The Roman Name of Canterbury and various misunderstandings

302 THE ROMAN NAME OF CANTERBURY AND LATER MISUNDERSTANDINGS anthony durham The Roman name of Canterbury is universally reported as Durovernum Cantiacorum, but is that correct? Here are the observed spellings in Roman sources: Duroaverno Cantiacorum (Ravenna Cosmography) Duroaverus (Peutinger Map) Δαρουερνον (Darwernon) (Ptolemy’s Geography 2, 3, 27) Durorverno, Durarveno, Durarvenno (Antonine Itinerary 2, 3, and 4) It is desirable to report exactly what ancient sources show, but modern typography always enforces slight compromises. Here v is used for a consonant and u for a vowel where ancient handwriting used just one letter u. The first two spellings can be checked carefully against original manuscripts, thanks to good photographs published by Richmond and Crawford (1949). Ptolemy’s original Greek is set out by Stückelberger and Graßhoff (2006). And the definitive version of the Antonine Itinerary1 is by Cuntz (1929). All this evidence was laid out by Rivet and Smith (1979). The usually quoted name spelling comes closest to that in the Ravenna Cosmography,2 which is often falsely accused of having especially corrupted manuscripts. But why does everyone drop out the letter A? An unbiased observer must surely wonder if the true original name of Canterbury was more like *Duroavernum. That naive question about one letter A unexpectedly developed into a huge enquiry into the early history of Kent. Almost all ancient names have several possible explanations, but only rarely do all viable candidates get open and equal treatment. Often a single etymology gets endorsed by some academic authority figure and is widely repeated thereafter. This article sets out to perform due diligence on all serious hypotheses to explain Canterbury’s ancient name. It turns out that there are at least seven. No explanation emerges as an unassailable winner, but there is a clear favourite: Duroavernum most likely meant something like ‘transport hub river island’ to Roman troops who had just arrived from similar locations on the Continent. The Alder-Swamp Theory Lyle (2002) confidently translated Durovernum Cantiacorum as ‘stronghold of the Cantiaci by the alder grove’, which is typical of how modern writers refer THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 303 to Roman Canterbury. For information about Roman place names, they rely on Rivet and Smith (1979), who took advice on language from the Celtic scholar Kenneth Jackson. His ‘Appendix on the Place Names’ to Rivet (1970) phrased its translation ‘Walled Town by the Alder Swamp’, slightly different from ‘the fort among the alders’ offered to Richmond and Crawford (1949) by Welsh-language expert Ifor Williams. That analysis collapses on close examination. Those language experts made an assumption without ever clearly stating it, or its component parts: (1) that almost all ancient British names were created by indigenous people; (2) that they used their own language; (3) that the closest living relative of that language is Welsh. Therefore, the experts were trying to answer an implicit question ‘What might the ancient name have meant in a precursor to medieval Welsh?’, when they should have been thinking more widely about all the people possibly in the picture when Canterbury received its ancient name. Scholars likened the -vern part of *Durovernum to Welsh gwern and Irish fern ‘alder’, because alder trees like wet ground, as occurs beside the river Great Stour at Canterbury. The argument usually stated (for example by Delamarre 2003) is that a proto-Celtic word *vern meant ‘swamp’ but later transferred to an archetypal tree that grows in swamps. Linguists have not found a convincing proto-Indo- European3 (PIE) origin for that sense of *vern, nor any attestation of gwern or fern earlier than 500 years after *Durovernum, but they take comfort from cognate words such as Occitan vernhe. There is nothing inherently wrong with alder trees contributing to place names, as Alresford (Essex) and Alderton (various counties) show, but current thinking about a Celtic *vern has missed a huge objection. The alder is a pioneer colonizer of disturbed soil and one of the first trees to produce pollen in spring. Therefore, plausibly, *vern- is related to Latin verno ‘to bloom’, from ver ‘spring’, which descends, along with cognates in other languages, from a PIE root *wesr. A sense of ‘vernal, early blooming’ may show up also in Latin verna ‘child born into slavery’ and in the Roman place name Vernilis (or Vernalis) reported by the Ravenna Cosmography in or near Cornwall. Vernodubrum was a Roman place name in the south of France, on the river Verdouble, which has a scenic gorge that was presumably once full of trees that like wet places. The other classic, but unconvincing, argument for a proto-Celtic word *vern ‘alder’ is that Endlicher’s Glossary4 of nominib[us] Gallicis ‘Gaulish names’ (Toorians 2008) translated Areverni into Latin as ante obsta. This has been argued to refer to front-rank warriors’ shields, because alder wood was good for making shields, and obsta looks like an imperative of Latin obsto ‘to stand in the way’. Julius Caesar faced tough opposition from Arverni people, whose name survives in the Auvergne region of France. Once again, a simpler explanation has been ignored: the prolific PIE root *wer- led to a range of words appropriate to an obstinate people, such as warn, warden, and warren in English. The Name Element Duro- Likening the name element Duro- to Latin durus and Irish dúr ‘hard’ plus Welsh dur ‘steel’ amounts essentially to guessing that the medieval walls of Canterbury ANTHONY DURHAM 304 had more substantial Roman or indigenous precursors than archaeology has managed to find. More seriously, Duro- is now thought to come from PIE *dhwer- ‘door’, which also led to Welsh dôr, Russian двор, Greek θύρα, Latin forum, etc., all words for pass-through zones. All 57 Duro- and durum places whose names have survived from Roman times across the Empire were surveyed by Durham (2021b), building on earlier work by Rivet (1980). Many explanations, including ‘fort’, can fit some Duro- and durum places, but only one characteristic fits all of them – to be a transport nexus, or what geographers often call a Central Place. Canterbury is the convergence point of six recognized Roman roads, plus several other roads and the river Great Stour. In that respect Duroliponte (Cambridge) and Durocatalaunum (Châlons-sur-Marne) are similar. Many post-Roman place names also fit that pattern, for example Dorchester-on-Thames and Dorking in England and some passes through hills in Scotland. France has dozens of place names whose medieval forms may have begun with Duro- or ended with durum. In Britain, Duro-named places seem to be a marker for the earliest phase of Roman occupation and road construction, since there are three instances in Kent, plus eight north of the Thames and into East Anglia. Rivet (1980) recognized the distribution of place names beginning with Duro- as very ‘Belgic’ (from the northwest of the Roman Empire), whereas those ending in durum tended more towards southern Gaul (Celtica). However, it might be unwise to focus on particular ethnic groups, because the elements that combined with Duro- or durum, on the Continent as well as in Britain, are remarkably diverse, with plausible relatives in multiple Indo-European language families. Other Published Guesses When Mortimer Wheeler (1932) summarized the history of Canterbury, he described the Roman spelling Durovernum as ‘certain’ and ‘plainly Celtic’, but then went on to stress that the earliest post-Roman spellings (on texts, coins, and charters) began with Dor-. To explain the name, he invoked a person called *Vernos. Few places in Roman Britain appear to be named from an individual other than an emperor, and positing an unattested personal name usually shows that a place-name researcher has run out of better ideas. If the Romans did indeed want to commemorate anyone at Canterbury, it could have been Quintus Laberius Durus, a senior Roman officer mentioned by Julius Caesar as killed during his 54 bc invasion. Goormachtigh and Durham (2009) pointed out that *Durovernum resembles Dutch doorvaren and German durchfahren ‘to pass through’. Linguists dislike that idea because words for ‘through’ and ‘door’ are probably not related by etymological descent, but only by popular re-interpretation, or folk etymology. Linguists also object that the second element of a name should not be a verb, but against that view are Lindisfarne and the people called Lindisfaran (Green 2012), containing relatives of Old English faran ‘to fare, to travel’. So far this article has noted three main published explanations of a name *Durovernum, and none looks convincing. Nor does a suggestion by Olmsted (2019, p. 330) that *vern might mean ‘wound’. It is time to look at name spellings containing a letter A. THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 305 *Duroarvernum? Maybe the observed spellings were trying to represent an original name *Duroarvernum. Every literate Roman would have heard of the Arverni people, who were major players in the politics of Gaul, who hired German mercenaries, and who interacted with Julius Caesar both before and after his excursions to Britain. Arverni might have seemed a logical name for similarly hostile natives encountered in Kent. The letter R is notoriously liable to shift around inside a word, by a process known as metathesis, and it could get left out or duplicated whenever a scribe copied an unfamiliar name. So maybe *Duroarvenum is a better guess, which comes close to the observed Durarven[n]o in the Antonine Itinerary. Arva was Latin for ‘cornfield, arable land’, from the adjective arvus ‘ploughed’, with relatives that include Welsh erw, Greek αρουρα, and Sanskrit urvarā, probably descended from PIE *arə-‘to plough’. Arva also had a specialized meaning, of outdoor sacred space or holy grove, like Latin lucus, Greek τέμενος, Celtic nemeton, Germanic thing, etc. The fratres arvales ‘Arval Brethren’ were a very ancient order of Roman priests whose elaborate rituals are known best from the Iguvine Tablets (Poultney 1959), written in one of the ancient Italian dialects that gave way to Latin. Arven, the precise spelling needed to match the Durarven[n]o of the Antonine Itinerary, is attested in those Tablets, meaning something like ‘on the sacred field’. Was Canterbury perhaps considered sacred5 before Roman times? One candidate site, under the ruins of St Augustine’s Abbey, lies next to the original Roman church of St Martin, where Æthelberht’s queen Bertha used to pray. It has a curious link to *arva ‘sanctuary’, because Bede reported (book 1, chapter 25) that Augustine and his companions approached Canterbury singing a litany whose words are recognizable as from the Catholic service of Rogation (Wood 1994), which preserves a tradition of the Arval Brethren. Bede provided several examples of rulers donating pagan sacred sites for Christians to found churches, in one case (4, 3) specifically calling it a nemu ‘grove’. He also quoted (1, 30) advice from Pope Gregory to Augustine that ‘the idol temples of [the English people] should by no means be destroyed, but only the idols in them’. Bede described (2, 13) the destruction of a pagan temple at Godmunddingaham in east Yorkshire, after its chief priest converted to Christianity and ‘defiled and destroyed the altars that he had himself consecrated’, about which Bede used Latin words fanum ‘shrine’, idolum ‘idol’, and septus ‘enclosure’. Any pre-Christian sanctuary would probably have been made of wood in the open air, and likely leave few archaeological traces. *Duroavenum? A fifth, and possibly the best, explanation of Canterbury’s ancient name lies in a neglected remark by John Leland (1540). He made a ‘probable conjecture’ that the attested Roman spellings derived from *Duravonum or *Duravennum, containing a former river name *Avona. Avon is now the name of up to 18 rivers in or near Britain (Durham 2021a), but ANTHONY DURHAM 306 there is no evidence that Canterbury’s river was ever called Avon (unless books are wrong in explaining Harbledown as derived from a personal name Herebeald). Canterbury’s river name of Great Stour goes back to at least as early as ad 605, when a charter mentioned Sturigao ‘Stur district’ at modern Sturry. The core meaning of Stour is uncertain, though the name is shared by at least five rivers in England, with many similar names on the Continent. Ekwall (1928) devoted five pages to the name Stour, which may be related to the Old English adjective stor ‘great, strong’. Welsh speakers like to claim that their word afon ‘river’ passed into English, to become the proper name Avon. However, the earliest attested use of afon known to the Welsh historical dictionary Geiriadur Prifysgol Cymru was centuries after it was being used in Old English, where afon (with oblique case spelling afene or avene) was a generic word for ‘river’ (Ekwall 1928, p. 22). For example, the river Cray, along Watling Street from Canterbury, was called Avene in charter S175 of ad 814, whose bounds were worked out by Hogg (1941). So it is conceivable that Canterbury’s river was once called the stor afon ‘great river’ and only the adjective has survived. Many water-related geographical names around Europe appear to derive from PIE roots *ap-/*ab- ‘water’, and/or *an- ‘to pour’, but their linguistics are complex. Ancient people may have distinguished ‘flowing water’ from ‘standing water’ (Witczak 2015), created words for ‘haven’ (Koch 2019), or used a related word with vowel U or O in some river (or river-bank) names in ancient Iberia and western Britain (Villar 2000). The prime example of a possible afon in Roman Britain was Abona, the name of both the estuary of the Bristol Avon and of a place upstream on it near Keynsham, occupying a site analogous to Canterbury. Another analogue of Canterbury was beside the Danube, at modern Silistra, on the border of Bulgaria with Romania (Ivanov 2000), upstream from a complex delta, and also an important centre of early Christianity. Its ancient name Durostorum suits a key crossing point of a strategic river. The Danube’s early name Ιστρος/Ister has never been convincingly explained, though it might be linked to Latin aestus ‘slow burn, tide’ and aestuarium ‘salt marsh, tidal inlet’, which was John Leland’s guess for the origin of the name Stour. The prominence of a letter R in names for Canterbury directs attention to three rivers in Britain now called Arrow or Yarrow (Ekwall 1928, pp. 16-17, 311, 478- 479) plus an unidentified Arewan in the Anglo-Saxon Chronicle. France has at least six similar river names (Avre, Auve, Erve, Orvanne, etc.), which French scholars trace back to Roman Arva, still the name of one river in Romania. These rivers are all very bendy, the opposite of arrow-straight, for which the explanation may be that both Latin arcus ‘bow’ and Germanic words for ‘arrow’ came from a single PIE root *arku ‘bow and arrow’. River Islands No single name analysis has emerged as an outright winner, but it does seem that Canterbury owed its original name to its river and that part of *Duroavernum (or similar) signified ‘riverside land’. In modern English place names, the ending THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 307 -ey (in Sheppey, Oxney, Bermondsey, etc) is often translated as ‘island’, but its characteristic usage in settlement names is for an area of raised ground in wet country (Gelling and Cole 2003). The equivalent place-name ending in Germany and its neighbours is au. Related dictionary words include eyot in English and Aue in German. The ancient precursor of these name elements was something like *awjō ‘river island, floodplain, water meadow’. That may be an extension of a root word that Beekes (1998) argued was possibly pre Indo-European, and is most directly represented by Latin aqua ‘water’ and Gothic ahwa ‘river’. In Old English, that word for ‘river’ became ea, which led to modern river names such as Yeo and Eau, but it is hard to distinguish from original eg, usually translated as ‘island’, but which Smith (1956) expanded to ‘land partly surrounded by water, a piece of dry ground in a fen, well-watered land’. Plenty of names based on *awjō were mentioned by Roman authors such as Pliny and Tacitus, including an island called Austeravia and peoples called Aviones, Chamavi, Frisiavones, Hilleviones, Ingaevones, and Istaevones. Most relevant to Canterbury was Batavia, a large river island in the Rhine delta, whose main settlement (near modern Nijmegen) was called Βαταουοδουρον by Ptolemy and Batavodurum by Tacitus. Batavians conscripted into the Roman army were valued as horsemen. One remaining puzzle is the prominence of –avern in the observed Roman spellings. Every Roman naval officer involved in the ad 43 landing would have heard of Lake Avernus, the water-filled crater of an ancient volcano near Naples, named from Greek αϝορνος ‘birdless’ because of the noxious gases it emitted. Lake Avernus was the inland basin of Portus Julius, the main Roman naval base created by the Emperor Augustus and named after his precursor, Julius Caesar. Its location some way inland hid it from anyone offshore, in much the same way as Canterbury, up the Stour estuary, was hidden behind the Isle of Thanet and accessible from the sea only via the Wantsum Channel. Julius Caesar’s Visit Ancient Canterbury was essentially a seaport. It makes sense as a substantial settlement only once substantial trade straight across the English Channel had built up, after Caesar had conquered Gaul and smashed the Veneti, who traded between Brittany and western Britain. Wheeler (1932) wrote that ‘Even until the 17th century coastal craft were able to penetrate to Fordwich, only two miles below the city. It may be, therefore, that the Roman town was placed as a convenient crossing and landing-point at the then navigable limit of the Stour.’ A map of the likely Roman coastline in Moses (2005) shows a strikingly wide Stour estuary. Caesar’s 54 bc beachhead camp6 has recently been identified at Ebbsfleet, on the Thanet side of Pegwell Bay. As described in de bello Gallico 5, 9, his troops marched 12 miles overnight before the Roman 7th legion captured a native fortification in woods uphill from a river. This almost certainly refers to Bigbury, 2 miles west of the modern city (Sparey-Green 2021). After that battle, Caesar forbade his troops to hunt down the defeated Britons because it was late in the day and he wished to use the remaining time munitioni castrorum ‘to fortify camps’ (plural). ANTHONY DURHAM 308 The Roman practice of building defensible overnight camps is amply documented by writers such as Frontinus, Polybius, and Vegetius. Modern comments tend to focus on physical structures (tents, ramparts, etc), but what really mattered was logistics. Catering for several thousand men and horses, who had marched overnight and then fought a battle, was a major logistical challenge. The Roman camps near Canterbury in 54 bc cannot have been mere hasty bivouacs, because Caesar described leaving many troops there for ten days. The importance of army logistics has been emphasized by authors with backgrounds as a soldier (Peddie 1987), a pilot (Selkirk 1995), a historian (Roth 1999), or a geologist (Kaye 2015). All these authors stress the scale of supplies that Romans needed to move to the site of any battle from their support base. In 54 bc many tons needed to move the 12 miles from Caesar’s Ebbsfleet beachhead camp to the vicinity of Canterbury. Transport on water is hugely more efficient than on land, and Caesar could readily send heavy supplies up the Stour, having brought 800 ships across the Channel. No major Roman camp near modern Canterbury has been identified archaeologically, but any camp must have been close to the river, partly to meet the huge need for fresh water for men and horses, and partly to use river channels as ready-made defensive ditches. Roman marching camps were often situated in the angle of a river confluence, and some developed into proper forts, such as those that became the cities of York and London. The obvious guess for that hypothetical camp at Canterbury is where the river Great Stour now divides into two distinct channels, to enclose a large river island, which was called Binnen ea ‘within the river’ in Charter S176 of ad 814, and later Binnewith Island (Potts 1943). Until recent work by the Canterbury Archaeological Trust is fully published, only one comment about occupation of the Stour floodplain and island(s) is safe to offer here: the passage of a large Roman army must have shifted the focus of settlement off the hills towards low ground beside the river. Caesar did not stay long at Canterbury, but he would have been sure to leave someone in charge there who was at least nominally Roman-friendly, most likely Amminus. Most importantly, everyone in Kent now knew that the Roman army could not be stopped by any hill-fort or tribal warband, but also that there was money to be made from trading across the Channel. And the Romans knew that the main route into the heart of Britain lay up the Thames estuary. De Bello Gallico supplies the closest approach to hard evidence available about the speech of pre-Roman Kent, in the names of local leaders (Cingetorix, Carvilius, Taximagulus, Segonax, and Lugotorix), other individuals (Cassivellaunus, Commius, Mandubracius, Adminius, and Verica) and peoples (Cenimagni, Segontiaci, Ancalites, Bibroci, Cassi, and Trinobantes). Such names are heavily Latinized from indigenous originals, but finding good parallels in medieval Welsh, Irish, English, or Dutch has been surprisingly hard. The Roman Conquest Everything about Caesar’s visit applies even more to the definitive Roman invasion of ad 43, when the army of Aulus Plautius was as multi-ethnic as the army of Wellington at Waterloo. Perhaps the best-known Roman troops were some THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 309 auxiliaries called κελτοι ‘Celts’ by the historian Cassius Dio. They won a key battle by swimming in full armour across a river, generally presumed to be the Medway. Most historians think that they were Batavians, who would later star in Agricola’s battle at Graupius mons, and may have been in Caesar’s personal bodyguard.7 Modern historians accept that after ad 300 a large proportion of the Roman army was recruited around the Empire’s frontiers on the Rhine and the Danube, but there has been a general reluctance to admit that the same was true much earlier. The Roman army in ad 43 may have had Italian commanders, Greek technicians, and Syrian archers, but much of its rank-and-file might nowadays be called Dutch, German, or Austrian. All the ad 43 Roman invasion troops came to Britain having been last stationed at places whose identity was tied up with water transport. The 2nd Legion came from Strasbourg, a town on a river island next to the Rhine. The 14th Legion came from Mainz, at a confluence lower down the Rhine where a Roman fleet was based. The 20th Legion came from near Dusseldorf, where the river Erft (formerly Arnapa) entered the Rhine. The 9th Legion may have come from a river confluence port at modern Sisak, near the Danube, following Aulus Plautius. In theory, legionaries were supposed to be Roman citizens, but that might include the sons of Roman fathers and local women. The Cantiaci At the start of this article, the second part of Duroaverno Cantiacorum seemed like a trivial issue. People whom Ptolemy called Καντιοι lived in a region that Caesar called Cantium and Strabo called Καντιον. The Cosmography merely added an adjectival ac to the basic name, much like ic in English, as a tribal label for the main settlement. However, probing deeper reveals problems. The common claim that Cantiaci were a unified tribe before Roman times is unjustified, notably because four regional reges ‘rulers’ and one dux ‘leader’ attacked Caesar’s beachhead camp in 54 bc. Also unjustified is the notion that Cantium was much the same as the later county of Kent. Ptolemy assigned to the Καντιοι three πολεις ‘cities’, at London, Canterbury and Richborough, but not Rochester, Dover, or Lympne. Caesar stated that Cantium was regio maritima omnis ‘entirely a maritime district’. An inscription ci cant found at Colchester suggests that a Roman civitas existed at Canterbury, but although civitas is often translated as ‘city’, ‘local government region’ is better. A promontory of Καντιον in Britain was mentioned by Diodorus of Sicily around Caesar’s date, and later by Ptolemy, most likely having been made known in the Mediterranean by navigators such as Pytheas, a Greek from Massalia (Marseilles), who visited Britain in the 300s bc. Ptolemy’s Geography places Καντιον on the same latitude as Richborough and slightly north of Canterbury, so it must be the North Foreland, near Broadstairs, and not the South Foreland, near Dover, as mistakenly suggested by Rivet and Smith (1979). The North Foreland was an important coastal landmark for navigators, because sailing ships could not manage its angle of turn between the Channel and the Thames estuary if the wind was in the wrong direction. That problem persisted until quite recently; hence the anchorage called The Downs, between the coast ANTHONY DURHAM 310 near Deal and the Goodwin Sands offshore, where sailing ships often used to wait for days on end for the wind to change. Once the Romans understood east Kent’s geography and tides, their shallow-draft vessels could take the Wantsum Channel, inside the Isle of Thanet, which stayed open for centuries but eventually silted up. Rivet and Smith (1979, pp. 279-300) discussed Cantium and Καντιον at length, and declared ‘corner land, land on the edge’ the most likely meaning. However, they ducked a difficult question: did *cant denote something angular or something rounded? They were influenced by the prevailing notion that ancient Britain was ‘Celtic’. To the limited extent that classical writers mentioned ethnicity, they never applied the term Celtic to the British Isles. To them, Celts were a range of peoples living north and west of the Mediterranean, whose common feature was to be alarmingly violent, not that they spoke a single type of Indo-European language. The term narrowed down to one part of Gaul, Celtica, distinct from Germania and Belgica, and gradually died out. Almost a millennium later it was resuscitated to describe the language family to which Welsh and Irish belong. Cant Breeze (2020) highlighted the use of cant in medieval Welsh to mean a stable wall or the cutting knife in front of a plough, and as part of Cantscaul, a battle site named in the Historia Brittonum of about ad 830. He did not remark that these examples of *cant, and many others, share a common core meaning of ‘subdivision’. That semantic notion is particularly clear in the English verb to cant, as in a heraldic canting badge. The Oxford English Dictionary reports that cant and its relatives were not common before 1600 and were hardly attested at all before 1400. Welsh cant and German Kant ‘edge, border, side’ are said to be loan words from Latin, where canthus (or cantus) ‘iron ring around a carriage wheel’ is sparsely attested and was described by the poet Quintilian as a barbarous word from Spain or Africa. Greek κανθός is better attested, with primary meaning ‘corner of the eye’ and secondary meaning ‘tyre of a wheel’, but Beekes (2010, pp. 633-637) called it ‘not well explained’ and ‘pre-Greek’, possibly having originated beyond Indo-European languages. Similar-looking words in Latin merely add to the puzzle. Cantherius could mean vine trellis, rafter or roof beam, low-grade equid (gelding, ass, mule, etc), or a gantry for suspending sick horses. Cantharis beetles are long and thin. There are real problems in explaining the deep etymologies of canto ‘to sing’, candeo ‘to shine’, canis ‘dog’, canna ‘reed, cane’, and canus ‘white’. Breeze (2020) was probably wrong in writing ‘Translations of Kent as ‘corner land, land on the edge’ must be jettisoned’. He was too much influenced by the way that (in the words of James 2020): ‘In the Celtic languages, *canto- has the senses of ‘a circumference, a boundary’ and ‘a division, a share of land’. However, James goes on: ‘in northern place-names the reference seems to be generally to ‘a corner, an oblique angle’.’ No easy answer comes from delving more widely into ancient names of overall form K-vowel-N-(T), where the initial K sound can show up in writing as C, CH, THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 311 or QU, and the final T sound can be written D or can disappear altogether. The vowel appears as A in Cantium and as E in Kent, but (arguably) it may take many other forms too. Names to be investigated include those now spelled Keynsham, Knutsford, Cynwyd, Ceint, Cound, and Cuntewellewang, as well as the more obvious Kenchester, Canterton, Candlet, and Canwick. The English words canton, hundred, and century are usually explained as derived from a root *kmt- that arose as ‘ten tens’ from *dekm- ‘ten’. But there is another range of words (canal, canon, cannon, cannabis, and qanat) that are usually linked to the tall wetland plants loosely known as ‘reeds’, that provided early human societies with measuring sticks, hollow tubes, and paper. All those items could have been involved in the subdivision of people and land, a very ancient process that is much older than Roman centuriation. None of this constitutes hard evidence, but it needs to be pursued because Breeze (2020) argued that ‘the fundamental sense of Celtic cant- is not ‘corner’ or ‘circle’, but ‘rim, edge; wall’’, which appears to be wrong. He regarded Cantabria in northern Spain as named from its inland mountains, without noting that its coastal plains lead up to a corner land inhabited in Roman times by Conisci and Concani. He noted that Cant Hill in Cornwall sticks up, but not that it lies next to wedgeshaped Cant Cove. And for Cantsfield and Cant Beck in Lancashire, he was not impressed that they lie near an acute-angled river confluence and an important travel nexus of Roman Britain. The hypothesis favoured here is that a conical or wedge shape gave rise to K-vowel-N-(T) words in many languages. They diverged and cross-fertilized to produce the wide range that is visible today, from anatomical features (κανθός ‘corner of the eye’, cunnus ‘vulva’), to landscape structures (cunette, chine), to animals with protruding muzzles (PIE *kwon- ‘hound’, Latin cuniculus ‘rabbit’). At places, sometimes an acute-angle shape is very obvious, as at Cuneo in Italy or Kentisbeare in Devon, but often the ancient landscape is too uncertain to be helpful. Early names regularly mentioned as possibly related to *cant include Cantuctune in King Alfred’s will and Canza in the Cosmography, probably now Cannington, in Somerset. The nearby Quantock hills recall Cantucuudu in a charter of ad 682. quantia, written on a gold coin minted around ad 600 at Quentovic, was a trading place with strong links to Britain, on the river Canche in France. Coantia in the Cosmography may be the modern river Kent flowing past Kendal in Cumbria, with Cantaventi somewhere nearby. Julius Caesar mentioned an angulus ‘angle’ that was ad Cantium, quo fere omnes ex Gallia naves appelluntur ‘onto Kent, towards which almost all ships from Gaul are directed’. It would also be logical if the people he called Ancalites lived around Thanet, with their name pulled towards Latin ancala ‘bend of the knee’. Some places have an obvious angle, whereas books currently endorse another explanation. For example, the ancient Conii (or Cynetes) people in the Algarve are usually linked with dogs, but they lived around Cape St Vincent, the south-western tip of Europe. Cantelupe Farm, near Cambridge, is said to come from chante-loup ‘howling wolf’, but sits in an angular river confluence. Early K-vowel-N-(T) names raise a host of questions. Why are they common where Roman roads crossed rivers? Examples include Cunetio, beside the river ANTHONY DURHAM 312 Kennet near Mildenhall in Wiltshire, with a namesake Kennett past Mildenhall in Suffolk. Or Kennetpans beside the Clyde in Scotland. Or Knettishall beside the Little Ouse in Suffolk. Why does an element spelled Chen- contribute to 200 names in Domesday Book? Did Cunia (Christchurch harbour behind Hengistbury Head) and Cunis (Ailsa Craig) owe their names to a wedge or cone shape? This whole Cant section may not convince everyone of the acute-angle hypothesis, but that is not its real purpose. It serves to show how many names based around *cant (or similar) existed in Britain and the Roman world. The name Kent does not belong uniquely to England’s south-eastern county. Kent/*cant is not the only ancient proper name about which historians may have jumped to wrong conclusions. The core issue now is to recognize that multiple locations can have similar characteristics that prompted people to give them similar names, albeit in different languages or dialects. The Adventus Saxonum The foundation myth of the English people and language declares that three boatloads of Germanic pagan warriors were invited to settle in the east of post- Roman Britain, as protection against northern invaders, but that they revolted against their Romano-British, Christian patrons. The key written account of those events is in this portion of the Anglo-Saxon Chronicle: ad 449. Hengest and Horsa ... landed in Britain in a place that is called Ipwinesfleet. ad 455. This year Hengest and Horsa fought with Wurtgern the king on the spot that is called Aylesford. His brother Horsa being there slain, Hengest afterwards took to the kingdom with his son Esc. ad 457. This year Hengest and Esc fought with the Britons on the spot that is called Crayford, and there slew four thousand men. The Britons then forsook the land of Kent, and in great consternation fled to London. This text comes from a modern English translation (from the Chronicle’s original Old English) that is readily available on the Internet because it was produced during the 1800s. Any serious investigation must go back to all available manuscripts and carefully check the spellings of all names. Then the Chronicle’s account must be cross-compared with all other sources, which include The Ruin of Britain by Gildas, the Ecclesiastical History of the English People by Bede, and the Historia Brittonum, which is commonly misattributed to Nennius. Information behind those early sources seems to have circulated among the literate monks of post-Roman Britain, who used three main languages – Latin, early English, and early Welsh. In addition, there were wide dialectal variations, plus interference from Irish, Pictish, Frankish, Danish, etc. So, there was plenty of scope for geographical misunderstandings by early chroniclers, even before modern historians injected their own prejudices. There is no single ‘smoking gun’ to prove beyond doubt where real events behind the Adventus Saxonum happened. Instead, there is a huge pile of weak evidence that must be analysed (much like an archaeological dig, a medical diagnosis, or a THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 313 police enquiry) in the spirit, if not the formal mathematics, of Bayesian inference. That includes careful examination of nearly 50 versions of proper names in almost as many ancient sources, and relating them to the landscape of 1,500 years ago. Another article (Durham 2022) lays out all that evidence, aimed primarily at a Yorkshire audience, allowing this article to focus on Kent. Translators and editors always face difficult decisions on how to render early proper names into modern English. In the text quoted above, they printed five geographical names that demand scrutiny: Ipwinesfleet, Aylesford, Crayford, Kent, and London. Ipwinesfleet is a fair representation of what the Chronicle’s manuscripts show for Hengist’s landing place: Ypwinesfleot, Hypwinesfleot, or Heopwines fleot. That is best explained as géap wine’s fleot, referring to the estuary of the Gypsey Race, next to Bridlington, east Yorkshire. Géap and fleot are well-established Old English words, related to modern gape and float, but wine ‘river valley’ has only recently been understood (Durham 2021c). Attribution to Ebbsfleet on Thanet is a mistake, caused by jumping to the conclusion that Hengist landed at the same place as both Caesar and Augustine. Ægeles þrep or Æglesþrep or Agælesþrep is where the manuscripts site that first battle, in ad 455. Aylesford is a bad translation, because the name element þrep is a variant of Old English þrop or þorp, most familiar spelled thorpe, meaning ‘farm, small village’, and it does not occur in Kent. A much better location is Haisthorpe, near Bridlington, which was called Aschiltorp in Domesday Book. Near that site is the Rudston Monolith, England’s tallest standing stone, which may explain the memorial to the slain Horsa mentioned by Bede and by the Historia Brittonum. Crecganford or Creacanford is where the manuscripts place the second battle, in ad 456 or 457. At first sight, Crayford is a plausible location for the battle, because it lies where the Roman road later known as Watling Street crossed the marshy estuary of the rivers Cray and Darent. The earthwork known as Faesten Dic (Hogg 1941) appears to guard that site’s southern flank, but its shape suggests that forces on the Kent side were defending against an attack from London, which does not suit the traditional narrative of Saxon invaders. The river Cray, responsible for the name Crayford, was the epicentre of a massive medieval industry of deneholes, vertical-shaft chalk mines. So, it is highly likely that the name Cray is related to craie, French for chalk, and to its Latin precursor creta. The history of Cray and its many relatives is too long to explain fully here, but its resemblance to Welsh crai ‘new, fresh’, noticed by Ekwall (1928) and widely repeated, is a red herring. Crai is cognate with the English word raw, and Breeze (1999) explained that in Welsh it mostly meant ‘raw, rough, crude’. Several historians (listed by Breeze) have pointed out that Crecganford or Creacanford evolving into Crayford does not fit the normal processes of linguistic evolution. As always with place names, this is not a conclusive argument, but it does allow another name explanation to claim first place: Crecg- or Creac- may represent the name element that shows up at dozens of modern places as Crick-, Creek-, Cruc-, Crayke, etc. Many lexical words are related, such as circus, chorus, and rick in English or crug in Welsh, but their key descendants are church and kirk, which Allcroft (1927, 1930) thoroughly explained as fundamentally meaning a structure on a circular base. ANTHONY DURHAM 314 The most likely site of that ‘Crayford’ battle is near Kirkham, in Yorkshire, at a crossing of the river Derwent. Kirkham Priory was founded more than 600 years after the battle and is now a ruin, but nearby at Crambeck are the remains of a prehistoric earthwork, inside which operated a major late-Roman pottery industry. A possible strategic context of that battle was explained by Storr (2018). ‘Kent’ actually appears as Centlond in manuscript A of the Chronicle, which becomes Kentland in the fractionally later manuscript B. One possible location for this Centlond might be the whole large wedge of land behind Flamborough Head and into the Yorkshire Wolds, an idea weakly supported by the modern place name Cans Dale. Possibly better is a wedge-shaped flood plain of the river Derwent, at whose apex sits Cant Bogs. On the map it is only a tiny name, of unknown history, but it is perfectly positioned to be where survivors might flee from a mass-casualty battle near Kirkham. ‘London’ appears in the Chronicle manuscripts as Lundenbyrg or Lundenbyrig, which make best sense as modern Londesborough, at a good fleeing distance (about 23km) from the Derwent. It is also near the apex of a large area of marshes, through which a small boat might have travelled to the Humber estuary by a channel functionally equivalent to the later Market Weighton Canal or the modern river Foulness. Storr (2018) suggested that one group of Germanic settlers in that area still supported the Romano-British establishment in their fight against Hengist’s immigrants. The Venerable Bede Bede’s Ecclesiastical History of the English People, completed in ad 731, reported the Adventus Saxonum in this passage (book 1, chapter 15), taken from a good modern English translation (Colgrave 1969) of Bede’s original Latin: ad 449 … the race of the Angles or Saxons, invited by Vortigern, came to Britain in three warships and by his command were granted a place of settlement in the eastern part of the island … First they fought against the enemy who attacked from the north and the Saxons won the victory … They came from three very powerful Germanic tribes, the Saxons, Angles, and Jutes. The people of Kent and the inhabitants of the Isle of Wight are of Jutish origin and also those opposite the Isle of Wight, that part of the kingdom of Wessex which is still today called the kingdom of the Jutes … the country of the Angles, that is the land between the kingdoms of the Jutes and the Saxons, which is called Angulus … Their first leaders are said to have been two brothers, Hengist and Horsa. Horsa was afterwards killed in battle by the Britons, and in the eastern part of Kent there is still a monument bearing his name. They were the sons of Wihtgisl, son of Witta, son of Wecta, son of Woden, from whose stock the royal families of many kingdoms claimed their descent. The core story of three ships landing in the east is taken straight from Gildas, and the supposed ancestors of Hengist have parallels in a complex of surviving documents known as the Anglian Collection of Royal Genealogies. However, THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 315 Bede was the first to mention three specific Germanic tribes and to supply forms of the personal names Vortigern, Hengist, and Horsa. Hengist and Horsa might be noms de guerre of real warriors or poetic inventions, but two brothers with horsey names certainly hint at early north-European warrior tales. In several Germanic languages Hengest or Hengist was an ordinary word for various types of horse; in Old English it formed compounds that mostly referred to ships. One Old High German text glossed hengist with cantarius, eunuchus, equus castratus ‘gelding’, though in Latin cantherius actually meant ‘ass, mule’. Was the similarity to Cantuarius mere coincidence or a deliberate poetic pun? Evidently Bede had sources of information that are now lost. He probably spent his entire life on Tyneside, and right on page one of his History acknowledged that much of his information came from Canterbury, through two named individuals, Albinus and Nothelm. Anything they could not find in written records came from seniorum traditione, literally ‘tradition of old people’, a phrase taken from the Bible, where it meant something more like ‘customary practices’ than ‘old wives’ tales’. Even if Bede had wanted to check his sources as carefully as a good modern reporter, he could have found out little about events in ad 449, six generations before the arrival of Augustine at Canterbury in ad 597. In fact, Bede knew remarkably little about life in Roman or early post-Roman Britannia. To some extent that was a natural consequence of the early church’s hostility to paganism. After Christianity was legalized in ad 313, more classical books were burned and statues destroyed than is commonly realized (Nixey 2017). Bede did not see key texts that modern historians rely on, such as Tacitus, Roman maps, and archaeological reports. He knew Caesar’s De Bello Gallico only indirectly via Orosius. A second major reason for Bede’s ignorance about the ad 400s was the prolonged cold spell after ad 540, when ash clouds from volcanoes in Iceland blocked sunlight. Crops failed, leading to famine and disease, including notably the Plague of Justinian and its after-shocks, which definitely reached Britain, as has been proved only recently. Mass mortality interrupted transmission of written records and helped to cause political and linguistic changes right across Europe. Darkness may even have inspired the apocalyptic tone of Gildas’ writing. A likely vehicle for the oral tradition that reached Bede, and that might have semi-independently reached the monks behind the Historia and the Chronicle, was itinerant entertainers, minstrel-poets who sometimes settled down as the court scop at a particular royal court. Minstrels were expected to remember their lords’ ancestors, whose deeds might grow more glorious with each retelling. Bede wrote so much about the kings of Kent that, with help from a few other sources, modern historians can reconstruct a reasonably full genealogy of that dynasty (see p. 316). This list, with 18 names over 282 years, is subject to many uncertainties due to joint reigns and much more, but it testifies to much political activity among the kingdoms of post-Roman Britain, with alliances often cemented by marriages. According to Bede, Hengist’s son Oeric was given the cognomen ‘nickname’ Oisc, probably the same as Æsc or Esc ‘ash’ (spear shaft) in the Chronicle passage above, so the Kentish dynasty were called Oiscingas. ANTHONY DURHAM 316 Jutes Gildas stated firmly that Germanic warriors were invited in ‘to repel the invasions of the northern nations’, and Bede agreed. Then how did Bede’s Iuti, commonly (mis)translated as Jutes, end up in the far south of England? No place name in Kent or the Isle of Wight or the adjacent mainland has been convincingly linked with them, though one red herring is often cited. A chronicle by medieval Winchester monks (for whom Bede was a revered source) reported the place where King William Rufus was killed, in ad 1100, as Nova Foresta, quæ linguâ Anglorum Ytene nuncupatur ‘New Forest, known in English as Ytene’. The most likely meaning of *ytene is ‘gorse’ (like Cornish eythin, Welsh eithin, and the Scottish river Ythan), because the Rufus Stone is at Canterton (Domesday Cantorton), an area rife with gorse bushes. As a consonant, Latin I sounded like Y in modern English and could be represented by G in Old English, as in modern Swedish. So Bede’s name Iuti looks suspiciously similar to the Γυτωνες (Gytōnes) from Gotland in the Baltic and the Γουται (Gutae) from modern Poland, mentioned in Ptolemy’s Geography. Also to people called Ytum in the Old English poem Widsith (Chambers 1912; Stenton 1940). Other authors called such people Goths. Roman and post-Roman Europe was full of footloose Germanic peoples, much given to migrating with their flocks, fighting with their neighbours, forming alliances, and serving as mercenary soldiers. Their diversity bewilders modern historians and may also have bewildered Bede, who supplied (5, 9) a list of more Germanic peoples from whom the Angles and Saxons in Britain derived: Fresones, Rugini, Danai, Hunni, Antiqui Saxones, and Boructuari. He also knew about Franks from whom Queen Bertha came to Kent. Yet nowhere did he mention Goths, unless they are hiding under the name Iuti. Germanic peoples encouraged to settle, as what the Romans called foederati, may explain some place names, including Suevi at Swaffham, Taifali at Tealby, and Vandals at Wandlebury. Goths invited in as protection (Old English mund) for the head of the Humber estuary may explain Godmunddingaham (Bede 2, 13), now Goodmanham, near Londesborough, mentioned above. They might have lived in that whole part of east Yorkshire, around the Roman road heading north from Hengist | Oeric | Octa | Eormenric | Æthelberht I | Eadbald | Eormenred | Ecgberht | Hlothhere | Eadric | Mul | Swæfheard | Swæfberht | Oswine | Wihtred | Alric | Eadberht | Æthelberht II THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 317 Petuaria (Brough). Storr (2018) interpreted some dykes in that area as evidence that ‘In an interesting display of tribal politics two dykes were in practice defending one group of Germanic settlers from another. … The geography suggests that (at least at this stage) the Sancton Germans remained loyal to the Romano-British’. Cantuarii In Bede’s Latin text it is abundantly clear that he intended Cantuarii to mean ‘people of Kent’, but just as more than one geographical name was based on *cant, more than one group of people might be called Cantuarii. Indeed, some place names far outside Kent, notably Canterton (Hants.) and Conderton (Worcs.), had early forms apparently containing *Cantware. It is usually claimed that these names arose from Kentish absentee landlords, but both places have topographical wedge shapes visible with 3D mapping tools that have only recently become available on the Internet. Would early farmers have seen that too? No one would accuse Bede of deliberate dishonesty. He was a Benedictine monk and a diligent scholar, but he was strongly partisan for Canterbury’s Roman version of Christianity. As Brooks (2000) put it, the core function of ‘the English origin myth [was] to assert the primacy of Kent among English dynasties’. De facto, Bede acted as a mouthpiece for king Æthelberht II of Kent, whose dynasty aspired to control a territory as large as the Roman Καντιοι (Kantioi) to whom Ptolemy had assigned three πολεις ‘cities’, at London as well as Canterbury and Richborough. Bede used Cantia for the territory of Kent, echoing the Roman forms for the promontory Καντιον/Cantium and the people Καντιοι/Cantiaci. Can his spelling Cantuarii be dismissed as showing trivial changes of compositional vowel and adjectival ending? Or does it indicate a freshly created name built around Germanic ware ‘dwellers’? Smith (1956) cited nine instances of ware in early English words or place names, noting that it was common in Kent. The element associated with ware generally referred to a relatively small feature of the landscape, rather than a political unit. That ending showed up in tribal names from north-west Germany reported by Roman authors, such as Angrivarii, Ampsivarii, and Chattuari. *Tentware people, deduced from the 1178 name Tentwardene of modern Tenterden, have been explained as absentee landlords living on the Isle of Thanet, but it would make better sense if *tent came from PIE *ten- ‘to stretch, to extend’, because Tenterden sits on a former promontory sticking into small-ship-navigable waters towards the former Isle of Oxney. The town’s website says that ‘It is difficult to believe today that Tenterden had access to the sea at Smallhythe … a centre for shipbuilding and produced wooden ships that were large for the time’. Words derived from *ten with a core sense of ‘sticking out’ exist in languages all over Europe; for example, English has tentacle and tendril, plus tetanus from Greek. Tenterden and Thanet are two sticking-out regions of Kent with a T-vowel- N-(T) pattern in their names. Many authors have looked at the name Thanet, and how it evolved over the centuries,8 but without properly understanding the geography involved. In particular, the Historia Brittonum (written in about ad 830 about events in ad 440) reported that [Vortigern] ‘handed over to them the island which in their language is called Thanet, and in British speech Ruoihm’. ANTHONY DURHAM 318 As usual, it is vital to go beyond the most readily available English translation, to check all spellings in the manuscripts: Tanet, Taneth, Teneth, or Tenech and Ruoihm, Roihin, or Ruim. Scholars seem not to have thought about east Yorkshire, where Ruoihm resembles the place names Rudston, Argam, and Reighton, plus the Welsh word rhwym ‘bonded’. And ‘Thanet’ might be the peninsula behind Flamborough Head, cut off from the mainland by Danes Dyke. People living around there could logically take a group name from the *cant wedge of land to be *Cantware. Their leaders, who founded the kingdom of Deira according to Storr (2018), may have displayed the ambition and military skill typical of recent immigrants. Victuarii A second category of ware people may be the Victuarii, one of Bede’s two categories of Iuti. He clearly thought they lived on the Isle of Wight, perhaps because of political events during his life. Kent was briefly controlled by Caedwalla of Wessex, who went on to conquer the Isle of Wight, slaughter its pagan inhabitants, and replace them with Christians from the mainland. In English place names, wiht seems to apply particularly to river bends (Smith 1956), with Canterbury’s Binnewiht island a prime example. The Isle of Wight sits in the bend of the Solent. Wiht got Latinized into Vectis, whose meaning as an ordinary Latin word was ‘carrying pole’, and hence ‘lever’ or ‘door bar’. A parallel on the Continent might be the Victuali (Doleža 2019). People living around any river bend might validly be called*wihtware. Dwellers in the Brough/Goodmanham/Londesborough area discussed above as possible Goths/Iuti may reasonably comprise Bede’s Victuarii. In support of that idea, there are place names now spelled Whitton, Whitgift, and Market Weighton near where the Humber estuary is joined by the rivers Ouse and Trent. And the south bank of the Humber, on the Lincolnshire side, is shaped uncannily like the Isle of Wight. In Germanic personal names, Wiht has generally been translated as ‘companion spirit, wight’, a notion that also suits the offshore location of the Isle of Wight (Durham 2011). An element Wiht- occurs repeatedly in names of Anglo-Saxon rulers, notably in a surviving complex of documents known as the Anglian Collection of Royal Genealogies, which may be related to the source that supplied Bede with names of ancestors of Hengist, such as Wihtgisl. Wihtred was the king of Kent while Bede was working on his History, and Centwine was king of Wessex while Bede was a child. What did Bede make of all those names among all the alliances, takeovers, and intermarriages of the English kingdoms? He would have faced biased assertions about the ethnic and religious characters of people and regions, as bad as any in modern politics, so it is understandable that he might have made a mistake in locating his Iuti. Conclusion This article started from a simple, naive question about one place name, not expected to be very hard to answer. By accident, it expanded into doing what rarely happens – trying to list all the potential explanations for that single ancient name. Usually, practical limitations of time and space dictate that one expert makes a THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 319 judgment, based on their years of experience, and only one final winning name makes it into print. Almost never does one name get so thoroughly researched as that of ancient Canterbury, nor do so many potential explanations get fully spelled out. In summary, Duroavernum looks preferable to Durovernum as the best guess for Canterbury’s Roman name. It probably originated among people who were culturally in tune with Continental, wetland Germanic speakers. One final suggestion may shock many people: the ‘Father of English History’ got something badly wrong. Any modern investigator of proper names will instantly sympathize with the Venerable Bede, because the available information is always fuzzy and inadequate, so mistakes are inevitable. Nevertheless, the cumulative evidence strongly suggests that Hengist landed in east Yorkshire, not Kent. Canterbury had a long feud with York over whose cathedral should control Christianity in England, and Canterbury won. But Yorkshire was where Anglo- Saxon England was born. references Allcroft, H., 1927, 1930, The Circle and the Cross, Macmillan, London. Beekes, R., 1998, ‘The origin of Latin aqua, and of *teuta people’. www.robertbeekes.nl/ wp-content/uploads/2019/08/b92.pdf Beekes, R., 2010, Etymological Dictionary of Greek, Brill, Leiden. Breeze, A., 1999, ‘The river Cray, Kent’, Archaeologia Cantiana, cxviii, 372-374. Breeze, A., 2020, ‘The place name Kent and Welsh Cant ‘Rim; Wall’’, Archaeologia Cantiana, cxli, 286-293. Brooks, N., 2000, ‘The English Origin Myth’, chapter 5 of Anglo-Saxon Myths: State and Church 400-1066, Hambledon Press, London. Chambers, R.W., 1912, Widsith: a Study in Old English Heroic Legend, CUP. Colgrave, B., 1969, English translation of Bede’s Latin Ecclesiastical History of the English People, republished as an Oxford World’s Classic in 1999 with an introduction and notes by J. McClure and R. Collins. Cuntz, O., 1929, Itineraria Romana, Teubner, Leipzig. Delamarre, X., 2003, Dictionnaire de la langue Gauloise, Errance, Paris. Doleža, S., 2019, ‘Constantine’s military operations against the Goths and the Sarmatians in 332 and 334’, Eirene. Studia Graeca et Latina, 55, 231-257. Durham, A., 2011, ‘The origin of the names Vectis and Wight’, Proceedings of the Isle of Wight Natural History and Archaeologial Society, 25, 93-97. Durham, A., 2021a, ‘Rivers called Avon’. www.romaneranames.uk/a/avon.pdf Durham, A., 2021b, ‘Duro- in early place names’. www.romaneranames.uk/d/duro.pdf Durham, A., 2021c, ‘Vindo- in Early Place Names’, Academia Letters, Article 4033. https://doi.org/10.20935/AL4033 Durham, A., 2022, ‘Hengist was a Yorkshireman’, East Yorkshire Historian 23, 88-105. Ekwall, E., 1928, English River Names, OUP, reprinted 1968. Gelling, M. and Cole, A., 2003, The Landscape of Place-Names (revised edition), Shaun Tyas, Stamford. Goormachtigh, M. and Durham, A., 2009, ‘Kentish place names – were they ever Celtic?’, Archaeologia Cantiana, cxxix, 279-293. Green, T., 2012, Britons and Anglo-Saxons in Lincolnshire AD 400-650, History of Lincolnshire Committee, Lincoln. Hogg, A.H.A., 1941, ‘Earthworks in Joydens Wood, Bexley, Kent’, Archaeologia Cantiana, liv, 10-27. ANTHONY DURHAM 320 Ivanov, R., 2000, ‘Romans on the Danube: Durostorum’, Athena Review, 2, 3. James, A., 2020, The Brittonic Language in the Old North. A Guide to the Place- Name Evidence: Volume 2 Guide to the Elements. https://spns.org.uk/wp-content/ uploads/2020/05/Alan_James_Brittonic_Language_in_the_Old_North_BLITON_ Volume_II_Dictionary_2020_Edition.pdf Kaye, S., 2015, ‘The Roman invasion of Britain, AD 43: riverine, wading and tidal studies place limits on the possible locations of the two-day river battle and beachhead’, Archaeologia Cantiana, cxxxvi, 227-240. Koch, J.T., 2019, ‘Rock art and Celto-Germanic vocabulary’. https://www.rockartscandin avia.com/images/articles/a19koch.pdf Leland, J., 1540, ‘Itinerary’ Notebooks edited by L. Toulmin Smith (1906-1910), 4, 69. Lyle, M., 2002, Canterbury: 2000 Years of History, Tempus, Stroud. Moses, C., 2005, ‘Interim report on Beaches At Risk’. www.sussex.ac.uk/geography/ researchprojects/BAR/publish/Interim%20Report.pdf Nixey, C., 2017, The Darkening Age: the Christian Destruction of the Classical World, Pan Books, London. Olmsted, G., 2019, The Gods of the Celts and the Indo-Europeans, revised version, available online, originally published by Innsbrucker Beitrage in 1994. Peddie, J., 1987, Invasion: Roman Conquest of Britain, Sutton Publishing, Stroud. Potts, R.U., 1943, ‘St. Mildred’s church, Canterbury: further notes on the site’, Archaeologia Cantiana, lvi, 19-22. Poultney, J.W., 1959, The Bronze Tablets of Iguvium, American Philological Association, Baltimore. Richmond, I.A. and Crawford, O.G.S., 1949, The British Section of the Ravenna Cosmography, Society of Antiquaries of London, Oxford. Rivet, A.L.F., 1970, ‘The British Section of the Antonine Itinerary’, Britannia, 1, 34-82. Rivet, A.L.F., 1980, ‘Celtic Names and Roman Places’, Britannia, 11, 1-19. Rivet, A.L.F. and Smith, C., 1979, The Place-Names of Roman Britain, Batsford, London. Roth, J.P., 1999, The Logistics of the Roman Army at War (264 bc - ad 235). Brill, Leiden. Selkirk, R., 1995, On the Trail of the Legions, Anglia Publishing, Ipswich. Smith, A.H., 1956, English Place-Name Elements (Parts 1 and 2), CUP. Sparey-Green, C., 2021, ‘Bigbury Camp and its associated earthworks: recent archaeological research’, Archaeologia Cantiana, cxli, 31-58. Stenton, F.M., 1940, ‘The historical bearing of place-name studies: the English occupation of southern Britain’, Transactions of the Royal Historical Society, 22, 1-22. Storr, J., 2018, King Arthur’s Wars, Helion and Compass, Warwick. Stückelberger, A. and Graßhoff, G., 2006, Klaudios Ptolemaios Handbuch de Geographie (volume 1 of 3), Schwabe Verlag, Basel. Toorians, L., 2008, ‘Endlicher’s Glossary, an attempt to write its history’. www.academia. edu/34275718/Toorians-Endlichers_Glossary.pdf Villar, F., 2000, Indoeuropeos y no indoeuropeos en la Hispania prerromana : las poblaciones y las lenguas prerromanas de Andalucía, Cataluña y Aragón según la información que nos proporciona la toponimia. Universidad de Salamanca. Watkins, C., 2011, The American Heritage Dictionary of Indo-European Roots (3rd edition), Houghton Mifflin, New York. Wheeler, R.E.M., 1932, ‘Towns’ in Kent (VCH), 3, 60-101. Witczak, K.T., 2015, ‘Flowing and stagnant water in Indo-European’, Linguistica Brunensia, 63, 7-19. Wood, I., 1994, ‘The mission of Augustine of Canterbury to the English’, Speculum, 69, 1-17. THE ROMAN NAME FOR CANTERBURY AND LATER MISUNDERSTANDINGS 321 endnotes 1 Arguing from the dates of composition of these classical sources is risky. They all drew on earlier sources, survive only as medieval copies, and were compiled long after the initial Roman conquests of Kent. The Antonine Itinerary supplies some of the hardest information about the locations of named places in Roman Britain, but it is actually quite late. 2 The Cosmography was produced by an unnamed monk in about ad 700 at the Italian city of Ravenna, while it was under Roman (Byzantine) control between periods of Germanic control (by Ostrogoths and Lombards). He seems to have conscientiously transcribed names off maps or itineraries that were already centuries old, dating from Roman imperial times. Medieval Celtic languages offer limited help for understanding its name spellings. 3 The proto-Indo-European language is well explained for non-specialist English speakers by Watkins (2011), whose typography and definitions are generally followed here. An asterisk before a word indicates a theoretical reconstruction that has not actually been observed. 4 Endlicher’s Glossary probably originated in the late 700s near the border of France and Belgium as an attempt to explain some regional place-name elements that were no longer words in mainstream use. The ‘Gaulish’ referred to was probably not a single language closely related to early Welsh, because around that date in that area the late Latin that became French was supplanting local Germanic dialects. 5 There seems to be no hard evidence for an ancient sacred spring or holy well in the Canterbury area, although several streams ran into the river Stour. Ancient water-source deities often served to protect a valuable community resource. 6 See https://archaeology.co.uk/articles/features/ebbsfleet-54-bc-searching-for-the-launch-siteof- caesars-british-invasions.htm. 7 Caesar (Gallic War 7, 13) mentioned ‘about 400 German horsemen, whom he had determined, at the beginning, to keep with himself’. See also Smith (2021) ‘The Germanic Horse Guards of the Julio-Claudian Emperors’ https://retrospectjournal.com/2021/02/07. 8 See particularly Rivet and Smith (1979, pp. 468469) and Green (2015) www.caitlingreen. org/2015/04/thanet-tanit-and-the-phoenicians.html. Linguistic parallels with Celtic words for ‘fiery’ or with Phoenician words are probably red herrings. ANTHONY DURHAM 322