Where was Brunanburh?

The battle

In late summer 937 (if the Irish annals have the order right) Anlaf Guthfrithson, the Viking king of Dublin, whom the Irish annals call Amlaíb, sailed from Dublin. He had spent that summer raiding the Irish midlands from a base on Lough Ree, a lake on the River Shannon. He was sailing to join two kings coming by land: Constantine II, king of Alba, the Gaelic kingdom that became Scotland, whom a twelfth-century chronicler calls Anlaf's father-in-law; and the king of the Britons of Strathclyde, the kingdom on the Clyde whose people spoke a language close to Welsh and whom English writers called the Cumbrians. Their enemy was Æthelstan, grandson of Alfred the Great. Æthelstan had ruled Wessex and Mercia, the southern and midland English kingdoms, since 924; in 927 he had taken York, the Viking kingdom that Anlaf's father Guthfrith had ruled until Æthelstan drove him out, and had become the first king to rule the whole of England; in 934 he had invaded Scotland by land and sea.

Æthelstan and his brother Edmund defeated the three armies at a place the English called Brunanburh. The Norse survivors got back to their ships and sailed for Dublin; Constantine fled north; among the dead were Constantine's son, five young kings and seven of Anlaf's earls. The victory was set in verse in Old English, the English of the period, and the poem was copied into the Anglo-Saxon Chronicle, the year-by-year record of events kept in English monasteries, under the year 937. It names the field, Brunanburh, and the water over which the Norse fled, Dingesmere, and it says that never yet on this island had a greater slaughter been made by the sword's edge since the Angles and Saxons first came over the sea (ll. 65–70). The contemporary Annals of Ulster, a year-by-year record kept in Irish monasteries, call it "a great, lamentable and horrible battle" in which several thousand Norsemen fell; Æthelweard, an English nobleman who wrote a Latin chronicle around 980, says it was still known as "the great battle".

Why it is a mystery, and who says what

The poem names the field but does not say where it was, and no place in Britain bears the name Brunanburh today. No other source says where it was for more than a century and a half. The Annals of Ulster give no place at all. The Welsh Annals give one word, Brune. Æthelweard says the fight was at Brunandun, and a text from Durham calls the place Wendune; neither is located. Then, between about 1095 and 1140, John of Worcester, a monk of Worcester writing a Latin chronicle of England, wrote that Anlaf's fleet "entered the mouth of the river Humber", the great estuary on England's east coast. Every later medieval writer who names a place copies him.

More than forty sites have been proposed, from Devon to Dumfriesshire; four are still argued in print. The first is Bromborough, a village on the Wirral, the peninsula between the Dee and Mersey estuaries opposite Liverpool: its medieval spellings descend from Brunanburh, it lay in Norse-settled English territory a day's sail from Dublin, and since 2019 a community group, Wirral Archaeology, has been recovering metal finds on Bebington Heath nearby. This is the case of Paul Cavill, a place-name scholar at the University of Nottingham, and Stephen Harding, a Nottingham scientist who writes on the Wirral's Viking past, whose book Finding the Battle of Brunanburh (University of Chester Press, December 2025) argues it at length; the university's press release said it "paves the way for ending the debate". The second is the Humber. Michael Wood, the historian and broadcaster, takes John of Worcester at his word and puts the field south of York, first at Brinsworth near Rotherham (1980) and later in the Went valley near Doncaster (2013), and others have proposed sites along the Humber's banks. The third is Burnswark, a flat-topped hill in Dumfriesshire with an Iron Age fort on its summit and two Roman camps on its slopes, near the Solway Firth (the inlet between Cumbria and south-west Scotland) and so nearest of all to Dublin and Strathclyde; Kevin Halloran, an independent historian, argues that the war was a Scottish enterprise rather than a Dublin one and was fought there. The fourth is Lanchester, a Roman fort in County Durham above a river called the Browney, argued by Andrew Breeze, a Celtic philologist at the University of Navarra, who reads the name as "the fort on the Brune". There is also a sceptical position, held by Tim Clarkson, a historian of Strathclyde, and Thomas Williams, a historian who reviewed the archaeological claims, among others: the original name may have been lost and the finds do not prove a battle, so no identification is secure. Historic England, the public body that keeps the official register of battlefields, does not list Brunanburh, because the register requires a battlefield to be securely located.

How the case was judged

This report weighs those cases against the sources in four steps. It reads every medieval text that mentions the battle in its original language (Old English, Latin, Irish, Welsh and Old Norse: 22 texts in all), asks which of them are independent witnesses and which are copies, and finds that no source before John of Worcester locates the battle at all, and that his sentence is a stock formula. It tests each candidate's name against the grammar of Old English, in which Brunanburh means "Brūna's stronghold" and the spelling of the first element tells you what kind of name it was. It computes, on one stated basis, how far and how long each of the three allied armies and Æthelstan's would have had to travel to reach each site. And it combines those judgements in an explicit evidence model: each candidate starts with a prior probability, each line of evidence multiplies it up or down by a stated factor with a stated margin of doubt, and the calculation is re-run 20,000 times with every factor varied within its margin, to show how much the ranking depends on any one judgement.

The result is a ranking with a stated uncertainty; it does not identify the site. Bromborough is the leading candidate, at roughly 60 per cent; there is about one chance in five that the battle was fought somewhere nobody has proposed; and no named rival reaches 5 per cent at the point estimate, though in the re-runs the strongest of them, the Lancashire coast, reaches about 10 per cent at its 95th percentile. The percentages encode a weighting of the evidence rather than any measurement, and the later sections set out each weight so that a reader can substitute their own.

What this report covers

The sections follow the four steps above, with a section on the candidates between the routes and the model. First, what the medieval sources say, quoted in the original languages with translations, and which of them are independent. Second, the name, and whether each candidate's recorded spellings can descend from Brunanburh, with a separate section on the poem's second name, Dingesmere. Third, geography: the routes and the sailing and marching times. Fourth, the candidates one by one, with what has changed since 2011, when Michael Livingston of The Citadel military college in South Carolina edited The Battle of Brunanburh: A Casebook, which collected all the sources with translations and essays on the site; the changes since then are the Wirral Archaeology finds and the Cavill and Harding book. Fifth, the evidence model that turns those judgements into the chart above, and how fragile its ranking is. After that come the findings that would change the answer, what this report adds to the published scholarship and what it could not check, a table of where each historian stands, the references, and the files and code.

Claims on which the argument depends are cited to an edition, page, manuscript transcription or URL. The dossiers listed at the end record, item by item, which texts were read directly and which only through another author's quotation, and the code and data are there so that the model can be re-run with different judgements.

What the sources say

Three layers of testimony survive. The contemporary layer is the Old English poem entered in the Anglo-Saxon Chronicle under 937, the Annals of Ulster, Æthelweard's Latin chronicle (c. 980), the Annales Cambriae (the Welsh Annals, a year-by-year record kept in Wales), and the Chronicle of the Kings of Alba, a short Latin king-list with notes kept in Scotland. The Anglo-Norman layer (c. 1100 to 1140, after the Norman Conquest) adds Symeon of Durham, a monk writing at Durham around 1100 to 1110; John of Worcester; William of Malmesbury, a monk of Malmesbury Abbey in Wiltshire writing about 1125; Henry of Huntingdon, an archdeacon whose History of the English was written between about 1129 and 1154; and Geffrei Gaimar, who put English history into French verse in 1136 to 1137. The late layer is derivative except for Egil's saga, an Icelandic saga of about 1220 to 1240 whose hero, the poet Egil Skallagrímsson, is said to have fought for Æthelstan, and the Annals of Clonmacnoise, an Irish chronicle that survives only in an English translation of 1627; both carry independent but problematic traditions. The key point is that no source earlier than the twelfth century gives any locational information beyond the names themselves, the sea, Dublin, and the word "north" for Constantine's flight.

The poem (Anglo-Saxon Chronicle, manuscript A, under the year 937)

The other early witnesses

The Humber sentence

This sentence is the only independent statement placing the fleet in the Humber. Every later Humber statement can be derived from it, and on the stemma (the family tree of copies) accepted by Cavill (2011) and by Clare Downham, a historian of Viking-age Britain and Ireland at the University of Liverpool (2021), does derive from it: the second, Worcester-derived section of the Historia Regum (Arnold, vol. 2, p. 125, verbatim), the Chronicle of Melrose (a thirteenth-century chronicle from the Scottish border abbey, and the first text to combine John's Humber with Symeon's 615 ships), Roger of Wendover (a monk of St Albans, writing before 1236), and Pierre de Langtoft, a Yorkshire canon whose French verse chronicle of about 1300 has "A Bruneburge sur Humbre", the only medieval text to put the burh itself on the river. The strongest case for believing the sentence is Michael Wood's: that John was transmitting a lost York annal of about the year 1000. If such an annal existed, the sentence would be strong evidence. No independent witness to it has been found; the Durham tradition, which does rest on lost northern annals, gives 615 ships and the name Wendune and does not mention the Humber.

Three things weigh against the sentence itself. It opens an entry that otherwise summarises the Old English poem, and it departs from the poem by sending both kings ad naves, "to ships", where the poem sends Constantine north to his own country and gives ships only to the Norse. It is built from John's stock formula for fleet arrivals (a pagan king, valida cum classe, "with a strong fleet", ostium … fluminis ingreditur, "enters the mouth of the river"), the same frame he uses for the Danish king Swein's invasion of 1013 and for the landing of Tostig and Harald Hardrada of Norway in 1066 (citato cursu ostium Humbre fluminis intraverunt), where the Humber entry is genuine; in this formula only the river-name varies, and no earlier or independent source supplies the Humber for 937. And when John records raiders regaining their fleet he writes, more than twenty times in Cavill's count, that they "returned to their ships" (a verb in re-); for 937 he writes only that the kings were forced to flee ad naves, which Cavill (2022) reads as a signal that the fugitives did not regain the fleet that had brought them. On Cavill's reading, even if John is believed, the Humber is where the fleet entered rather than where the survivors re-embarked.

The later and legendary witnesses

The full dossier with each quotation, translation, edition and URL is in sources.md. Two points are stated here because summaries often blur them: the Welsh "Bellum Brune" is in the A-text of Annales Cambriae only, not in B or C; and the Beverley vow story (the tale, told at Beverley Minster in Yorkshire, that Æthelstan left his knife on the altar of St John of Beverley as a pledge before the campaign) belongs in its earliest form to the 934 Scottish war, with no Anlaf and no Brunanburh, as the historian Alice Law already noted in 1910. Both are checks against the editions rather than new findings.

The name

Old English Brūnan burh is "Brūna's stronghold": a personal name, Brūna, in its possessive form, plus burh, "stronghold". Old English formed possessives in two ways. Names ending in -a, like Brūna, belonged to the "weak" class and took -an (Brūnan); most other nouns and names were "strong" and took -es, the ancestor of the modern 's (Brunes-). The distinction survives in later spellings, so it tells you what kind of name the first element was. The Durham forms Brunnanwerc and Brunnanbyrig, Æthelweard's Brunandune, the Gaelic Duinbrunde, Gaimar's Bruneswerce and the Welsh Brune all agree on the first element and vary the second between fort, fortification, hill and simplex. Only one candidate, Bromborough, has recorded forms that descend regularly from Brūnan burh.

Forms from J. McN. Dodgson, The Place-Names of Cheshire 4 (English Place-Name Society vol. 47, 1972), pp. 237–8, as sequenced by Cavill, "The Place-Name Debate", Casebook (2011), p. 344. Bromborough is not in Domesday Book, the survey of England made for William the Conqueror in 1086.

Three points carry the philological case. The early forms exclude the old dictionary etymology "broom-fort" (Old English brōm, the broom plant), since there is no -m- until the assimilated Brumburg'. The personal name Brūna is rare: Briggs's 2021 index of personal names in English place-names lists four (as reported by Cavill 2022), and three of them (Bromborough, Brimstage, Brimston) are on the Wirral, plausibly one landholder. And, as Cavill (2023) argues, the strong form Brūn, whether the colour word "brown" or a river-name, gives a genitive Brunes- (Brownsall, Dorset, Bruneselle; Bruneshurst, Cheshire, 1275), not the weak Brunan- that every early witness has. That last point answers the two live objections at once: Michael Deakin's 2022 claim that Bromborough is "the brown fort" and so cannot descend from Brunanburh, and Andrew Breeze's Browney (Brune c. 1190, Brun c. 1195), which as a river-name would give Brunes-.

None of the rival sites has a recorded name that can descend from Brūnan burh. Brinsworth is Domesday Brynesford, "Brȳni's ford", with a persistent -s- genitive. Burghwallis is Domesday Burg, a bare burh with no Brun- element, and the River Went (Wenet, Wente) is a Celtic river-name that cannot yield Wendune or Weondune ("holy hill"). Burnswark is first recorded in 1541–2 (Burniswerkhill, Burnyswarke), six centuries after the battle, and every early form has a strong -s- genitive of burn (or of Birrens, the Roman fort nearby) plus Scots wark, "work, earthwork"; the -wark resembles Symeon's -werc and Gaimar's -werce, but nothing in the forms points to a weak noun Brūnan. Lanchester's own name means "the long fort" (Old English lang + ceaster, "Roman fort": the Roman Longovicium); Breeze's argument transfers the name from the river. Bourne in Lincolnshire and Bourn in Cambridgeshire go back to words for "spring" or "stream" (Old Norse brunnr, Old English burna), which are not weak nouns inflected in -an. The sceptic's reply, from Alistair Campbell in 1938 onward, is that a same-named place need not be the battle site and other Brunanburhs may have existed and been lost. That is true, and it is why the model gives the "unidentified site" a large share; it does not favour any named rival whose own name fails the test.

Dingesmere

The poem's second name, the water from which the Norse departed "over deep water" for Dublin, was long emended or left unexplained. Cavill, Harding and Judith Jesch, professor of Viking studies at Nottingham, read it in 2004 as þinges-mere, "the wetland of the Thing", the Thing being the assembly of the Norse settlers, whose site survives in the Wirral name Thingwall (Old Norse Þingvǫllr, "assembly field"), six kilometres from Bromborough, with the initial D- for þ- explained by Gaelic-Norse pronunciation as in Dingwall in Ross-shire, and mere in its sense of "marsh, wetland", the tidal margins of the Dee estuary. The reading is plausible but depends on three separate hypotheses (the etymology, the phonology, the sense of mere), each with alternatives on record: Downham allows "Dynne's mere" (a personal name attested in charter names such as Deanshanger); Livingston prefers Wallasey Pool on the Mersey to the Dee; Deakin takes dingesmere as a poetic term for the sea; Halloran reads it as a "sea of noise", the Solway; Breeze emends it to dingles-mere, "pool of the dell". The assessment of the Wirral finds commissioned by Wirral Council notes that the location of Dingesmere "is unknown, despite much discussion and conjecture". A parallel Thing-name exists at Tinwald (Old Norse þingvöllr) near Burnswark, and at Thingwall by the Mersey in Lancashire, so the argument favours the Wirral without being unique to it. The model treats it accordingly (an evidence factor between 1.5 and 5 for Bromborough, up to 3 for Burnswark). Independently of the etymology, the line shows that the Norse survivors re-embarked and sailed directly to Dublin.

Geography and logistics

Distances are great-circle sums (shortest distances over the earth's surface) along coast-hugging and Roman-road waypoints, and are lower bounds. Speeds: sea, 40 to 70 nautical miles a day for a fleet coasting by day (trials of replica Viking ships give 4 to 6 knots under sail; the Sea Stallion, a full-size replica longship built at Roskilde in Denmark, averaged about 5.5 knots on open-sea legs of its 2007 voyage from Roskilde to Dublin), 120 as an upper bound; land, 15 to 25 km a day for an army with baggage. Computed for this report by routes.py; the full table is routes_output.csv. Dave Capener of Wirral Archaeology gives 120 miles a day for laden ships, a figure cited by Cavill (2022) and used as the upper bound here; Cavill calls it "optimistic for a voyage around the north of Scotland in autumn".

The candidates, one by one

The evidence model

The model's purpose is to set the reasoning out in a form that can be inspected and altered; its numbers are not precise. Each candidate starts with a prior, the probability assigned to it before any evidence is weighed. Each of eight lines of evidence then has, for each candidate, a likelihood ratio: a factor saying how much more, or less, probable that piece of evidence is if the battle was at this candidate than if it was at the catch-all "unidentified site", which is fixed at 1. The factors are written as ranges. A ratio of 3 to 12 for Bromborough on the name means: the observation "the only regular descendant of the name is here" is three to twelve times as probable if the battle was fought at Bromborough as if it was fought at some unknown place whose name is lost. The final probability (the posterior) is the prior multiplied by all of a candidate's factors, scaled so that the candidates sum to 100%. The ranges are judgements, argued in the sections above; the point estimate uses the midpoint of each range on a multiplicative scale (its geometric midpoint); and the 20,000 re-runs (Monte-Carlo draws) pick each factor at random within its range, on the same scale, to show how much the ranking depends on any one judgement.

Two further outputs are more informative than the point estimate. The leave-one-out table shows what happens when a line of evidence is dropped altogether. Bromborough survives the loss of any single item except the name: without the name it falls to 24% and the unidentified site leads at 45%. Dropping the Humber tradition, which is the only item that favours the Yorkshire sites, raises Bromborough to 72%. The factor-to-tie list shows how much stronger each rival's total case would have to be to draw level: sixteen-fold for the Lancashire coast, eighteen-fold for the Went, thirty-one-fold for Burnswark, seventy-eight-fold for Lanchester.

The extremes need stating too. If every one of Bromborough's factors is set to its low end and every rival's to its high end, Bromborough drops to 2.5% and the Lancashire coast leads at 24%; at the opposite extreme it reaches 98%. Those extremes show the range of the assigned weights; a reader who gives the name argument no weight and the Humber sentence great weight will get a different ranking without any new evidence. At that joint extreme the Lancashire coast, a stretch of coast rather than a single site, comes first; in none of the 20,000 re-runs, where the weights vary independently, does any rival rank first. Every named rival fails at least one test that Bromborough passes, the name test in every case.

What would change the answer

Three findings would overturn the ranking. The first is a demonstration that Bromborough's early forms do not descend from Brūnan burh: if Deakin's "brown fort" reading were sustained against the Brunes- objection, the name factor would fall to about 1 and the result would become "unidentified, 45%; Bromborough, 24%". The second is an independent, pre-Conquest witness to a Humber landing: a northern annal behind John of Worcester, if one were found, would raise the Humber-tradition factor for the Yorkshire sites, and combined with a genuine Brun- fort-name near the Humber it could tie or overtake the Wirral. The third is a securely located, dated tenth-century battle assemblage anywhere else, or a charter or saint's-life text naming Brunanburh at another place.

The Wirral case would harden if the Bebington Heath material were re-surveyed on a grid with recorded findspots and dated deposits, and if the isotope programme announced by Cavill and Harding (2025) showed that a significant number of the iron objects originated north of the Iapetus suture, the geological join along the Anglo-Scottish border that separates the rocks and ores of Scotland from those of England. Their pilot tested eight objects, only one of them from Bebington Heath, and the authors themselves say that so few objects do not support conclusions (the limitations "are all too clear").

What this adds, and what it does not

Nothing in the primary evidence is new: no text, inscription or find has been added, and every argument about a candidate has an author in the published literature. What this report adds is method and a few verifications. It contains what appears to be the first explicit, reproducible probability model of the question, with a prior, evidence factors given as ranges, re-run intervals and leave-one-out sensitivity, in place of the qualitative "on balance" verdicts in print; the model shows that the case turns on one philological fact and quantifies how far. It computes routes and times for every candidate on a single, stated basis, where the figures in print are scattered (Capener's 120 miles a day, Cavill's "1,200 miles round Scotland"). Its source table was checked against letter-for-letter transcriptions of the manuscripts rather than repeated from secondary works: the superscript n in the Parker manuscript, the A-text-only Welsh entry, Æthelweard's actual wording, the first combination of Humber and 615 ships in Melrose, and the 934 date of the Beverley story, all of which are in the editions or the older literature. It also gives "somewhere else" an explicit share, which the debate has treated as an aside and which, on these weights, is the second-ranked hypothesis. The report re-analyses existing evidence; the percentages depend on the candidate list and the weights assigned; and the spread of the re-runs measures sensitivity to the weights and says nothing about their accuracy.

The conclusion agrees with the position argued by Cavill, Harding, Jesch, Coates and Downham and cautiously endorsed by Sarah Foot, Æthelstan's biographer, by Woolf, and by the Cambridge Old English Reader, a standard university textbook. This is not a consensus position: the Battlefields Trust, the charity for battlefield preservation, calls the site "unlocated", and Clarkson calls the identification unproven.

What could not be checked

Barbara Yorke's 2026 review of Cavill and Harding (Yorke is a historian of Anglo-Saxon England at the University of Winchester), Halloran's 2005 article, Wood's 2013 article, Deakin's 2022 and 2025 papers, Neil McGuigan's 2014 review of the Casebook in the Scottish Historical Review (McGuigan is a historian of northern Britain before 1300 at the University of St Andrews, and his review doubted the Wirral on strategic grounds: a Scottish king would have marched into a small peninsula hundreds of miles from his own territory) and the 2025 book itself were paywalled or unobtainable and are cited from abstracts, extracts and replies. Dodgson's Cheshire volume was not consulted directly. No manuscript images were examined beyond published transcriptions, and no fieldwork or archive search was possible. The evidence factors are judgements, and a specialist may reasonably set several of them differently; the code is provided so that they can. Nobody has yet re-run the model independently.

Where the argument stands (September 2026)

No formal consensus exists and no poll of specialists has been taken. To judge by reference works and reviews, opinion has moved towards the Wirral since 2011: the philological case has been re-argued in peer-reviewed venues and each rival answered in print, while Wood, Breeze and Halloran have published nothing new on the site since 2021 and Deakin's Cambridgeshire case (2025) is the only new rival. Clarkson warns that "widely accepted as Bromborough" risks becoming a "pseudo-fact". The University of Chester's press release of December 2025 ("paves the way for ending the debate") and the popular history website Medievalists.net ("may be solved") go further than the authors' own "likeliest location", and the archaeological claims of 2019–2021 have been scaled back to "consistent with" a camp.

Further textual argument is unlikely to settle the question. Physical evidence could: a gridded survey of Bebington Heath with every findspot recorded, isotope ratios showing that the iron came from Scotland, or a located tenth-century battle assemblage elsewhere. Until such evidence appears the site remains unidentified; what this report adds is a stated probability whose inputs can be inspected and changed.

References

Editions of the primary sources

1. Anglo-Saxon Chronicle, MSS A–F: letter-for-letter transcriptions by Tony Jebson, asc.jebbo.co.uk; A. Campbell, The Battle of Brunanburh (London, 1938); Dobbie, Anglo-Saxon Poetic Records VI (1942), pp. 16–20; Bately, The Anglo-Saxon Chronicle: MS A (1986), pp. 70–2.
2. Æthelweard, Chronicon: Petrie and Sharpe, Monumenta Historica Britannica (1848), p. 520; A. Campbell, The Chronicle of Æthelweard (1962).
3. Symeon of Durham, Opera omnia, ed. T. Arnold, Rolls Series 75 (1882–85), vol. 1 p. 76 (Libellus II.18), vol. 2 pp. 93, 125 (Historia Regum); D. Rollason, Libellus de exordio (Oxford Medieval Texts, 2000).
4. John of Worcester: B. Thorpe, Florentii Wigorniensis … Chronicon (1848), vol. 1, pp. 132, 226; Darlington and McGurk, The Chronicle of John of Worcester II (Oxford Medieval Texts, 1995), pp. 392–3, 602–3.
5. William of Malmesbury, Gesta Regum, ed. Stubbs, Rolls Series 90 (1887), vol. 1, pp. 142–4, 151–2; Mynors, Thomson and Winterbottom (Oxford Medieval Texts, 1998).
6. Henry of Huntingdon, Historia Anglorum, ed. Arnold, Rolls Series 74 (1879), pp. 159–60; Greenway (Oxford Medieval Texts, 1996).
7. Gaimar, Estoire des Engleis, ed. Hardy and Martin, Rolls Series 91 (1888), vol. 1, p. 147; I. Short (Oxford, 2009).
8. Annals of Ulster, ed. Mac Airt and Mac Niocaill (1983); CELT G100001A / T100001A. Annals of Clonmacnoise, ed. Murphy (1896), pp. 150–1. Annals of the Four Masters, ed. O'Donovan (1856); CELT G100005B. Chronicon Scotorum, CELT G100016.
9. Annales Cambriae: Williams ab Ithel, Rolls Series 20 (1860), p. 17; H. W. Gough-Cooper, Annales Cambriae: A, B and C in Parallel (Bangor, 2016). Brut y Tywysogion, Rolls Series (1860), p. 20.
10. Chronicle of the Kings of Alba: Skene, Chronicles of the Picts and Scots (1867), p. 9; B. T. Hudson, Scottish Historical Review 77 (1998), 129–61.
11. Egil's saga, ed. Sigurður Nordal, Íslenzk fornrit 2 (1933), chs. 50–55; text at heimskringla.no.
12. Chronicle of Melrose, ed. Stevenson (1835), p. 28; Roger of Wendover, ed. Coxe (1841), vol. 1, p. 392; Langtoft, ed. Wright, Rolls Series 47 (1866), vol. 1, p. 330.
13. Electronic Sawyer (the online catalogue of Anglo-Saxon charters), charters S 407, 425–428 (934) and S 432–439 (937), esawyer.lib.cam.ac.uk. Alia Miracula S. Johannis Beverlacensis, ed. Raine, Historians of the Church of York I, Rolls Series 71 (1879), pp. 294–7.
14. M. Livingston (ed.), The Battle of Brunanburh: A Casebook (Exeter, 2011): texts and translations of 53 sources.

Scholarship

15. P. Cavill, "The Place-Name Debate", in Livingston, Casebook (2011), pp. 327–49; pre-print nottingham.ac.uk.
16. P. Cavill, S. E. Harding and J. Jesch, "Revisiting Dingesmere", Journal of the English Place-Name Society 36 (2004), 25–38.
17. P. Cavill, "The Battle of Brunanburh in 937: Battlefield Despatches", in Harding, Griffiths and Royles (eds), In Search of Vikings (CRC Press, 2015), pp. 95–108; PDF.
18. P. Cavill, "Escaping from Brunanburh: The Evidence of John of Worcester", English Studies 103 (2022), doi:10.1080/0013838X.2022.2067708 (open access).
19. P. Cavill, "Spellings of Brunanburh Revisited", Northern History 59.2 (2022), 162–76, doi:10.1080/0078172X.2022.2088437.
20. P. Cavill, "The Battle of Brunanburh: The Yorkshire Hypothesis", English Studies 104 (2022/23), doi:10.1080/0013838X.2022.2154045.
21. P. Cavill, "The Battle of Brunanburh: The Lanchester Hypothesis", Neophilologus 107 (2023), 431–49, doi:10.1007/s11061-022-09758-1 (open access).
22. P. Cavill, "Bromborough, Brunanburh, and Dingesmere Revisited: A Response to Michael Deakin", Notes & Queries 70.3 (2023), 224–31, doi:10.1093/notesj/gjad077.
23. P. Cavill and S. E. Harding, Finding the Battle of Brunanburh (University of Chester Press, December 2025), ISBN 978-1-910481-15-8; reviewed by B. Yorke, Northern History (2026), doi:10.1080/0078172X.2026.2639392.
24. C. Downham, "A Wirral Location for the Battle of Brunanburh", Transactions of the Historic Society of Lancashire and Cheshire 170 (2021), 15–32, doi:10.3828/transactions.170.5.
25. PHC Services (P. Sherman, C. Downham, R. Philpott), The Wirral Archaeology "Search for the Battle of Brunanburh" Project: An Initial Assessment, for Wirral Borough Council (July 2020; redacted March 2022), wirral.gov.uk.
26. S. E. Harding et al., "Combining isotope ratios for provenancing Viking Age iron artefacts in the British Isles: a pilot study", RSC Advances 13 (2023), 31292–302, doi:10.1039/d3ra06367d.
27. J. McN. Dodgson, "The Background of Brunanburh", Saga-Book 14 (1957), 303–16; The Place-Names of Cheshire 4 (English Place-Name Society 47, 1972), pp. 237–41.
28. R. Coates, "A further snippet of evidence for Brunanburh = Bromborough", Notes & Queries 45 (1998), 288–90.
29. M. Wood, "Brunanburh Revisited", Saga-Book 20 (1980), 200–17; "Searching for Brunanburh: The Yorkshire Context of the 'Great War' of 937", Yorkshire Archaeological Journal 85 (2013), 138–59; "The Spelling of Brunanburh", Notes & Queries 64 (2017), 365–9.
30. K. Halloran, "The Brunanburh Campaign: A Reappraisal", Scottish Historical Review 84 (2005), 133–48; "The Identity of Etbrunnanwerc", Scottish Historical Review 89 (2010), 248–53; "Anlaf Guthfrithson at York: A Non-Existent Kingship?", Northern History 50 (2013), 180–5. Earlier: G. Neilson, "Brunanburh and Burnswork", Scottish Historical Review 7, no. 25 (October 1909), 37–55, with the rejoinders of J. Wilson (vol. 7, 1910, 212–14) and Alice Law, "Burnswork and Brunanburh" (vol. 7, no. 28, July 1910, 431–6, with a note by Neilson).
31. A. Breeze, "The Battle of Brunanburh and Cambridge, CCC, MS 183", Northern History 53 (2016), 138–45; British Battles 493–937: Mount Badon to Brunanburh (Anthem, 2020), pp. 121–8; reviews by S. Morillo, The Medieval Review 21.04.01 and T. Pickles, Northern History 60 (2023).
32. M. Deakin, "Bromborough, Brunanburh, and Dingesmere", Notes & Queries 69 (2022), 65–71; "Brunnanburh and Brunanburh: Two Names, One Place and a Battle in a Cambridgeshire Valley", Battalia 3 (2025), 49–70.
33. T. Clarkson, Strathclyde and the Anglo-Saxons in the Viking Age (Birlinn, 2014), ch. 5; Senchus blog, 9 Dec. 2014 and 28 May 2020.
34. A. Woolf, From Pictland to Alba, 789–1070 (Edinburgh, 2007), pp. 168–73.
35. S. Foot, Æthelstan: The First King of England (Yale, 2011); M. Livingston, Never Greater Slaughter (Osprey, 2021); T. Williams, review in British Archaeology (Sept.–Oct. 2021), p. 58.
36. A. C. Grant, "Locating the site of the Battle of Brunanburh (937)", v4 (2019–21), online; Á. Majorosi, "Where was Brunanburh?", Hadtudományi Szemle 16 (2023); A. Gautier, "La bataille de Brunanburh", Tabularia (2023), doi:10.4000/tabularia.6873.
37. M. Gelling and A. Cole, The Landscape of Place-Names (Stamford, 2000), on dūn; K. Briggs, An Index to the Personal Names in English Place-Names (English Place-Name Society, 2021).
38. C. E. Blunt, "The Coinage of Athelstan, 924–939", British Numismatic Journal 42 (1974), 35–160, at 89–92.
39. Battlefields Trust, "Brunanburh", battlefieldstrust.com; Wirral Archaeology CIC statement, 4 April 2024.
40. Sailing and marching rates: O. Crumlin-Pedersen, Viking-Age Ships and Shipbuilding in Hedeby/Haithabu and Schleswig (1997); Viking Ship Museum, Roskilde, trial-voyage reports for the Sea Stallion from Glendalough, Roskilde to Dublin 2007; D. W. Engels, Alexander the Great and the Logistics of the Macedonian Army (1978); J. Haldon, Warfare, State and Society in the Byzantine World (1999).

Files and code

All the materials are in one folder with the report: the dossiers of sources and cases, the evidence model, the route calculations, the map and chart builders, and the extracted texts. The scripts are pure Python with no dependencies and run from the code/ folder in this order:

python3 routes.py                 # sea and land distances and days -> routes_output.csv
python3 evidence_model.py 20000   # posterior, Monte-Carlo, leave-one-out -> evidence_model_output.json/.md
python3 make_map.py               # map.svg from data/british_isles_coast_50m.json
python3 make_chart.py             # chart.svg from evidence_model_output.json
python3 frag/make_tables.py       # HTML tables from the CSV and JSON

Report

brunanburh_report.html (this page)

Dossiers

dossiers/sources.md (the medieval sources, original language, editions, URLs); case_wirral.md; case_humber.md; case_north_and_others.md; debate_status.md (2011–2026)

Model

code/evidence_model.py with the likelihood table and its per-cell justifications; outputs in JSON and Markdown under data/

Routes

code/routes.py, routes_output.csv, make_map.py, and the Natural Earth coastline under data/

Texts

Extracted texts of the council assessment, Cavill 2011/2014/2022/2023, the Rolls Series editions and the annals, under texts/

To change a judgement, edit the corresponding range in EVIDENCE in evidence_model.py and re-run; the chart and tables regenerate from the JSON.