How old is English?



Critical remarks

Etymology is very tricky. Although Michael and Anthony worked together to write this article, they did not agree upon all proposed etymologies. Example: river Stour. As it is pronounced 'stower' Michael prefers the explanation 'stowing river', river which 'stows', pushes up the water. The Stour is a tidal river. The word could refer to the tides.

However, essential is that a serious attempt was made to interprete the ancient place names using the Germanic approach. It is possible that some of the proposed etymologies here will change as our insights steadily progress.

[1a] Bononia comes from *bona (plural = bonan) meaning 'wooden floor' (Bühne in German) and by extension granary or storage house in general. Bononia was the main port ot sail to Britain. No wonder it had storage facilities.

[1b] Sometimes we still have doubts or we change our mind. This is such a case: Dio wrote 'Boudovica', Tacitus 'Boudicea'. A more likely etymology could be a word derived from Old Germanic *bowanan, to dwell, inhabit, to live, to farm, a widely spread PIE word. Modern cognates in Dutch are boer = farmer, bouwer = builder , bouwen = to build, verbouwen = to cultivate; many also in German e.g. Bauer = farmer, etc. Also in modern English: neighbour = 'close farmer'. As 95% of people of the first century AD were farmers, one can understand that it was also a synonym for 'settler', 'inhabitant', 'people'. Bowed-wicca could mean '(important) lady (of the) people' or simply 'our lady'. As so many 'names' at the time, Boudicca was a title. Its modern equivalent is queen. I know of no female first names in those days which refer to 'victory', that is except goddesses. Although I first suggested 'humiliated woman', I now think that this is linguistically incorrect and that naming your queen as such could have been a serious offence.
Lollia Bodicca is an attested name for a British woman who married a Roman centurion. 'Lollia' is the female version of 'Lode-' and means loud (praised), famous. 'Lolle' is still used as a first name in Friesland. Bodicca was her proud title: lady Lolle.

The Celtic interpretation 'victorious woman' can be rejected as this title or nickname would have been valid only between her first and only victory and her defeat. A brief period of a few days. The use of 'Victoria' as a name came much later and is the female version of the original macho-male 'Victor'. But even 'Victor' appeared later, initially as a self attributed surname .

The word 'lady' itself means 'first woman of the loaf', as monitoring the quality of the home made bread (loaf) was trusted to the first lady of important families. The word 'queen' originally meant 'female, woman'.

[2] Today, we prefer the meaning derived from *than
= needle, sticking out, perhaps related to 'thin'. German: Tannenbaum (spruce)  So Thanet would mean: "what sticks out in the sea".

[3] We now think that Rutupiae means 'red tops'. The word 'red' would refer to the salt winning pans and the 'top' were the small hills whereon salt was won. Salt used to be boiled dry during the Iron Age. Many such hills were found in Britain, always next to the sea.

[4] A different and equally tempting etymology is "stowing river". The Stour is a typical tidal river. Occam's razor: "keep it simple".

[5]  We did extensive research on the subject and are now convinced that Dover means 'two shores' or 'spit'. Dover = 'two ofer' where 'ofer' means river (sea) bank.


Kentish Place Names
Were They Ever Celtic?

(as it was published in "Archaeologica Cantiana") by Michael Goormachtigh & Anthony Durham

All the place and river names in Kent that are commonly claimed to derive from Brythonic roots have at least as plausible Germanic roots. This weakens the common assertion that Kentish people were “Celtic” in Roman times and shifted to a Germanic language only after the Romans left.

For hundreds of years British history books have taught that the Anglo-Saxons arrived after the Romans left. In this traditional view, hordes of Germanic immigrants, led by Hengest and Horsa, stormed into a largely Celtic Britannia around 450 and forced the native Britons back into Wales and the west.

This version of history originated with chroniclers such as Bede, Gildas, and Nennius, who had political agendas to promote and were a bit hazy about geography. Modern scholars are sceptical, partly because contemporary Roman authors appear to have noticed no “ethnic cleansing” in Britain, but mainly because the implied movements of peoples would have been logistically impossible.

Nevertheless,most people still believe that throughout the Roman period the vernacular language in south-eastern England was some sort of Celtic dialect, commonly called Brythonic (or Brittonic) or proto-Welsh. This belief requires that the Germanic/Celtic linguistic and cultural boundary, which unquestionably existed on the Continent, did not continue across the English Channel.

An alternative view is that many people on both sides of the Channel at this time spoke some sort of Belgic language or proto-English (Goormachtigh, 2008). In support of this view, we now show that many place names in Kent which have in the past been traced to Celtic linguistic roots may in fact have Germanic roots.

Languages of the North-Western Roman Empire

When the Romans invaded Britain in ad 43, Aulus Plautius and his legions had to deal with a well-established people in Kent whom they called Cantii or Cantiaci. No ancient author has left a record of what language those early Kentish people spoke at home, which suggests that the Roman army had little difficulty in communicating with the locals.

Julius Caesar famously described Gaul as divided into three parts (Belgic, Gaulish, and Aquitanian), each with its own laws, customs, and language. He also wrote that coastal Britain was inhabited by tribes that had earlier migrated from ancient Belgium to seek booty. Besides the tribe explicitly called Belgae (with their capital at Venta, modern Winchester), there were Atrebates (with their capital at Calleva, modern Silchester), plus at least colonies of Suessiones, Armoricans, and Ambiani. Modern archaeology, especially finds of coins, suggests that several early kingdoms in Britain were in regular contact with their continental Belgic counterparts.

On the Continent there was a Germanic/Celtic linguistic boundary, with the precursors of modern Dutch and German to the north and east and Gaulish (later supplanted by French) to the south and west. In Roman times that language boundary hit the coast of the Fretum Gallicum (English Channel) well to the south of the modern boundary between Flemish and French, but exactly where is uncertain – indeed, much argued over.

Classical authors such as Tacitus and Strabo clearly understood that northerly tribes (in modern Luxembourg and Belgium) were Germanic, while to the south inscriptions on stones, coins, etc strongly suggest that Gaulish was a p-Celtic language. Caesar wrote that the dividing line between Belgae and Gauls was the river Seine. I n about ad 400, the Byzantine historian Zosimus unequivocally described Bononia (modern Boulogne) as “a city in lower Germany”.[1a]

Immediately opposite Kent on the Continent were the Morini tribe, who inhabited a marshy landscape akin to the poldered Netherlands and resisted the Roman army better than inland tribes. The Romans eventually imposed a civitas at Tarwanna (Terwaan in the Middle Ages, modern Thérouanne) in the drier, but forested, Pas de Calais region. Until the late Middle Ages, the whole territory was part of the county of Flanders, and its place names are generally traceable to a Germanic language. The vernacular language of the Morini was probably a precursor of the West Flemish dialect of Dutch, which is the surviving modern language closest to Kentish Old English.

On balance, therefore, it seems most likely that in Roman times the linguistic boundary on the Continent lay somewhere near the river Somme, which is well south of Kent. It was probably quite fuzzy, with merchants, soldiers, sailors, priests, and rulers moving around and using several languages. It is often suggested that some southern Belgae shifted to speaking Gaulish before they shifted to Latin and/or French. Also one must remember that “Celtic” culture was not necessarily the same as Celtic language.

The Traditional Idea of a Celtic Kent

The modern English language has taken strikingly little vocabulary or grammar from Celtic roots – far less, for example, than it has taken from Vikings, Indians, Australians, or Amerindians. Nevertheless, many people (and encyclopaedia entries and tourist brochures) still doggedly assert that in Roman times Kentish people generally spoke a Brythonic language. Their motives include prudent conservatism, romantic Celtic (and French) nationalism, and respect for the patient labours of generations of scholars.

Genetic evidence suggests that the Celtic/Germanic divide in the British population long pre-dates the Roman invasion (Oppenheimer, 2006). After the last ice age, founder populations arrived by two distinct, main routes: one up the western seaboard of the British Isles, the other down the Rhine and across the North Sea. In general, populations grow far more by natural increase than by migration. For early people, rivers and seas were highways rather than frontiers. However, one must remember that languages move with military and technological power as well as by accompanying the genes in human bodies.

Then there is the timescale of the alleged wipe-out of Brythonic from south-east England. The Roman army left Britain in ad 407. The political (if not linguistic) border between Anglo-Saxons and Romano-Celts between ad 497 and 517 probably lay on an arc from Bath via Wroxeter to York, where the warlord(s) behind stories of King Arthur fought key battles (Reno, 1996). By ad 880, English politics was mostly about Anglo-Saxons versus Danes, and by the Domesday Book in 1086 no trace remained in eastern England of any Celtic language.

One wonders how the Brythonic language could be squeezed out (except in its Welsh and Cornish heartlands) so much faster than persecuted languages elsewhere. Coptic lasted in Egypt for more than 1000 years after the Arab conquest. Welsh, Breton, and Occitan survive to this day. Gaulish took many centuries after Caesar’s De Bello Gallico to yield to Latin (Adams, 2003). For a real example of a language that vanished without trace one must look to Basque in southern France (Coates, 2007).

Rulers’ names (which were usually titles rather than personal names) are often cited as being Celtic, on the basis of spurious logic. For example, it is commonly asserted that Boudicca [1b] was a proto-Celtic feminine adjective *boudīka ,meaning 'victorious’, whereas the proto-English bowghede-wicca (modern English cognates bowed witch ), meaning humiliated woman, is a far better fit to her personal story of terrible abuse by the Romans.

That just leaves place names as the main defence of the traditional interpretation of history. Generations of etymologists (often amateur, mostly British) have scoured Welsh dictionaries for plausible cognates in English place names. Their assumed chronology (Celtic then Latin then Old English) is almost certainly valid for Wales, Cornwall, and Cumbria, but further east it may be wrong.

Place Names in Kent

The latest thinking about place name etymology is being assembled by the Survey of English Place Names, which is still incompletely published (but see Briggs, 2008). According to Coates (2007) it finds almost no names of Brythonic origin in many counties of south and east England, notably Bedfordshire, Cambridgeshire, Hertfordshire, Middlesex, Leicestershire, Sussex, and Rutland.

For place names in or near Kent, see fully published books such as Mills (2003) or Room (1988). For river names, Ekwall (1928) is still the definitive source. Readily accessible sources of information on the Internet include BBC Kent (2008), Maximus (2008), and Even (2008). For a well-reasoned exposition of traditional pro-Celtic thinking applied to Kent see Evans (1965).

Within these various sources, the following modern place names stand out as being allegedly (or at least possibly) derived directly from a Celtic origin:

Kent (and Canterbury), Chatham, Chevening, Crayford (etc)

Dartford (etc), London, Lyminge, Sarre, Thanet

So do these river names:

Cray, Darent, Dour, Limen, Medway, Stour, Thames

And the original Latin names of these places are usually said to have come from Celtic:

Dover, Lympne, Ospringe, Reculver, Richborough, Rochester, Springhead

English, Welsh, Latin, etc all arose from a common root language, proto-Indo-European (PIE), so it can be hard to say how or when a particular word moved from one stream of these languages into another. Note, however, that Welsh and Latin are closer together than either is to English, and that Germanic languages are rich in words that have no equivalent in other European languages. One should also never forget that the British Isles were inhabited by sophisticated peoples, with unknown languages, for thousands of years before the Romans came.

Place Names in the Thanet Area

Nowadays it is hard to see that the Isle of Thanet used to be a real island, separated from Kent by the Wantsum Channel, through which ships heading to and from London would sail in order to avoid the difficult rounding of the end of Thanet. A map of the ancient Thanet area based on the work of Cunliffe (2005) and others is reproduced by Even (2008). Much of the Kent coast is surrounded by ever-shifting sandbanks and the Wantsum Channel seems to have silted up as much from seaward as down the Stour river from Canterbury. No one appears to have claimed a Celtic root for Wantsum, but it has obvious cognates in Germanic languages, so that a likely proto-English form was something like want sich um , meaning literally 'winds itself around’.

Thanet is said to be named from Celtic tan-arth , 'bright island’ or 'beacon’, for which there is no archaeological evidence. Rivet and Smith (1979) discuss possible etymologies via Roman spellings, e.g. Tanatis . Thanet has long been confused with the river Tanat in North Wales. Around ad 833, the historian Nennius wrote that its British name was actually Ruoihm . Variant spellings, such as Ruoichim , hint at modern English reach ,discussed below. In fact, Thanet is very easy to explain as proto-English, something like modern Dutch ten ende , at the end .[2]

Reculver , at the northern end of the Wantsum channel, in Roman times was Regulbium. At least five different (not very convincing) Celtic etymologies have been proposed, among which a meaning of 'great headland’ is most commonly cited. A more plausible proto-English root is something like reg-gewelber . PIE *reg- means to move in a straight line, which passed into most Germanic languages, including the English sailing term reach , still much used on the Thames. For the second part of the name, alveus is tempting (Latin 'channel’, legal English 'riverbed’), but there are better cognates in Germanic languages meaning 'to make bow-shaped’: Dutch welven , German wölben . The position of the Roman fort hints that it name ought to have been 'ships turn into the channel here’, but it seems more likely that the Romans simply gave it the proto-English name for the whole Wantsum Channel. So Regulbium might be directly translated as curved reach.

Richborough has an obviously English type of name (in which rich nevertheless suggests reach ). In Roman times it was Rutupiae , whose etymology has been much, but inconclusively, discussed – for example, by Rivet and Smith (1979). Rutupiae was then a major Channel port, at the eastern end of the Wantsum channel, near the mouth of the river from Canterbury. The similarity between the ancient Isle of Thanet and the modern Isle of Wight makes it tempting to suggest that a proto-English version of Rutupiae (especially its alternative spellings Ritupis and Ruthubi) began with rith , Old English for 'small stream’, as in modern Ryde. However, once one starts thinking geographically, it becomes apparent that the ancient port’s key characteristic was to lie inside a fringe of sand and shingle banks, a bit like modern Poole harbour. Until quite recently, the patch of sea between Deal and the Goodwin Sands, known as The Downs, was possibly the busiest anchorage in the world, where sailing ships awaited suitable winds to continue their journey. Presumably in Roman times the sandbanks and the sheltered zone lay a little further north, off Rutupiae. In that case the key cognate word in modern English is road , where ships ride at anchor (with analogues in most Germanic languages). The second element ( tup or thub) may then be derived from oever or duver, discussed below under Dover. In that case the most likely original meaning of Rutupiae was something like sandbank anchorage .[3]

Sarre, a village on the edge of the channel, is said to be obscure in origin and meaning, possibly a pre-English name for the river Wantsum. In fact it has an obvious cognate in the German river Saar (French Sarre), possibly derived from a PIE root *ser-, meaning to flow.

Many river names in Eastern England are said to be Celtic or pre-Celtic (Ekwall, 1928) Those in Kent can more easily be explained by proto-English roots.

Cray is said to be Celtic, meaning 'rough, turbulent’ (Breeze, 1999) and to be cognate with the river Crai in Wales. This is manifest nonsense (since the modern river Cray is not turbulent and its lack of gradient suggests it never was) and as a possible cognate it is no better than modern Russian край (as in Ukraine), which means ' edge’. The best cognate is actually the English word crayer – a type of small merchant ship known in medieval times (and in several languages) for slowly carrying heavy cargoes. Among the most important of all the bulk cargoes carried by ancient ships was chalk: Latin creta, French craie, Old English cray, modern English crayon. As early as Roman times, chalk was in huge demand for marling (de-acidifying) new agricultural land and also for burning into lime for building. In Kent, chalk was obtained by digging downwards, into deneholes, often right beside or underneath the fields to be marled. But there is one particular area of historic Kent with a truly amazing concentration of deneholes – Stankey Wood, Joyden’s Wood, and Cavey’s Spring – right next to the river Cray. So many deneholes were there that

...shafts would have yielded a tonnage of chalk far in excess of that required for agricultural marl, even if they were worked intermittently over a long period ... probably excavated to provide building material and/or the raw element for lime burning. Le Gear (1979)

In other words, their industrial output would have needed to be exported and the only means of transporting such loads for long distances was by water. So it seems safe to suggest that Cray originally meant chalk .

Several places with names containing Crai or Craie (modern Cray) are mentioned in the Domesday Book, and Norman leases often referred to the marling of fields. Pliny the Elder wrote about British chalk mining around ad 70, and iron age archaeological traces have been found near this cluster of denenoles. But there is a further historical twist to add. There are no known etymological precursors for the words g rey and gray, which occur in most Germanic languages, including West Flemish where the word grisde includes one of the suffixes -te and -de used for emphasis in some languages. So perhaps Latin creta was derived from a Germanic phrase meaning 'grey stone’, containing an adjective like greyda .

Darent is said to be Celtic for 'river where oak-trees grow’ and to be cognate with the names of the Derwent, Darwen, and Dart rivers in the Celtic west of Britain. However, there is a better cognate in the river Trent in the east of England. Compare modern English trend , or Dutch drentelen,'to walk erratically’ or 'to walk back and forth’. So a putative original proto-English name would have referred to the river’s winding .

Dour – see below under Dover.

Limen (the old name for the river Rother, on which Lympne used to stand) is said to mean 'elm river’ or 'marshy river’, based on a Celtic word limen . There has been much discussion of Celtic words for elm trees in relation to river names (Ekwall, 1928). In fact there is an over-supply of alternative etymologies, based on the many word roots associated with wetness that begin with l , or the many rivers all over Europe with names like lee or lym . The most plausible proto-English root (discussed below under Portus Lemanis) simply means muddy.

Medway is said to be pre-Celtic in its way part (rather than the obvious Germanic root weg etc) and Celtic in its med part. This is strange, since the obvious meaning 'middle’ is accepted for the river Medina in the Isle of Wight, and between Dubris and Londinium the Medway is the midway river .

Stour is a common river name in England. A Celtic root has been suggested for the name of the Stour that runs through Canterbury, but there is no reason for preferring that to numerous Germanic words with a sense of strong, fierce, or violent. Possibly when it was first named the river had a strong current.[4]

Thames has received much attention from linguists, with a Celtic root meaning 'dark’ sometimes being incorrectly stated as certain. However, a proto-English etymology makes better sense, based on te (which has evolved into modern English at) plus something like eems ('estuary’, 'water body’, 'lake within the river’). Modern rivers with similar names include the Ems in Hampshire, the Emm Brook (which flows into the Loddon and thence into the Thames), and the Eem, Eems, Amstel, and Emme rivers on the Continent. The modern Belgian town of Temse probably retains that sense of a place on a river. So, according to Ekwall (1928), does Tempsford (Beds.), a village at the confluence of two English rivers remote from the Thames. The Ravenna Cosmography (Vatican codex) explicitly lists Tamese as the name of a place, not a river. So before Thames became applied to the river it probably started out as a place name (possibly even a generic one) meaning on the river.


Canterbury comes from the same root as Kent (discussed below). Its Roman name Durovernum Cantiacorum is also discussed below.

Chatham ought, at first glance, to be considered quintessentially Old English, like the closely similar Chartham, Chatterton, Chatwell, Chidham, and Cotham. However, place-name dictionaries prefer to link it with places further north-west, such as Cheadle and Cheetham, and say that the chat part is cognate with Welsh cêd, 'wood’, as in Betws-y-coed. (Early spellings did not include an h.) An alternative etymology is via cot, Old English for a cottage, shed, etc (compare dovecot), with cognates in many other Germanic languages. So Chatham could have originally meant village of small houses.

Chevening is said to come possibly from the Welsh cevn, 'ridge’, which is strange since its name is so similar to that of the Dutch coastal town Scheveningen. In Old English -ing commonly indicated a personal name. Cheven is cognate with various Indo-European words that mean 'lopsided’, 'crooked’, 'clumsy’, etc, such as Dutch scheef, or Old English scaff. So Chevening could have been derived from a local crookbacked person.

Crayford etc are related to the river Cray, discussed above. It is usually asserted that Crayford was the site of a bloody battle mentioned in the Anglo-Saxon Chronicle (written around ad 870) thus:

Anno 457. Her Hengest and Æsc fuhton wiþ Brettas in þære stowe þe is genemned Crecganford, and þær ofslogon feower þusend wera. þa forleton þa Brettas Centland, and mid micle ege flugon to Lundenbyrig

which is often assumed to be the same battle as Nennius described thus:

...Vortimer... valiantly fought against Hengist, Horsa, and his people ... Four times ... the second was upon the river Darent ...

The chroniclers were writing about events centuries before their time in places they had never visited and it is not credible that the Britons (Romano-Celts) could put an army into the field big enough to lose 4000 dead in Kent in ad 457. Part of the explanation may be that Crecganford is usually assumed to be a mis-spelling of Craegenford . However, crec is just an early version of the English word creek , while gan might be an early version of gang (meaning go ) but is more likely just to be a plural ending. So Crecganford may be a more generic word meaning 'creeks ford’ that should not necessarily be identified with Crayford.

Dartford (formerly Tarenteford) and Darenth come from the river Darent, discussed above.

Kent comes from the tribal region which Caesar called Cantium and which became Ceint after the Romans left. Words like kant, with a sense of 'edge’ or 'rim’, are found in many languages (e.g. cantle in English), often traceable back to Latin canthus for the iron rim of a wheel or PIE *kantho-. Maybe continental Celtic cultures were early adopters of iron technology and then donated the word root into many languages, including Slavic. Maybe not. Either way there is no reason to prefer a Brythonic root over a Latin or Germanic one for kant, but to early Channel-coast peoples Kent clearly meant something like edge region.

London is often described as etymologically mysterious, because the name has no obvious Celtic root (see Coates, 2002). In fact, it is easy to explain as proto-English. Landen in modern Dutch and German means simply 'lands’, which evolved from PIE *lendh-, meaning open land, as in modern English lawn or landscape. Note that its form is plural, which appears to have been picked up by the French in Londres (just as in Ypres), and that in about ad 700 the Ravenna Cosmography had duplicated entries for 'Landini’ and 'Londinium Augusta’. Ancient people sailing up the Thames estuary would have avoided its endless fringe of marshes until, at the limit of tidal floods (when sea level relative to London was as much as 4 metres lower than today), they would have reached a good, flat, relatively dry place to settle. There, in the sense of 'let’s make a home here’ they could have called the place simply lands.

Lympne is related to the river Limen discussed above and to Portus Lemanis, discussed below. Lyminge lies on an entirely different river, but presumably has the same root.

Rochester comes from Old English Hrofaescaestir, where Hrofa has been interpreted since Bede as a personal name, though it is also a small place in Iceland. In early English and Dutch, h was pronounced as a soft g, so there are multiple possible cognates in modern English, including grove, the grave element in Gravesend and Graveney (probably originally Old English graf, 'ditch’), and groove (which is very appropriate for the deep valley of the Medway). It is a delightful idea that Rochester may mean fort groovy.

Roman Place Names

Roman place names in Britain are often the earliest for which there is any written evidence, albeit often with variant spellings, grammatical endings, scribal errors, an old alphabet, and uncertain pronunciation to contend with. Many of their etymological difficulties disappear once one accepts that Germanic rather than Celtic roots preceded them (Scutt, 2008).

Dover is said to have taken its name from the Roman Portus Dubris, which is said to derive in turn from proto-Celtic dubras, 'waters’, cognate with dwfr in modern Welsh and deep in modern English. However, a Dutch speaker would point to te oever, 'at the shore’, while in the Isle of Wight there is a local word duver (pronounced duvver), meaning 'sand dunes’, whose pronunciation has mutated into Dover for one local street. In Roman times the river Dour had a mouth wide and deep enough to give Dover a significant harbour (which has long since silted up) that would, however, have been totally inadequate for landing an invading army. What made Dover a good entry port was not its 'waters’ but the gentle ascent up its valley through the white cliffs. Julius Caesar’s visits in 55 and 54 bc involved beach landings. To this day many small fishing vessels often find it easier to pull up on a shingle beach than to sail into a congested harbour. A simple Belgic interpretation of Portus Dubris, then, is beach port.[5]

Durobrivae Cantiacorum (or Durobrovum or Durobrivis) later became Rochester (discussed above), where the ro invites comparison with the later hrofa element. Celtic apologists have offered several possible Brythonic roots, including dourbruf, meaning 'swift stream’ or 'walled town with bridges’. For a proto-English interpretation of duro see below. The element briv (also in the ancient names for Dunstable and Water Newton) suggests 'swirling waters’, by analogy with modern English brew and its PIE root *bh-reuh-, 'strong (boiling) movements’, 'to cook’ (in water). So Durobrivae would have meant something like crossing over fast-flowing river.

Durolevum is a name on the Antonine Itinerary that has been linked with a ford across a river near Ospringe. Ancient travellers were probably connoisseurs of different types of river or wet ground, with a copious vocabulary to describe them. So the obvious guess for the element lev is a relatively small or slow river, related to Latin lavare or Welsh llif, thought to be from a PIE root *lou-, 'to wash’. If so, Durolevum would have meant something like crossing over small river.

Durovernum Cantiacorum (also spelled duro auernum ) later became Canterbury. The element duro is discussed below. For the element ver it is tempting to suggest the modern English word weir or river name Wear , both probably derived from PIE *uer- meaning to flow, but this does not explain the letter n . The Latin word vernus has been invoked in one of its meanings 'spring’, but curiously not in its other meaning of 'native’, for places called Vernometum. However, Durovernum has a modern English cognate – thoroughfare – that precisely fits ancient Canterbury’s topographical situation as both a destination for cargo boats travelling up the river Stour from Rutupiae and the crossroads where three routes from the coast merged into one route to London. Thoroughfare means 'to pass through’, and is etymologically related to Dutch doorvaren and various other Germanic forms such as deurvaeren and durchfahren . It is hard to find a concise modern English phrase that distils the meaning of the proto-English root behind Durovernum Cantiacorum better than Kentish crossing place .

Portus Lemanis invites comparison with Lacus Lemannus (modern Lake Geneva) where the local people were allegedly Celtic, and with liman , 'estuary’, from Russian лиман, derived from Byzantine Greek. An etymology that runs from PIE * lei - ('to be sticky’) to modern English loam etc makes best sense. A likely proto-English adjective leim would take an inflectional ending - en to match an accompanying noun such as hafe , so that Portus Lemanis most likely meant muddy harbour .

Noviomagus is a name on the Antonine Itinerary that has still not been conclusively identified with any archaeological site. Suggestions have included modern Crayford, Welling, and most recently West Wickham. Many places in the northern Roman empire had names that included the element magus, which appears to have meant 'market’, and for which an implausible Celtic origin has been proposed. Irish machaire (cognate with English marches) means 'flat, low-lying land’, and has passed into many Irish places as magh or maghera (Room, 1986), from which a proto-Celtic root * magos - has been deduced. The meaning is then supposed to have extended into 'field’, then 'fairground’, then figuratively into 'market’ (Falileyev, 2007). However, it seems far simpler to observe that the nearest Latin root is not mercatus (containing an awkward t like markt and merchant ) but mercor (meaning to carry on trade), or merces (meaning pay, reward, or wealth), which has passed into modern Italian as merce meaning goods. The clearly Germanic character of many magus places ( such as modern Nijmegen, Neumagen, Remagen, and Speyer) draws attention to the Germanic tendency to interconvert k and g sounds. An alternative possibility is that the m actually comes from a Germanic preposition um followed by a lightly sounded h in hage , meaning 'enclosure’ (modern English cognate hedge ). Then - omagus would have meant something like 'around fortified farm’. Either way there is no need to invoke a Celtic root to accept the obvious meaning of new market.

Vagniacae is a place on the Antonine Itinerary that has been identified as modern Springhead, “an archaeologist’s dream” (Wessex Archaeology, 2008), where:

at the head of the River Ebbsfleet was a pool fed by eight natural springs, an unusually large number that made the site sacred to the Celts, who began settling there around 100bc. They called the site Vagniacis ('the place of marshes’) ...

This etymology is based upon a speculative Celtic root *uagna-, for which there appears to be no corroborative evidence, so it seems reasonable to point out an alternative proto-English root wagen. So Vagniacae would mean something like 'cart people’. Early travellers would surely be less interested in the next marsh than in the next smithy or wainwright. To the Romans it probably meant waggon repair place.

Significance of the Duro Element

Besides the three distinct instances of the element duro in place names on the Roman road from Dover to London, there were at least nine more in other parts of south-east England. So it is useful to broaden the focus outside Kent in seeking the meaning of duro. The currently favoured explanation accepts the obvious translation of 'hard’ (which Celtic apologists claim for Welsh as well), so that duro indicates a fortified place.

A proto-English alternative might be like Old English duru (modern English word door and cognates in many languages), derived from a PIE root *ter- that had a sense 'to break through’ or 'to cross over’. Other possibilities in Germanic languages include the adverb through (door in Dutch, durch in German). These multiple possibilities are not mutually exclusive, because word meanings often get confused or conflated, and of course the modern English language is notorious for puns. They all converge on a suggestion that duro indicated something much less fortified than a Roman castrum or a British hill fort – more like a tollgate or checkpoint, a mildly guarded bridge, ford, ferry, or crossroads.

To modern eyes the Antonine Itinerary recalls a motorway map, which one might expect to be marked with péage (toll) points, or places to refuel or spend the night. The following table of known duro places makes it clear that they were not military strong points, capable of withstanding a serious siege. They were just natural places to control traffic, where the local political authority might well build a stockade nearby or plant a fort on a nearby hill, and where a local boatman or bridge warden might expect to be paid some money.

Modern place name Latin name Local topography Possible meaning
Cambridge Duroliponte river bridge river ferry
Canterbury Durovernum river crossing thoroughfare
Dorchester Durnovaria river roads ?
Dorchester-on-Thames ? river confluence ?
Dunstable Durocobrivis crossroads torrents?
Godmanchester Durovigutum river roads ways
Ospringe Durolevum river crossing brook
Romford? Durolitum road river ? river bank?
Rochester Durobrivae river crossing torrent
Towcester Lactodurum ford low place
Wanborough Durocornovium crossroads bends
Water Newton Durobrivae river crossing torrent

Readers knowledgeable about the Roman empire or philology may be able to widen the viewpoint outside Britain or to improve the link between local topographical features and cognate words in an early language. Here are some suggestions to start the process going.

Duroliponte is easy to pin down in modern Cambridge, where the Castle Mound overlooks Magdalene Bridge at a point where the river Cam bends and is least fringed by ancestral water meadows. The root li is very commonly associated with rivers. It is tempting to translate the ponte part as 'bridge’, but in fact 'quay’ or 'ferry’ is more likely for the topography and economics of ancient times. In modern Cambridge a punt is a boat.

Regarding Godmanchester, Green (1969) wrote that it was

...typical of a class of settlement which served the main roads as posting stations and administrative centres, and in many cases appear to have grown out of the civil settlements around early military forts .... a mansio or rest-house for official travellers using the imperial post service (curus publicus). Establishments of this type also had to provide facilities for changing relays of horses used for the carriages and gigs of the couriers, and for the storage of government goods in transit. In the later Roman empire mansiones tended to be used as police posts and collecting points for corn tax.

For the vigut in its name, conceivably way out and modern German weg are cognates.

In Lactodurum, the element lacto has usually been interpreted as 'milk’, but an alternative possibility is low place: compare laagte in modern Dutch. Wanborough stands out from the list as having no river nearby, but just a crossroads with a steep hill to climb. Both Saxon and Roman names may share a sense of 'bendy’: wan suggests PIE *wendh- (modern English wend or wind), while Roman corn is thought to come from PIE *ker-, 'curved’ (modern English cornice or crown).

Etymology rapidly descends into guesswork, not just for word roots but also for the declensions, local dialects, sound changes, and silly mistakes that have always moulded place names down the ages. Despite this huge caveat, this article shows that all the place names in Kent that have been claimed to be derived from Celtic roots have alternative Germanic roots that are at least as plausible. The evidence does not prove what language Kentish people spoke at home in Roman times, but it makes Belgic seem more likely than Brythonic.



The Internet provides copious information about all the specific words and place names mentioned here, plus translated texts of most ancient authors. Maximus (2008) and Even (2008) are particularly useful.

Etymologies were looked up mainly in the full version of the Oxford English Dictionary (1971), the American Heritage Dictionary ISBN 395090652, Etymologisch Woorden van het Nederlands (only A-R released yet) ISBN 9053566538 , Van Dale Etymologisch Woordenboek ISBN 9066483024 , Larousse Dictionnaire Etymologique ISBN 2037102283, and Internet lists of PIE roots.

Adams, J. M., 2003, Bilingualism and the Latin Language , Cambridge University Press, ISBN 052181714.

BBC Kent, 2008, 'Kent Place Names’, .

Breeze, A., 1999, 'The name of the river Cray’, Archaeologica Cantiana 118, 379-382.

Briggs, K., 2008, ' English Place-name Society – bibliographic data on survey volumes’,

Coates, R., 2002, 'A New Explanation of the Name of London’, Transactions of the Philological Society, 96, 203–29.

Coates, R., 2007, 'Invisible Britons: the View from Linguistics’, in Higham, N. J., ed., Britons in Anglo-Saxon England, 2007, Manchester University Press, 172–91. First published in 2004, available at http//

Cunliffe, B. C., 2005, Iron Age Communities in Britain: an Account of England, Scotland and Wales from the Seventh Century BC Until the Roman Conquest , 4th edition, Routledge, ISBN 0415347793.

Ekwall, E., 1928, English River Names , Oxford University Press, reprinted 1968.

Evans, J., 1965, 'Some Early Place and River Names in Kent’, (reproduced from Kent Archaeological Review ), accessed 5 September 2008 at .

Even, J.-C., 2008, Marc’heg an Avel, (French/Breton website with informative pages about English sites such as Richborough, Springhead, and Thanet.)

Falileyev, A, 2007, 'Dictionary of Continental Celtic Place Names’, http //:

Goormachtigh, M., 2008, 'How Old Is English?’ .

Green, M., 1969, Roman Godmanchester , accessed September 2008 at Le Gear, R. F, 1979, 'Deneholes Part 2’, Records of the Chelsea Speleological Society, volume 10.

Maximus, V., 2008, http// (A comprehensive source for almost everything about Roman Britannia)

Mills, A. D., 2003, Dictionary of English Place Names , Oxford University Press, ISBN 0198527586. (1991 edition inspected on paper, on-line edition checked 2008)

Oppenheimer, S., 2006, The Origins of the British , Constable and Robinson, ISBN 1845291581 .

Reno, F. D., 1996, The Historic King Arthur: Authenticating the Celtic Hero of Post-Roman Britain , McFarland, ISBN 0786402660.

Rivet, A. L. F. and Smith, C., 1979, The Place-Names of Roman Britain , Batsford, ISBN 0713420774. (For many relevant excerpts, see Even, 2008.)

Room, A., 1986, A Dictionary of Irish Place Names , Appletree Press, ISBN 0862811325.

Room, A., 1988, Dictionary of Place Names in the British Isles , Bloomsbury, ISBN 074750170X.

Scutt, W., 2008, 'Roman Place-Names’, .

Wessex Archaeology, 2008. 'Springhead – an Archaeologist’s Dream’