The Meaning of the Name Dover

315 HISTORICAL RESEARCH NOTES the meaning of the name dover The place name Dover is routinely described as derived from a Celtic word meaning ‘waters’. This article explains why that supposed etymology is almost certainly wrong and why Dover is far more likely to have been named from the shingle sandbanks around its ancient harbour entrance. A previous article suggested that Dover belongs among the 170-plus place names in England that contain the Old English word ofer ‘beach, bank, shore’ or its relatives (Goormachtigh and Durham 2009). Controversially, this requires that some word like Dover (and a Germanic language ancestral to Old English) already existed when the Romans were in Britain. The English language contains almost no early loanwords from Celtic, and the south-east of England contains almost no place names with a securely Celtic etymology (Coates 2002). So the idea that Dover is a rare Celtic survival tends to be cherished by anyone who still believes that everyone in early Roman Britannia spoke language(s) ancestral to Welsh. That is why the evidence needs to be examined in great detail. The earliest known mention of Dover is as ad Portum Dubris, occurring twice in the Antonine Itinerary, a document from about ad 300. As a single word, Dubris occurs once in the Notitia Dignitatum, once in the Tabula Peutingeriana, and twice in the Ravenna Cosmography though once mis-spelled Durbis. Dover is not mentioned in any known Roman-era inscriptions. Ptolemy, writing in about ad 140, did not mention Dover, though it is a weak candidate to be his καινος λιμην ‘new port’. The classic argument for a Celtic origin of Dubris was set out by Rivet and Smith (1979) in these words: ‘the British name was *Dubras ‘waters, stream’ (perhaps ‘streams’), plural of *dubro- ‘water’ (Welsh dwfr, dwr, Cornish dofer, dour, Breton dour; Old Irish dobur)’.1 They went on: ‘all records of the name, even those of the Antonine Itinerary set in a grammatical structure, show it as a locative plural in –is’, which is a grammatical mistake because Dubris was far more likely to be a genitive singular. Medieval Latin writers unhesitatingly used the genitive, not the locative, after ad portum. An apparent genitive plural occurs in classical HISTORICAL RESEARCH NOTES 316 Latin, where the Itinerarium Maritimum contains ad portum ritupium. And Julius Caesar wrote ad portum itium about his departure point for the invasion of 55 bc, where itium has been interpreted as a Latin adjective but looks suspiciously close to genitive plural ituum ‘departures’. It is not inherently important that classical scholars often misinterpret the noun cases in Latin place names (Arias 1987; Solopov 2005; Williams 2007) or even how many streams pass through Dover. However, this mistake does serve as a reminder that long-cherished ideas can be wrong, and it also points to a Latin nominative form of either *duber or *dubris. Most place names are duplexes (qualifier plus generic), including other post-Roman names such as Cantwareburh, Londonbyrig and Ythancaestir. So it is at least curious to see an apparent simplex in the earliest recorded Anglo-Saxon forms of Dofras, Dobrum, Doferum, Doferan, etc. In fact, plain ‘waters’ does not look ideal as the meaning of any place name, while ‘port of the waters’ looks distinctly odd. If a watery meaning for dubris is nevertheless considered acceptable, Celtic is not the only, or even the best, language family to supply it. The PIE2 root *dheub- has descendants in many languages. Examples in English include deep, dip, dive, and dimple. Dub (northern dialect or Scots for a dark or muddy pool) has cognates in other Germanic languages that are well represented in place names such as Dobbewatering. According to Pokorny (1959), the exact word dubris existed in Illyrian, which sounds remote but actually exemplifies a whole band of ancient languages that got squeezed out when Latin speech expanded towards Germanic and Celtic speech. Illyrian was also the first language of many of Rome’s best soldiers and sailors, including officers up to the level of emperor and part of the Britannia garrison between at least 105 and 400. And Julius Caesar overwintered in Illyria, near modern Dubrovnik, between his two trips to Dover. In short, the classic Celtic explanation looks unconvincing. Are there better alternatives? Dover as landmark? Old English ofer is a well-established descendant of proto-Germanic *obera- (Philippa, Debrabandere and Quak 2007), and has many cognates such as modern German Ufer. Its exact meaning is complex, and there may have been two distinct forms of the word. The ofer with a short O may be cognate with upper, hyper, super, etc., and may have evolved into modern English over. The ofer with a long vowel O could mean river bank or shore, and Ekwall (1960) thought that it applied primarily to a firm beach or gravelly shore. Goormachtigh and Durham (2009) took the link between Dover and ofer only as far as the ancient practice of grounding ships on tidal HISTORICAL RESEARCH NOTES 317 beaches. Another possibility to be considered is that Dover was named from its hugely visible notch in the white cliffs serving as a landmark. This was inspired by Gelling and Cole (2003), who showed that ofer, plus its variant spellings ufer, yfre, and ora, tended to occur at places that could serve as landmarks for ancient travellers on land or water. At least one ofer/ora name occurs near every port of any significance in Roman times, from Exeter in the west to Maidenhead high up the Thames. There is a cluster of more than twenty near the Isle of Wight, where even now boats navigate mostly by reference to lighthouses, seamarks, forts, church steeples, etc. And if a boat sailed from there, around Kent and up the Thames, it would pass ten more examples. The idea that Dover was an ofer/ora landmark was dismissed by Cole (1990) on the grounds that it did not have the perfect topography of a flat-topped ridge with a convex shoulder, like the end of an upturned canoe, which requires green hills not white cliffs. In fact, Dover has eight potential ofer/ora features, with its two valleys, plus Shakespeare Cliff to its left and St Margaret’s Bay to its right. Also, the crucial landmarks for ancient seafarers getting close to Dover must have been its two pharos lighthouse towers. A forthcoming book by Gavin Smith will argue that ora is often assoc-iated with early religious or secular enclosures on high ground, and only secondarily with landmarks. This fits the obvious linguistic parallels with Latin ora ‘limit’ and Greek ορος ‘mountain’. Also there is a word akin to Icelandic eyrr ‘gravelly bank’ that shows up in numerous place names such as Ayr and Elsinore on coasts sailed by Scandinavians. Especially in Orkney and Shetland, ayre is used for a tongue of shingle sticking into the sea, and the word passed from Germanic into Celtic languages. In short, the idea that Dover was named as a landmark is not really convincing. An ancient word *duber Dubris contains several phonemes that might be hard to write down in Latin, including D versus TH (etc.), U versus O, and B versus V or F. It seems best to take the spellings Dover and *duber at face value and focus on the main difficulty – that initial letter D. Many English and Irish place names have picked up a surprising initial letter, usually by transfer from a preceding word. Examples among ofer/ora places include several instances of The Nore (formerly atten ore) and of River (possibly from æt thære yfre), plus Hever (formerly Heanyfre). However, explaining Dubris as analogous to Old English æt ofer ‘at the shore’ or Dutch aan de oever ‘on the shore’, runs into the objection that definite articles and prepositions had not developed much by ad 300. The best explanation seems to be that the initial D had the force of HISTORICAL RESEARCH NOTES 318 two/twin/double, as in Latin duplex ‘two folds’, or in dubito ‘doubt’, which originated as ‘have two minds’. PIE *duo ‘two’ gave rise to the Old English prefix to- with a sense of division or separation, found in numerous words such as tofær ‘departure’ or tobriting ‘destruction’, which have all been superseded by modern English words beginning with dis-, di-, or de-. That D-to-T-to-D change, plus to’s propensity to join with verbs, suggests that any hypothetical *du‑obera as a precursor of *duber must have developed in a linguistic environment distinctly earlier than Old English. The critical question is whether translating *duber as ‘double bank’ or ‘separated beach’ fits the local topography at Dover and elsewhere. Inspiration for an answer came from Bruges, Belgium, where a historic boat-unloading basin called the Dijver or Dyver (with no certain etymology) had water flanked by two beaches. This suggested that the essence of a duver or dover in the Isle of Wight is to be a strip of land flanked by water on two sides. The Isle of Wight word duver Pope (1989) explained that in the Isle of Wight duver or dover is a generic local word for ‘a low-lying piece of land along the coast, subject to occasional inundation by the sea’. There are (or were until recently) four duvers, at St Helens, Ryde, Hamstead, and Seaview. Written records of a word variously spelled duver, dover, or duffer can be traced back only as far as 1774, but the physical objects being described existed much earlier. The common characteristic of all the Isle of Wight duvers is to be a sandbar or gravel spit partially closing off an estuary, so as to have sea on one side and river or marsh on the other. Simplified diagrams of their modern shapes are shown in Fig. 1.3 It is hard to map or diagram spits or sandbars at the mouths of estuaries because by their very nature they change over the years through natural forces or human interference. Indeed, both St Helens and Hamstead duvers have probably swapped sides of their estuaries in past centuries. In modern Ryde, many residents of Dover Street do not realise that it is named after ‘the dover’ along the shore, even though they always speak of ‘the duver’ at St Helens. The word dover can be pushed back before 1086 thanks to Domesday Book and two clear examples of the characteristic dover/duver topography on the English mainland. Doverhay (Domesday Doveri 4) is now just a street in Porlock, Somerset, but it lies at the apex of a huge alluvial fan of marsh leading towards a shingle ridge at the sea’s edge. Dovercourt (Domesday Druvercourt) is now a suburb of Harwich, Essex, but it occupies a tongue of land between Ramsey Creek and the North Sea that extends into one of the two spits across the mouth of the estuary between Harwich and Felixstowe. HISTORICAL RESEARCH NOTES 319 Doverow Hill in Gloucestershire ends in –ow because the western end (at position SO811053) of its ridge is a hoh ‘hock’ landform of the type recognised by Gelling and Cole (2003): when seen from the south by the river Frome, it is concave like the sole of a person lying face down. Its lower, westward extension (like the toe end of a foot) splits into two lower ridges (for example at SO809055) from which streams flow into formerly marshy ground below them, thereby plausibly fitting the dover/duver pattern. Big English dictionaries should really show duver/dover as an ordinary noun, not a proper name. In fact dictionaries already contain a related word: doab is defined as ‘the “tongue” or tract of land between two confluent rivers’. Doab is a recent loan-word into English from Hindi/Urdu, but its origin as do-ab ‘two waters’ in Sanskrit or PIE is obvious. A digression into river names Around Britain at least fifty places or rivers have names a bit like Dover, which are mostly attributed to proto-Celtic roots *dubo- ‘black, dark’, Fig. 1 Diagrams of the four duvers or dovers known in the Isle of Wight, at St Helens, Ryde, Hamstead, and Seaview. HISTORICAL RESEARCH NOTES 320 *dubno- ‘deep’, and *dubra- ‘water’, with descendants Welsh dwfr, Gaelic dobhar, dobur, Cornish dofer, and Breton dour. Watson (1926) listed more than twenty such names in Scotland, often with anglicised spellings that are hard to unscramble, where a derivation from Gaelic (rather than Norse, Scots, or Pictish) seems highly likely. Particularly noteworthy are Edradour and Edidovar, from Eadar da Dhobar ‘between two waters’. Analogous names exist in other areas thought to be formerly Celtic-speaking, including the rivers Douve, Dives, and Durance in France, Douro in Portugal, Deva in Spain, and Dora in Italy. Many people then seek to widen the bounds of ancient Celtic speech by pointing to the rivers Tauber in Germany, Tiber in Italy, and Douve in Belgium, plus places like Doeveren in the Netherlands and, of course, Portus Dubris. However, that argument can be extended much further afield: for example, linguistically to Russian дупло ‘hollow’ and дно ‘bottom’, Illyrian δυβρις ‘deep’, and Greek βαθος ‘depth’; or geographically to Lithuanian rivers like Dubysa, and to several Danish Dover places. Where does one stop? Even within Britain there are many rivers whose etymologies (and nearby places) might be fascinating to discuss, including Dawlish, Deveron, two Devons, Black Devon, Devy, Dewey, Divelish, Dore, Dove, Dovey/Dyfi, Dowlish, and Towy. However, one ends up dragging in lots of possible confusions (dove, devil, defile, the rivers Derwent, Dutch toeven ‘stay’, and French d’Evreux) without shedding much light. There was clearly an Anglo-Saxon word defer ‘river’, distinguished perhaps by Wessex regional dialect, perhaps by character as a chalk stream. The modern river Dever was Myceldefer (micel = ‘great’) in Anglo-Saxon charters from ad 900. Not far away, the modern river Candover was cendefer (cen = ‘keen’) from 824. The modern Deverill (upper reaches of the Wylye) was defereal in 968. Strung along these three rivers are (or were) sixteen places named from them (including Micheldever, Preston Candover, and Kingston Deverill), so it is reasonable to deduce that Andover (955 Andeferas) was also named from a defer whose name has been lost. Andoversford in Gloucestershire might perhaps be similar. A cendefrion in Cornwall was mentioned in an Anglo-Saxon charter of 967. Dofer was recorded as a component of river names at least as early as defer. It occurred more towards the Celtic west, but in duplex names with second elements that look Anglo-Saxon. The Doferic (recorded from 757, but now Shrawley Brook) joins the Severn by Holt Fleet (at SO815636). Not far away, the Doverdale (from 706, now Elmley Brook) meets the Salwarpe near Droitwich (SO868620). The Doferlan (from 940, now Doverle) meets the Little Avon at Berkeley (ST676992). The Doferburnan (from 977, now Knee Brook) meets the Stour near Tidmington (SP258377). The Dover Beck (from 1219) meets the Trent HISTORICAL RESEARCH NOTES 321 near Gunthorpe (SK694450). Doversgreen near Reigate, and The Dover near Angmering might conceivably fit this dofer pattern, but Condover and Wendover are just a distraction: contrary to what some books assert, they contain ofer after a first element that ends in D. All these dofer rivers appear to have river-islands or water-meadows at their mouths or confluences (or might have had in the past). Since rivers generally get named first at their mouths and only later in upstream tributaries (Nicolaisen 2001), one might argue that dofer meant duver. However, this is a very weak suggestion, because river mouths are where people choose to build harbours and sewage works, dig moats and gravel pits, and so on. Ekwall (1928, 1960) recognised real problems in squashing the early forms of defer and dofer names into the prevailing orthodoxy of a wholly Celtic Britannia. Förster (1941) devoted ten pages to invoking an archaic Welsh plural dyfr ‘rivers’. However, there is no need to explain defer and dofer as Celtic survivals picked up by Anglo-Saxon invaders, because they have an independent line of descent from PIE *dheub‑, through proto-Germanic *deupaz ‘deep’ and *deubjanan ‘to dive’ (Orel 2003), to Old English deop ‘deep’, dufan ‘to dive’, and dippan ‘to dip’. Place-name dictionaries envisage Old English words *dybb ‘pool’ and *dyfe/*dief/*def/*deof ‘hollow, valley’ as contributing to various place names. Perhaps dofer is just an alternative spelling of defer ‘river’, much as wold is a regional variant of weald. This kind of vowel change is easy to accommodate in the Germanic languages, which preserve a lot of PIE ablaut and umlaut (vowel changes such as sing/sang/sung/song, or long/length). Celtic scholars are puzzled that they cannot explain how their *dubron ‘water’ relates to recognised PIE roots that mean ‘deep’ and ‘dark’ (Delamarre 2003). Perhaps the best explanation is that there was an original word for ‘double bank’, which developed differently in different areas. For instance the Sanskrit word dvipa ‘sandbank, river island’ is etymologically ‘two waters’, with the ending ‑pa (from PIE *ab- ‘water’) that also shows up in Latin ripa ‘riverbank’, which led to the modern word river. So Dover has some previously unrecognised cognates that include Java and Socotra, plus the Maldives, which are essentially a series of sandbanks. While Sanskrit developed the original ‘double bank’ meaning towards ‘island’, Germanic, Baltic, and possibly Celtic moved towards ‘river’, while Celtic, plus possibly Illyrian and Greek, seem to have moved towards ‘water’ more generally. This is far from the only instance where an originally pan-European word survives as a general, lexical word in Insular Celtic, but survives mainly as a place-name element in England. Welsh cwm ‘valley’ versus English combe is the classic example. Less well known is Avon: that name belongs distinctively to rivers that were HISTORICAL RESEARCH NOTES 322 prehistoric trans-isthmus transport arteries (Sherratt 1996), but lost that importance before Celtic languages are thought to have crystallized. The way in which land sticks up from water in a duver/dover, while water goes down below land in a defer/dofer, need not be surprising. Many place-name elements form confusing doublets, one up, one down: for example dune/dene, dyke/ditch, moor/mere, grove/grave, and comb/ combe. Evidently what mattered in the creation of such landscape des-criptors was slope rather than absolute height. The word ‘bank’ itself originally meant a horizontal shelf, yet geographically it is meaningful only when adjacent to a slope and is frequently used in the plural ‘banks’. Figuratively ‘bank’ has developed to mean a storage pile. Linguistic discussions like this rarely lead to an overwhelmingly certain conclusion, so there remains a minute possibility that Dover was named from its river by people speaking a Celtic language. Yet it seems far more likely that Dover’s name came from its most salient characteristic as a port. How strong is the evidence that ancient Dover had a duver? Dover harbour Dover harbour is fundamentally a ria, a river valley with its lower portion submerged by rising sea level. The Dover Bronze Age Boat (Clark 2009), of about 1550 bc, was discovered a startling 6m below present ground level, at a spot just outside the medieval city wall, but 100m in from the modern sea front. The Boat seems to have been abandoned in a freshwater environment, suggesting that the sea was then some way off. Dover formerly had an extensive inner basin, whose approximate extent can be guessed from modern map contours. Layers of silt and the modern town now cover that basin, which was probably marshy rather than usable by shipping in Roman times. Somewhere closer to the sea, but still under modern streets and houses, lay the Roman harbour. In trying to work out what it looked like nearly 2,000 years ago one must bear in mind six main processes: - sea level rise, by 3m or more (Waddelove and Waddelove 1990). - cliff falls, totalling perhaps several hundred metres (Bates et al. 2011). - multiple earthquakes in the Dover Straits. - siltation, perhaps exacerbated by farming to feed the Roman navy. - diversion of water into sewers and drains instead of the river. - build-up of shingle spits and sandbanks. The last process is the key. Essentially the history of Dover harbour has been one long battle against shingle and silt, as engineers gradually learned how to deal with natural processes on that coastline, and discovered from bitter experience that their groins and jetties tended to make matters worse. Hasenson (1980) described how all the ports of south-east England tend HISTORICAL RESEARCH NOTES 323 to self-destruct. ‘The prevailing winds are south-westerly ... The waters tend to carry mud and shingle ... to fill in indentations and break down cliffs, rather than forming inlets, with the result that beaches get formed and river mouths blocked’. The modern Port of Dover that motorists see is entirely artificial and well to the east of earlier harbours. Before that, Dover harbour was well to the west, outside the river estuary and under the Archcliffe promontory. Written records of that harbour exist since about 1495, and chart its struggle against longshore drift of shingle. Earlier still (before 1400) the main harbour for small boats may have been on the east side, within bowshot of the castle walls (Jones 1907). The way that the river Dour then divided into two streams separated by a delta of shingle sounds very like the situation at Doverhay in Somerset. Domesday Book of 1086 mentioned a mill at the entrance to Dover harbour that was a nuisance to shipping. It is very clear that the focus of Roman and Saxon occupation was on the west side of the river, and it is highly likely that the Romans had the same experience as in later centuries, that their own building works caused a build-up of shingle that damaged their harbour (Bates et al. 2011). However, it is not easy to work out how that process actually played out. Rigold (1969) built upon the work of Amos and Wheeler (1929) and of Rahtz (1958) to deduce the Roman harbour layout. Unfortunately Rigold wrote just before Dover’s Roman forts were discovered (Philp 1977a, b). In 1992 a second Roman-era wooden structure, perhaps a jetty or mole, was discovered next to the Dover Boat, and in 1996 some rescue archaeology was performed under the Townwall Street petrol station. Early buildings all lay on the west bank of Dover’s estuary. The first Roman fort seems to have been started in about 115, but not finished before it was succeeded by a typical playing-card-shaped fort for about 600 men of the Classis Britannica. Most likely this was a naval logistics base for attempts to extend the Roman Empire northwards, 70 years or so after the initial conquest of Britannia. The Phase 3 fort was built much more strongly, on top of existing structures, probably about 270, while the Roman Empire was riven with internal conflict. It is hard to pull all these strands of information together into a single map, particularly as street layouts have changed so much since the Second World War. However, Fig. 2 shows the modern course of the river and modern street layout, derived from the authors’ own investigations on the ground, with the aid of Google Maps. Grey areas represent modern buildings or areas otherwise denied to pedestrians. Rigold (1969) deduced that the two almost right-angle bends of the modern river are a response to two sand-and-shingle spits. The latest thinking seems to be that his ‘new spit’ developed relatively late, as a HISTORICAL RESEARCH NOTES 324 response to Tudor harbour works. And his ‘old spit’ might be a remnant of that shingle delta between the two branches of the Dour. The present Castle Street was deliberately embanked in the 1830s over an area of marshy ground leading to a ford over the river, in a process that sounds strikingly similar to how the dover in Ryde, Isle of Wight, was built over. Fig. 2 Diagram of Dover, showing the locations of areas inaccessible to modern pedestrians (in grey), the modern river, Roman-era archaeological traces, and the locations of two spits across the river mouth deduced by Rigold (1969). HISTORICAL RESEARCH NOTES 325 Borehole evidence allowed Bates et al. (2011) to propose roughly where the eastern margin of Dover’s harbour area lay early and late in Roman times. They envisaged ‘the creation of saltmarshes and intertidal mudflats in the western part of the former harbour’ during the 2nd to 3rd centuries, which was relatively rapid and ‘may have been exacerbated by Roman engineering creating barriers at the harbour mouth’. So the Phase 3 fort may have been built out over mudflats caused by its Phase 2 precursor. The strongest reason for believing that Roman Dubris had at least one duver is that any harbour on its coastline was naturally doomed to have one. Huge quantities of sediment move around in the English Channel, a remnant of land that disappeared under the North Sea as recently as c.6000 bc, plus more recent erosion. On the other side of the Channel, Dover’s counterpart used to be at Wissant, which had a very similar history of a river that became clogged, but its port traffic moved further away, to Calais. Naming Portus Dubris after its estuary-mouth spit(s) and/or mudflats would just be a prosaic description of its most distinctive feature for mariners. Even today, careful pilotage among sandbanks is a vital survival skill in the Channel and the Thames estuary. The notorious Goodwin Sands now end about 8 km from historic Dover (Cloet, 1954), but they may have been more prominent in Roman times. The real question is why Dover, rather than a hundred other estuaries with similar features, was named from its *duber. The Romans had no need to bother with most ports that had a tricky entrance or needed expensive maintenance, and cargo ships could bypass Dover and head through the Wantsum Channel towards Canterbury or London. The answer must surely be that Dover was strategically important to the Roman Empire just as in later centuries – as a naval base and as a passenger ferry port. So the literal meaning in modern English of Portus Dubris is ‘port of the double bank’. acknowledgements Keith Parfitt of the Canterbury Archaeological Trust kindly communicated some of the latest archaeological findings relating to Dover’s topography at the relevant periods. anthony durham and michael goormachtigh references Amos, E.G.J. and Wheeler, R.E.M., 1929, ‘The Saxon-shore fortress at Dover’, Arch. J., 86, 47-58. Arias, G., 1987, Grammar in the Antonine Itinerary: a challenge to British archaeologists (in Cambridge University Library). HISTORICAL RESEARCH NOTES 326 Bates, M.R., Corke, B., Parfitt, K., and Whittaker, J.E., 2011, ‘A geoarchaeological approach to the evolution of the town and port of Dover: Prehistoric to Saxon periods’, Proc. Geol. Assoc., 122, 157-176. Clark, P. (ed.), 2009, The Dover Bronze Age Boat in context: society and water transport in prehistoric Europe, Oxbow Books, Oxford. Cloet, R.L., 1954, ‘Hydrographic analysis of the Goodwin Sands and the Brake Bank’, Geograph. J., 120, 203-215. Coates, R., 2002, ‘The significance of Celtic place-names in England’, in Filppula, M., Klemola, J. and Pitkänen, H. (eds), The Celtic Roots of English, University of Joensuu, pp. 47-85. Cole, A., 1990, ‘The origin, distribution and use of the place-name element ora and its relationship to the element ofer’, J. Eng. Place-name Soc., 23, 26-48. Delamarre, X., 2003, Dictionnaire de la langue gauloise, 2nd edition, Errance, Paris. Ekwall, E., 1928, English River Names, Oxford University Press, reprinted 1968. Ekwall, E., 1960, The Concise Oxford Dictionary of English Place Names, OUP, reprinted 1968. A later edition is Mills, A.D., 2003, Dictionary of English Place Names, OUP. Förster, M., 1941, ‘Der Flußname Themse und seine Sippe: Studien zur Anglisierung keltischer Eigennamen und zur Lautchronologie des Altbritischen’, Sitzunberichte der Bayerischen Akademie der Wissenschaften, Munich. Gelling, M. and Cole A., 2003, The Landscape of Place-Names, Shaun Tyas, Spalding. Goormachtigh, M. and Durham, A., 2009, ‘Kentish place names – were they ever Celtic?’, Archaeologia Cantiana, 129, 279-293. Hasenson, A., 1980, The history of Dover harbour, Aurum Special Editions, London. Jones, J.B., 1907, Dover. A Perambulation of the Town, Port, and Fortress. Dover Express. Nicolaisen, W.F.H., 2001, Scottish place-names, John Donald, Edinburgh. Oppenheimer, A., 2006, The origins of the British, Constable and Robinson, London. Orel, V., 2003, A handbook of Germanic etymology, Brill, Leiden. Philippa, M., Debrabandere, F. and Quak, A., 2007, Etymologisch Woorden van het Nederlands (Ke-R), Amsterdam University Press. Philp, B., 1977a, ‘The British evidence – Dover’, in Johnson, D.E., The Saxon Shore, CBA Research Reports 18, pp. 20-21. Philp, B., 1977b, The excavation of the Roman forts of the Classis Britannica at Dover 1970-1977. Pope, C.R., 1989, ‘The historic flora of Ryde Dover’, Proc. Isle of Wight Nat. Hist. Archaeol. Soc., 9, 33-40. Pokorny, J., 1959, Indogermanisches Etymologisches Woerterbuch, Francke, Bern. A revised edition is at http://dnghu.org/en/Indo-European%20etymological%20dictionary. Rahtz, P.A., 1958, ‘Dover: Stembrook and St. Martin-le-Grand’, Archaeologia Cantiana, 72, 111-137. Rigold, S.E., 1969, ‘The Roman Haven of Dover’, Arch. J., 126, 78-100. HISTORICAL RESEARCH NOTES 327 Rivet, A.L.F. and Smith, C., 1979, The Place-Names of Roman Britain’, Batsford, London. Sherratt, A., 1996, ‘Why Wessex? The Avon route and river transport in later British prehistory’, Oxford J. Archaeol., 15, 211-234. Solopov, A., 2005, ‘The imperial context of place-names in Roman Britain’, J. Eng. Place-name Soc., 37, 5-18. Waddelove, A.C. and Waddelove, E., 1990, ‘Archaeology and research into sea level during the Roman era: towards a methodology based on highest astronomical tide’, Britannia, 21, 253-266. Watson, W.J., 1926, The history of the place names of Scotland, Blackwood, Edinburgh. Williams, J., 2007, ‘New light on Latin in pre-conquest Britain’, Britannia, 38, 1-12. 1 Putting * in front of a word is a linguistic convention to indicate a hypothetical recon-struction, as distinct from an observed word. 2 PIE = proto-Indo-European, the reconstructed original language from which many modern languages began to diverge thousands of years ago. See Pokorny (1959). 3 For detailed information, readers should refer to aerial photos (for example on Google Maps), Ordnance Survey maps (for example at www.streetmap.com), and to various scenic photos (notably in the Shoreline Management Plan published online by the Isle of Wight County Council. 4 Ancient manuscripts often used u where modern practice (and this article) substitute a v.