the degree of confidence in detail that attaches to its record for the 5th-7th centuries must often be small. This notwithstanding, when the early medieval distribution is compared against a number of other data sets, some general conclusions may be drawn. One of them is with Roman roads, where there is some degree of patterning at a strategic level, though in reality it is only when examined at a very local level that significant conclusions might be drawn as regards the actual sites. Similarly, there appears to be a degree of concentration of battles along some boundaries between AngloSaxon kingdoms – but again these boundaries are often modern supposals, and ‘kingdom’ in the 7th century may been as much or more a matter of the extent of family or dynastic influence at a given moment than a geographical area with permanent borders. 4 This analysis can be extended to the patterning by chronology, where conclusions may be drawn about the spread of warfare across the country in different periods. However, none of this assists in the identification of the actual sites of the battles which is critical to the present purpose.
Table 1: Early Medieval Battles by Century century 5th 6th 7th 8th 9th 10th 11th (to1065)
number 7 14 17 5 36 15 20
Stenton, 1947, 29; Trans Royal Hist Soc, 4th Series, xxii, 19-20 Wood, 2007
Figure 11: Distribution of early medieval battles by century (5th–8th), compared to total distribution
Figure 22: Distribution of early medieval battles by century (9th–11th), compared to total distribution
Secure and substantial archaeological evidence has yet to be retrieved from any English battlefield before the 15th century. Indeed across Europe, apart from mass graves, the only major study of an earlier battle to have produced extensive battle archaeology is that at Kalkriese (Germany, AD 9), where exceptional preservation conditions prevail.5 Burials which may relate to the battle of Chester in the early 7th century have been located at Heronbridge, but the association is insecure.6 There is also a report of archaeological finds from the Fleet Valley in London which have been suggested as representing the battle of London in 1066; this needs to be assessed against later battlefield deposits.7 While chance discoveries of burials may cast light on warfare, it is only when spatial patterning 5
across a landscape is identified that the physical evidence can make a major contribution to the understanding of a battle. The iron spear is likely to have formed a substantial proportion of the projectiles used in the early Middle Ages, although to judge from manuscript and other evidence (e.g. sculpture), iron arrows were also significant in some actions. Since spears points will have been far easier to retrieve for re-use during or immediately after a battle, they are likely to have been so. Hence, where artefacts do survive their patterns of distribution may be far less dense or representative of the action than arrowhead spreads of the Hundred Years War or Wars of the Roses. A question not yet settled, and critical in relation to survival, is the extent to which early medieval equipment lost on the field will have been of ferrous or nonferrous metal. In addition, account must be taken of the length of time that artefacts will have lain in topsoil. If we accept the traditional dates for, say, Badon, Heavenfield (Denisesburn), and Ethandun, then these periods would be 1500, 1375, and 1130 years, respectively. Thus the artefact distributions for early medieval battles, if they survive, will probably be sparser than for later periods. It may only be where part of a battlefield surface has been buried beneath alluvium, waterlogged, or left uncultivated, that battle artefacts may survive in sufficient numbers to permit analysis. Until we are able to locate early medieval battlefields, a research strategy to address them archaeologically remains out of reach. Even Maldon, the only example to have been Registered, is doubtful (below, p.80). Study of the strategic landscape of the early medieval period may assist in focusing attempts to identify the general location of battles in this period, and in this context the research currently underway to map Anglo-Saxon civil defence could lead to substantial advances, and possibly offer methodological innovations that could be employed elsewhere (see pp 85-87). However, it is only through the presence and interrogation of intelligibly-patterned battle archaeology that resulting hypotheses can be tested, and unless or until such an assemblage can be identified the location of battles of the period is likely to remain conjecture. An exception may be through the discovery of mass graves, and there may also be potential to identify siege sites. Resolution of the problem calls for a two-fold approach. One path is to approach from Hastings backwards, for Hastings is the most accurately-located of this group of battles, and although it comes at the very end of the period, tactics and equipment of preceding centuries were still in use by the English army. The other is to seek one or more of the ‘battles’ of the period which appear to have been an attack on a defended location and where that location can be identified with some
confidence. For example on might choose an action directly associated with a Roman fort, or a located centre such as Bamburgh. However, problems would then arise regarding the conduct of a detecting survey around or on such sites. The most viable strategy, assuming favourable soil chemistry, thus seems to be to target Hastings. Maldon Maldon is the only battle from before 1066 where the battlefield has been identified with some confidence. The Anglo-Saxon Chronicle states that the battle took place near Maldon (clearly Maldon in Essex, because Brihtnoth, the English commander, was Ealdorman of Essex while the earlier plundering by the Viking force took place at Ipswich, 30 miles to the north east). The position of the battlefield has been interpreted from topographical evidence within the broadly contemporary poem The Battle of Maldon.8 The key topographical facts are that the Viking force landed or marched to a position on one side of a channel of the Blackwater estuary or a tidal section of the river, then called ‘Panta’. At high tide they were unable to cross but the channel was sufficiently narrow to enable the opposing forces to negotiate by shouting across the water. When the tide fell and they were allowed to cross they did so, across a causeway (brycg) or ford, in a westerly direction. The interpretation currently favoured is that the battle was fought on the mainland immediately opposite Northey Island, in the Blackwater estuary just to the east of the town of Maldon. However, this has been challenged,9 and there have been three main candidates for the site. One lies immediately north east of Maldon at the place now known as Heybridge, now completely built up as a suburb of Maldon.10 The Viking force was suggested as having crossed north eastward from just outside the Maldon defences to attack the East Saxon army on the north east bank. The case for this site has been rejected by more recent authors, partly because it does not appear to meet the criteria relating to the tides, and partly also because the poem makes no reference to Maldon town, which would have been immediately to the back of the Viking force.11 The second candidate was Osea Island, which lies in the north of the estuary and is connected to the mainland by a causeway. But even when sea level rise is
taken into account, the 1 km distance between the island and the mainland is too great to fit the events in the poem.12 The strongest candidate is Northey Island, which has been shown to fit closely the topographical features described in the poem, including the way in which the rising tide flows around both sides of the island to meet at the causeway.13 While today the distance from mainland to the island might be considered too great for a shouted exchange, it has been shown that in this area land has been sinking since the last glaciation and that sea level has risen by about 6ft since the 10th century. As a result there are now saltings where there was dry land at the time of the battle, which would have made the channel no more than about 110m wide at that time. The adjacent land was dry pasture, dipping gently to a steep river bank. In AD 991 the causeway would have been covered by 6-8 ft of water at high tide, while silts similar to those seen today would have made crossing the channel impossible except by a causeway.14 According to the poem there was a wood nearby, for one Englishman let his hawk fly to the wood before he advanced to battle, while others fled to the woods after Brihtnoth’s death. The parishes to the west of Northey Island have woodlandrelated names, Woodham and Hazeleigh and Purleigh.15 However, no reconstruction of the distribution of woodland in this area in the medieval period has apparently been attempted so it is unclear how closely the woodland might have extended towards the battlefield, or indeed at what date(s) it may have existed. Despite the apparent strength of the case, therefore, unless or until archaeological evidence for the battle is recovered, Maldon’s identification remains open.
Heavenfield The battle of Heavenfield, a celebrated encounter between king Oswald and Cadwallon in the early 630s, is widely assumed to have taken place just to the north of Hadrian’s Wall, a short distance east of Chollerford, at a place commemorated at least from the later Middle Ages by a chapel, and by an annual procession of canons from Hexham Abbey. If this were the battle site, then Heavenfield would certainly be a candidate for the closest archaeological survey. In fact, re-examination of all the available sources for the purposes of this project shows that Heavenfield and the battle were in
English Heritage, 1995a Laborde, 1925 14 Petty G.R. and Petty S., 1976. Petty and Petty, 1993 15 Dodgson John McNeal, 1991 13
different places, and that almost all commentators have been wrong about the location of the initial encounter.16 In summary, Heavenfield was north of the wall, but was the site of Oswald’s camp before the battle, not the battle itself. The battle concluded with the death of Cadwallon at Denisesburn, south of the Tyne. Denisesburn probably equates with today’s Rowley Water.17 It is not clear where the battle began. Leland, working from an older tradition, may provide a clue when he says: ‘There is a fame that Oswald won the betelle at Halydene, a two miles est from St. Oswalde’s asche.’18 Plummer identifies Halydene with Hallington,19 but this is north of the Wall and can be discounted.20 This would seem to suggest that Oswald was at Heavenfield, while Cadwallon was in the vicinity of Onnum/Halton: that Cadwallon fled south through Corbridge, but was caught at the far side of the Tyne. Oswald would seem to have attacked at night: Bede says he attacked incipiente diluculo, but Adomnan says clearly that the attack took place at night.
The leading source is Bede, writing in the earlier 8th century: HE III 1, Quo (Osualdo), post occisionem fratris Eanfridi, superueniente cum paruo exercitu, sed fide Christi munito, infandus Brettonum dux cum inmensis illis copiis, quibus nihil resistere posse iactabat, interemtus est in locol, qui lingua Anglorum Denisesburna, id est riuus Denisi, uocatur; HE III 2, on erection of cross: Fecerunt omnes, ut iusserat, et sic incipiente diluculo in hostem progressi, iuxta meritum suae fidei uictoria potiti sunt …Uocatur locus ille lingua Anglorum Hefenfelth, quod dici potest latine caelestis campus, quod certo utique praesagio futurorum antiquitus nomen accepit …Est autem locus iuxta murum illum ad aquilonem, quo Romani quondam ob arcendos barbarorum impetus totam a mari ad mare praecinxere Brittaniam … Other references to the battle site occur in Historia Brittonum 64, Oswald … Ipse occidit Catgablaun, regem Guenedotae regionis, in bello Catsgaul, cum magna clade exercitus sui; Annales Cambriae, s.a. 631, Bellum Cantscaul in quo Catguollaun corruit. Plummer, II, 121: ‘The name of the battle both in Ann. Camb. 631 and in Nenn. § 63 is ‘bellum Catscaul’ (= cath-is-gwaul, ‘the battle within the wall.’) Clearly it is not at Heavenfield (thus destroying the argument of Raine, Hexham, I, xi, ff., cited in Plummer, II, 122-3). On the other hand it is obviously south of the Wall. On Heavenfield: also Adomnan, Vita Columbae I 1, 14-15: Before the battle Oswald had a vision of Columba: ‘Sanctus igitur Columba haec ad regem in uisu loquens addit: ‘Haec sequente nocte de castris ad bellum procede. Hac enim uice mihi dominus donauit ut hostes in fugam uertantur tui, et Catlon inimicus in manus tradatur tuas, et post bellum uictor reuertaris et feliciter regnes.’ … Eadem subsequta nocte Osualdus rex sicuti in uisu edoctus fuerat de castris ad bellum cum admodum pauciore exercitu contra milia numerosa progreditur. Cui a domino sicut ei promisum est felix et facilis est concessa uictoria, et rege trucidato Catlone uictor post bellum reuersus postea totius Brittanniae imperator a deo ordinatus est.’ 17 Plummer, II, 123: Denisesburn is identified by a charter edited by Raine, Hexham, Appendix iv: ‘Dedit … archiepiscopus … xx acras terrae … in Ruleystal … inter Denisesburn et Diuelis.’ 18 Itin. Ed. 2, vii. 58 (cited by Plummer, II, 23) 19 Plummer, II, 123 20 There is, however, a Halton approximately four miles east of the supposed site of Heavenfield, just off Dere Street: In fact, Halton Chesters = Onnum is the fort adjacent to where Dere Street crossed the Wall.
One overall reconstruction would leave Heavenfield where everyone thinks it was: but it as Oswald's camp, not the battlefield. Oswald then launched a night attack on Cadwallon's camp, which was near Dere Street. Cadwallon and his army fled back down the Roman road, but were caught just after crossing the Tyne bridge at Corbridge, and the rout took place at Denisesburn, which flows into the Tyne effectively at the bridge crossing.21 This still leaves questions. There is nothing in the Roman or later road pattern to suggest why an army in flight would be anywhere near the Rowley Burn as we know it today, whilst if Cadwallon’s forces had broken, the greater bottleneck would be at the crossing of the Tyne. Is it possible that we are missing a key piece of the Roman or medieval road pattern? If the battle had been fought close to Corbridge, why was it not called the battle of Corbridge, as were a number of other battles? If it was fought at or near Halton or the fort there then why not name it again after Corbridge as the parochial area (admittedly much later though presumably reflecting some early pattern) or the Onnum fort or Halton? A part-answer to some of the questions may lie in the devotional background to the battle, to which Bede and Admonan give much prominence. Since Heavenfield was where Oswald and his army prayed for victory, in Bede’s mind, was where the battle was actually won, even though the physical events occurred elsewhere? ‘Heavenfield’ is clearly not a battle that can yet be located for archaeological purposes; it is a question whether any other important battle in England during the early middle ages can be located more closely. However, while discussion thus far has been tinged with pessimism, recent interdisciplinary work in Wessex offers a methodological way forward. This is now summarised.
The Rowley Burn as it is today is the stream that joins the Devil’s Water to the east of Juniper, some 6km south-west of Corbridge. Close to the head of the burn there is Rowley Head, recorded in the 18th century on Armstrong’s map. The OS 1st edition gives Devil’s Water for the stream that runs into the Tyne next to Corbridge; this is also the case on the Armstrong map, and is supported by the place name Dilston – which has a ‘devil’ derivation according to the English Place Names Society, though Mills (2003) gives ‘dark stream’.
Figure 23: Battle of Heavenfield or Denisesburn: of the alternative sites for the battle, the one adjacent to Hadrian's Wall is now known to be incorrect. The other lies somewhere in the vicinity of Rowley Burn
On locating early medieval battlefields22 Although primary textual sources for the early medieval period give only a limited amount of information about battles, a number of strands of evidence can be used to identify early battlefield locations and aspects of the historic terrain. These include: • • • •
The nature of the action (e.g. set-piece battle, raid, skirmish) The location of communications (e.g. roads, signalling systems, mustering points) Topography (geology, relief, and vegetation) Other monuments (e.g. fortifications, meeting places)
This section paraphrases the forthcoming work of J Baker, S Brookes, D Parsons and A Reynolds, Beyond the Burghal Hidage. We are extremely grateful for their contribution.
Two main forms of battle can be discerned from primary sources. Most of the battles are characterised by swift raids into enemy territory for the purpose of inflicting damage and capturing booty. A second – less common – form of battle is large setpiece confrontations. The purpose of these was to force the opponent into submission – by force if necessary - which was indicated through the giving of hostages and tribute. Both forms of engagement rest on a number of conditions which are potentially reconstructible by archaeology. For the second category of battle in particular the physical make-up of the landscape is highly significant. Firstly, gathering large forces together, and coordinating confrontations was very difficult. There are indications in the sources that places of assembly were an integral part of these battles, whether this meant the gathering together of friendly forces as part of a local or national mobilization, the meeting of hostile forces on the field of battle, or the coming together of opposing forces for the drawing up of a treaty. From first summons, through local and regional mobilization, to final arrival at the battlefield, this demanded a complex level of organisation and a pre-arranged network of assembly-places. The origins of the hundredal system are unclear, but it is a natural supposition that the mobilization of an army took advantage of existing governmental structures, with forces gathering first at local moots, then regional meeting places, before assembling with the rest of the fyrd at a predetermined spot. The meeting at Swinbeorg, mentioned in Alfred’s will, is traditionally taken to refer to a mustering at the assembly-place of Swanborough hundred (W) prior to the Battle of Ashdown in 871. In 878, before the battle of Edington, Alfred stopped first at Ecgbryhtesstane ‘Egbert’s Stone’ – on the borders of Somerset, Dorset and Wiltshire and perhaps an established focal point – then at Iley Oak, the meeting-place of the hundreds of Warminster and Heytesbury. The battle site itself is near the probable meeting-place of the hundred of Whorwellsdown, in the vicinity of Cresswell Down Farm. Perhaps the most essential feature of all meeting-places, whether for mustering, diplomacy or battle, was their accessibility. In many instances, this meant the proximity of major roads or waterways, but it could equally mean their location on the borderland between two hostile territories. The place-names of numerous meeting places give quite precise topographical information which can potentially be used to define the location of battlefields. The place-name, OE ēagemōt ‘junction of streams’, in reference to the confluence of the Lowther with the effluent of Ullswater lies beside the line of the old Roman road from Brough to Carlisle. Another place-name compound that may be relevant to a discussion of
mustering is OE here-feld ‘army field’. These are often very close to major boundaries. Halsall suggests that the limited number of campaigning routes and the basic nature of medieval scouting made selection of a battlefield a question of convenience as much as tactical choice, noting that twenty-two of the twenty-six located Anglo-Saxon battlefields of the period c.600 to c.850, were at river-crossings or significant landmarks.23 He nevertheless points to examples of armies stationing themselves at significant legal or administrative centres, perhaps as a form of challenge to the opposition. The activities of the Vikings at Cuckhamsley in 1006 might perhaps be viewed in this light. The Viking occupation of a shire meeting-place on a prehistoric monument was perhaps a confident (and ultimately successful) challenge to the West Saxon levies to meet them in battle. The Cuckhamsely incident demonstrates that distinctive topographical features appear to have been highly relevant. The high number of later Anglo-Saxon military engagements with place-names in ford – Fulford, Stamford Bridge, Brentford, Thetford, Hereford and Castleford – suggests that river-crossings were still often the sites of battles, and by extension, that accessibility was of prime importance. This is hardly surprising, but it is notable that many battles fought between c.850 and 1066 were in close proximity to Anglo-Saxon fortified sites, Viking encampments, or prehistoric monuments – Hastings, Wilton, Maldon, Hereford, Cynuit/Congresbury, Thetford and Reading – many of which might be expected to have been focal points for the mustering of troops. They were also, of course, notable landmarks and therefore perhaps useful for armies attempting to locate each other. For the purposes of mustering and engagement in battle, accessibility would need to be combined with ease of identification, especially in cases where the fyrd was operating outside its own shire. In many cases, shire meeting-places were marked by distinctive landscape features, identifiable through fieldwork. If established meeting-places were convenient general locators for battle, the specific choice of battle site was subject to other considerations of terrain. Although contemporary sources seem relatively unconcerned with the tactical and spatial details of battle, recording only the general locations and the outcomes, terrain was obviously an important factor in the battles between Anglo-Saxons and invading armies, influencing the choice of landing places and the deployment of troops. The location of several battles involving the ‘seizure of ships’ are constrained by topographical factors that include terrestrial and submarine contours; harbours;
Guy Halsall, Warfare and society in the barbarian west, 450-900, 2003, 156
navigable waterways; extent of the foreshore, marsh or sandbars; extent of the intertidal zone; the tidal head for each creek and river; deep water channels; the extent of artificial harbour works; and access points – all of which are potentially accessible archaeologically. In other cases such features as marsh, sharp relief, woodland or causeways might also play a determining role in the identification of battlefields. It thus appears that at least some early medieval battlefields are potentially accessible to landscape archaeology, but that this requires a detailed knowledge of the civil administrative structure of middle and late Anglo-Saxon kingdoms, early medieval logistics, and topographical conditions. It is a conclusion of the research by Baker, Brookes et al that these various evidences can be confidently reconstructed and used in combination to locate battlefields on the ground. The data in their forthcoming study Anglo-Saxon Civil Defence in the Viking Age should be assessed to identify sites for which this potential can be explored.