Explanatory alphabetical list of Romano-British place-names


Part 9:  N → R



[For information as to which names are included in this list and which not, and an explanation as to the abbreviations employed, click on Alphabetical List menu provided above]




(Celt) (Rav) (Ptol)     (Mod)
(Navi)magno Navimago (Regentium) Noeomagus    

The Trundle



          (West Sussex) 

Note 1:

Rivet and Smith see this name as a duplicate of Noviomagno. They see the first part of the name as being derived from hypothetical Celtic novio, taken by some to mean 'new', though Watson and Holder took the view that the word had additional meanings such as 'fresh, lively, vigorous'. The second element of the name is thought to be derived from hypothetical British magos, taken to have meant 'field, plain' originally and to have taken on the meaning 'market' later. The name as a whole is thus taken to mean 'new place' or 'new market'.


But the mago element of Navimago Regentium will have been magno originally, this being an old-style compound in the hill-letters m and n2, where the gn element means ‘steep hill’. It will have been the name of the hillfort known as The Trundle, this being apparently the hillfort nearest to Chichester. The Navi element, which presumably should be Novio, meaning  ‘new’, was probably added by the Romans when they moved the inhabitants of The Trundle to a new settlement on the site of modern Chichester.  The Regentium element will have been added to Navimagno when the local king died and his kingdom became a civitas of Roman Britain. It is not entirely clear whether the Regni were a people separate from the Atrebates at the date when The Trundle was seized from its previous occupants, who used the hill-letter m. But we see this seizure of control at three other hillforts on the South Downs, namely Cardadonecon (Ardaoneon) at Harting Beacon, Carnis (Armis) at Butser Hill and Claducendum (Clausentum) at Old Winchester Hill. The hill-letter n2 is the latest hill-letter in all three names, as it is in Navimagno.  At some point the second n of (Navi)magno was dropped or lost and so the full name became Navimago Regentium. The forms given in Ravenna and by Ptolemy refer to Romano-British Chichester rather than to The Trundle, as does the Regno appearing in Iter VII of the Antonine Itinerary.  


Note 2:

Example of omission of letter - n 


[The entry for Navimago Regentium was last modified on 22 September 2019]




(Celt) (Rav)       (Mod)
Navione Navione (106)       Brough-on-Noe
(a river)         (Derbyshire)



Williams, apparently following Walde-Pokorny, accepts a hypothetical root sna, taken to mean 'to flow', forming a basis for a hypothetical river-name Nava. Rivet and Smith then regard Navio, which they see as the nominative form of the place-name, as that hypothetical river-name Nava, taken to mean 'fast-flowing water', with a hypothetical derivational suffix io for the fort on its banks.


The river-name was probably somewhat of the form Bavione originally, comprising the river-letter b and an avione ending. The b will have changed to v and then the v switched to n (for another example of this process see the entry for Levioxava). The Romans simply transferred the name of the river to the fort which they built at Brough-on-Noe. The modern river-name Noe is of course derived from the Celtic name.


Note that Navione was the name of the Flavian fort at Brough. It would appear that the later fort, thought to have been founded around AD154 and occupied until around AD350, was the Virosido of the ND.




(Celt)     (AI)   (Mod)
Bito     Nido   Neath
(a river)     (Iter XII)   (Glamorgan)


Note 1:

Rivet and Smith see this name as being somewhat obscure. They refer to a hypothetical root neid or nid proposed by Pokorny and taken to mean 'to flow', and they mention one or two other Nida rivers.


The river-name will originally have been Bito, a compound in the river-letters b and t . The b will first have changed to v and then the v changed to n. For another example of the same process see the entry for Levioxava. The river-name was simply transferred by the Romans to a fort which they built on the banks of the river concerned, in this case the river Neath in South Wales. It is not clear when the B and t of Bito changed to the N and d of Nido.


Note 2:

Example of t → d

Example of b → v → n







(Celt) (Rav)       (Mod)
Noviomagno Noviomagno (39)       Maiden Castle


Rivet and Smith see this name as a duplicate of Navimago Regentium, which they locate at Chichester, though it is clear from the order of names in Ravenna that the two names refer to quite different places. The first part of Noviomagno is thought to be derived from hypothetical Celtic novio, taken by some to mean 'new', though Watson and Holder took the view that the word had additional meanings such as 'fresh, lively, vigorous'. The second element of the name is thought to be derived from hypothetical British magos, taken to have meant 'field, plain' originally and to have taken on the meaning 'market' later. The whole name is thus taken to mean 'new place' or 'new market'.


It is probable that the Maiden Castle hillfort was simply called Magno, this being an old-style compound in the hill-letters m and n2, where gn means ‘steep hill’, and that Noviomagno was a new settlement built slightly to the north, after the hillfort had been seized by or surrendered to the Romans, but probably not on the actual site of the later Roman town at Dorchester. Ravenna's Noviomagno will have been a Roman fort built somewhere in the vicinity of that new settlement. There was a similar development further east where the Trundle hillfort was probably just called Magno and a new settlement called Navimagno (presumably the same name as Noviomagno) was built on lower ground, perhaps in this case on substantially the same site as the later Romano-British town at Chichester. Returning to Dorset, at a later date the Romans will have decided to build a new town in the Roman style, complete with forum, basilica, public water supply, amphitheatre, etc. The new Roman town will have retained the name Noviomagno but with the Duro prefix seen in a number of place-names in southeast England. It is still not entirely clear how this prefix was used but it is suggested here that it was applied to new towns built near existing Celtic settlements and intended to persuade the locals of the superiority and desirability of the Roman way of life. The full name of the new Roman town will have been Duronoviomagnoaria, but it is still not clear whether the aria ending was added to the name of the earlier settlement just called Noviomagno or to that of the new Roman town. The latter is perhaps more likely, since it is Noviomagno, without that aria ending, which appears in Ravenna. The aria ending is unusual, the only other examples which come to mind being Calcaria (Tadcaster), which is a Latin name, and Decuaria/Petuaria (Brough-on-Humber). But the very long name Duronoviomagnoaria was shortened by deletion of internal letters to leave Duronovaria.  There are other examples of long names being shortened by deletion of internal letters, e.g. Lincoviglavicus Lincovicus Longovico, or Anderelionubita Anderita Anderitos/Anderidos. It is possible, then, that the Ravenna names (S)aranus to Noviomagno do not in fact record the capture of various hillforts, as the writer has suggested elsewhere on this website, but that they relate to a period very shortly thereafter, i.e. after the Roman fort had been built inside the Iron-Age hillfort at Hod Hill, and after the residents of Maiden Castle had moved into the new settlement of Noviomagno, but before the foundation of Roman Dorchester.


[The entry for Noviomagno was last modified on 11 February 2021]




(Celt)     (AI) (Mod)
Magno     Noviomago Crayford
      (Iter II) (Kent)

Note 1:

Rivet and Smith appear to regard the first part of the name as being derived from hypothetical Celtic novio, taken to mean ‘new’. Watson took the view that novio might additionally mean ‘fresh; lively, vigorous’. For the mago part of the name Rivet and Smith see a derivation from hypothetical magos, taken to mean ‘field, plain’, later ‘market’. The name as a whole might then mean ‘new place’, perhaps ‘new market’.

It seems much more likely, however, that the earlier form of the name was Noviomagno (c.f. Ravenna’s Noviomagno at Dorchester, Dorset). Omission of n is also seen in the Notitia Dignitatum which gives two forms for the name of the fort at Carvoran, an early form Magnis (manned by an old-fashioned cohort) and a later form Magis (manned by a new-fangled numerus). The original Magno will have been a Celtic settlement, most probably a hill-fort, somewhere up on the high ground to the south or west of Crayford. The Romans then moved the inhabitants down to a new settlement at Crayford, this being called ‘new Magno’ by the Romans, i.e. Noviomagno. Then at some later date the second n was dropped or lost to leave Noviomago. The fact that the name appears in the Antonine Itinerary suggests that the Romans built a posting station there.


Note 2:

Example of omission of a letter: n


[The entry for Noviomago was inserted on 22 September 2019]




(Celt) (Rav)       (Mod)
Olcaclavis Olcaclavis (188)       Elginhaugh


Rivet and Smith see Oleiclavis as being the correct reading, with variants Olea Clavis and Oleclavis. They see the name as being a corrupt form of Horrea Classis, a Latin name taken to mean 'granaries of the fleet, storehouses of the fleet', and so they assume their Oleiclavis to have been a coastal base somewhere in southern Scotland or Northumberland. Richmond and Crawford took the view that if the name were really Olcaclevis, then it might mean 'famous fertile fields', based on a word corresponding to olcas, which was apparently used in Gaul in the sense of 'fertile fields', and hypothetical clevos, taken to mean 'fame' or 'something heard'.


But, as it stands, Olcaclavis comprises the inversion-type element olc meaning 'hill steep' and the old-style element cl meaning 'steep hill'. Sometimes, where the same hill-letter appears in two elements of a name the elements are simply arranged in chronological order within the name, with the earlier element first, as in Bannatia. But in Olcaclavis the inversion-type element olc is qualified by the old-style element cl, that is to say the qualifying element lies to the right of the generic term within the name. One has the impression that the people who coined the inversion-type element didn't actually recognise cl as an element with topographical meaning - for them it was just a name. This might be possible if those who coined inversion-type names in the hill-letter l were newcomers from the Continent who had coined inversion-type names for as long as anyone could remember and had forgotten, or never knew, that there had been old-style elements in the hill-letter l. Another example in that same general region of the country is Lagubalium at Carlisle. In this name the inversion-type element lag, meaning 'hill steep', is qualified by the old-style element bal meaning 'high hill'. Against this argument, perhaps, is the fact that there are one or two place-names with transitional elements in the hill-letter l, for example the element bulg in Bladobulgio (the original form of the AI's Blatobulgio). The answer to this little puzzle is that the hill-letter l  in the cl of Olcaclavis, the bal of Lagubalium and the bulg of Bladobulgio is l1, where the hill-letter l1 was used by people who came to Britain at some time prior to 150BC. The hill-letter l in the Olc of Olcaclavis and the Lag of Lagubalium is, however, l2, and this hill-letter was used by Celts who came to Britain sometime after 150BC. Returning to Olcaclavis, the steep hill referred to in both elements of the name is of course the steep river embankment at the top of which the fort stands. 


[The entry for Olcaclavis was last modified on 18 May 2021]




(Celt) (Rav)     (ND) (Mod)
Olerisca Olerica (121)     Axeloduno Maryport

Note 1:

Rivet and Smith prefer to see Olerica as a corrupt form of Ptolemy's Olicana, for which they think Olenacum might be the correct form. They are then in some doubt as to an etymology, though say the name might possibly mean 'property of a man called Olen' where Olen is a hypothetical name. If Olerica be regarded as the correct form then, following Richmond and Crawford, they see a possible derivation from hypothetical Celtic oler(i)ca taken to mean 'swan', though they express some doubt as to whether a simple noun such as this could be used as a place-name without a suffix.


Olerica might be correct as it stands, a compound in the hill-letters l and r with an ica ending, or the name may include the inversion-type element ric meaning 'hill steep'. It seems more likely, however, that the Celtic name had been Olerisca, where the inversion-type element isca, meaning 'hill steep', refers to the cliffs on the western side of the fort - the slope at a place with an isca-type name usually is very steep, sometimes nearly vertical, as at Isca at Exeter and Esica at Haltwhistle Burn. Then, as happened with a number of other names, this name was simply reversed at some stage to yield Acsirelo. This form was then given a duno-ending to yield Acsireloduno, which with the ir dropped gives us Acseloduno, which is of course the Axeloduno of the ND. It would thus appear that Olerica was a Flavian fort at Maryport and Axeloduno the later Hadrianic fort. Axeloduno is thus in its correct position in the ND list and has no connection, other than name similarity, with the Uxelodunum at Castlesteads. Note that the river at Maryport is the Ellen, probably earlier simply Elen, this comprising the river-letter l applied to minor rivers by people who used the hill-letter n, this hill-letter being seen in the duno element referred to above.


Regarding name reversal, it would appear that Ravenna's river-name Novitia was at some point reversed and the v dropped, leaving the t and n of the modern Tyne (in East Lothian). Similarly it would appear that Ravenna's Tagea has been reversed to yield the geath of modern Strageath, and Decha reversed appears to yield the keith of modern Inverkeithing, of course with d → th and ch → k. Furthermore it would seem that Ravenna's Uxelis (Ucselis) was reversed to yield the Lisk of modern Liskeard. Finally, it would seem that Ptolemy's Iscalis was reversed to give Lacsis, that the l was at some point taken to be the river-letter l, and the acs survived as the modern river-name Axe (in Somerset).


Note 2:

Possible example of missing letter - s

Example of name reversal

Example of addition of ending - duno

Example of deletion of letters - ir


[The entry for Olerica was last modified on 15 February 2021]







(Celt) (Rav)       (Mod)
Omirededertis Omiretedertis (25)       Ham Hill

Note 1:

Rivet and Smith see Omire and Tedertis as being two separate names, though they regard Tedertis as being a gross corruption or a fragment, and they remove the first two or three letters of Omire to form the ending of a version of Sorviodunum. The argument is too complicated to repeat here - the interested reader should consult Rivet and Smith 1979.


Omiretedertis is, however, substantially correct as it stands, though the first t was probably a d in the original Celtic name. It seems unlikely that the inversion-type element ret meaning 'hill high' would be followed by another inversion-type element in the same hill-letter, dert, meaning 'summit of hill high'. But there are names in which an old-style element in one hill-letter is followed by an inversion-type element in that same hill-letter, for example Bannatia, in which the old-style element ban is followed by the inversion-type element nat, both elements referring to a location adjacent a high hill. It thus seems more likely that the Celtic name had been Omirededertis, in which old-style element red meaning 'hill summit' is followed by inversion-type element dert meaning 'summit of hill high'. The m at the beginning of the name is the earliest element of the name - it is just the hill-letter m and qualifies the later, but still old-style element red, meaning 'hill summit'.


Note 2:

Example of d → t


[The entry for Omiretedertis was last modified on 27 February 2021]




(Celt) (Rav)       (Mod)
Conda, or Onna (40)       Horsebridge
Conva         (Hampshire)


Note 1:

Jackson proposed a derivation from hypothetical British onno, taken to mean 'ash-tree', whereas Williams preferred hypothetical ond, taken to mean 'stone, rock'. Rivet and Smith preferred derivation from hypothetical onna, taken to mean 'stream, water'.


The order of names in Ravenna appears to indicate that Onna lay between Old Sarum and Winchester, and so most probably at or near the crossing of the river Test, at Horsebridge. There are two Iron Age settlements in that area, both of which have yielded material showing occupation also in the Roman period. The first is an Iron Age enclosure at Fir Hill, Bossington, on the crest of an east-west ridge just west of Horsebridge. The ridge drops quite steeply down to the valley of the Test, so a name of the form Conda (meaning ‘steep hill summit’) would seem appropriate for that enclosure. But occupation of that area seems to go back to the early Iron Age, so if the earliest settlement had been on the side of any of the steep slopes in that area the Celtic name might have been Conva (meaning ‘steep hill slope’). (For a detailed report on the excavations at Bossington see Brown, 2009). The second settlement is the univallate hillfort on the steepish slopes of Ashley Down (at SU 394 301) to the east of Horsebridge. Since the hillfort actually stands on the slope, below the summit, the name Conva would be entirely appropriate. A change from Conda/Conva to the Onna of Ravenna is entirely possible. There are several cases of Ravenna dropping the initial consonant of a name (e.g.  (C)ardaoneon, (C)armis, (Lu)coganges and (L)itucodon), the change of nd to nn is seen in GabaglandaAmboglanna and the change of nv to nn is seen in Banva Banna.

Note that it is not clear whether Ravenna refers to one of the two Iron-Age settlements discussed above or to a Roman fort built to control the crossing of the Test, the name then being transferred to the Roman fort from one of the two Iron Age settlements.


Note 2:

Example of omission of initial C

Example of nd → nn, or nv → nn


[The entry for Onna was last modified on 11 May 2019]




(Celt) (Rav)     (ND) (Mod)
Condo Onno (146)     Hunno Halton Chesters
or         (Northumberland) 


Note 1:

Rivet and Smith see a derivation from onno or hypothetical onna meaning 'stream, water'. They regard the initial H of Hunno as being intrusive, "a mere learned decoration".


It seems more likely, however, that the H of Hunno was originally C and that the initial C had been dropped from the Ravenna form, as in a number of other cases. The original name might then have been Condo meaning 'steep hill summit' (as in the cond of Condecor) or Convo meaning 'steep hill slope'. The Hunno of the ND was the Wall fort at Halton Chesters, whereas Ravenna's Onno was a Trajanic fort. That Trajanic fort may have been on the same site as the later Wall fort, hence Condo, presumably because the Fence Burn runs in a deep ravine just west of the fort, so that the latter can be said to stand at the top of a steep slope. Alternatively the Trajanic fort may have been somewhere down the slope to the south of the later Wall fort, hence Convo. It is to be noted that all the way from Kirkbride over to Newbrough the Trajanic frontier forts stood a little to the south of the later Wall - the same may have been true at Halton. The change from the nd in Condo to the nn in Onno is known in other names, for example in Cambaglanda → Amboglanna. Likewise the change from the nv of Convo to the nn of Onno is known in other names, as in Banva → Banna.


Note 2:

Possible example of loss of initial C

Possible example of nd → nn

Possible example of nv → nn




(Celt)   (Rav)     (Mod)
Bamvocalia   Pampocalia (125)     Ilkley
          (West Yorkshire)

Note 1:

Richmond and Crawford see this name as being a conflation of the two names Cambodunum and Calcaria, a view apparently shared by Rivet and Smith.

But Pampocalia appears to be a straightforward topographical name, though one which has been altered. The Celtic name will have been Bamvocalia, an old-style compound in the hill-letters m and l1, where Bamv means ‘high hill slope’ and cal means ‘steep hill’. The b/v/p interchange is fairly common in place-names, so the change Bamv → Pamp need cause no surprise. The name is a correct topographical description of the location of the Roman fort at Ilkley and Ilkley is in the right area to be Pampocalia, i.e. between Camulodono at Skipton and Lagentium at Castleford. The name Bamvocalia will of course have been that of a Celtic settlement transferred to the Roman fort. That settlement may have been on the same site as the fort or somewhere else in the vicinity.


Note 2:

Example of b  → p

Example of v → p


[The entry for Pampocalia was inserted on 14 September 2019]




(Celt)   (AI)     (Mod)
Benvocrucio   Pennocrucio     on river Penk, south of Penkridge
or   (Iter II)     (Staffordshire)


Note 1:

The first element is generally considered to be derived from hypothetical British penno, taken to mean 'head, hill; end; (as adj.) chief'. The second element is thought to be derived from hypothetical British croco, earlier crouco, taken to mean 'hill, mound, heap, stack; tumulus'. The name as a whole might then mean 'chief mound' or 'tumulus on the hill'. Some scholars have thought it might be the name of a native hill-fort (though there are none in the immediate vicinity) which the Romans simply transferred to their post on the river Penk.


The Romano-British settlement was built on a slope on the south side of the river Penk, though the name was presumably originally that of the Roman fort a little to the northeast, this also lying on a slope. The Penn element may then originally have been Benv (an element of the banv-family), meaning 'high hill slope'. The second element might just comprise the old-style element cr meaning 'steep hill' with an ucio ending, or it may originally have been the old-style element cruv meaning 'steep hill slope'. But note another possibility - the Celtic name may have been Bendocrucio, where Bend is an old-style element meaning 'high hill summit', cr an old-style element meaning 'steep hill' and ucio is simply a name ending. But Bendocrucio would be the name of a hill-fort transferred to the new settlement on the river Penk, with B → P and nd → nn, both known changes in Romano-British names. The nearest hill-fort appears to be Castle Ring, just north of Cannock Wood in Staffordshire. The name Bendocrucio is entirely appropriate for the site of that hill-fortThe hill-fort is some 15 kilometres from Romano-British Pennocrucio, but this distance may not be a valid objection to the name transfer. Presumably if the Romans thought it in their own best interests to transfer the inhabitants and the name of the hill-fort to a new settlement on the river Penk, then they would effect the transfer quite regardless of the distance of 15 kilometres.   


Note 2:

Example of b → p







(Celt) (Rav)       (Mod)
Becsa Pexa (193)       Mumrills

Note 1:

Richmond and Crawford appear to have thought that the name should be Dexa, perhaps related to the name Dexia, the goddess of Good Fortune. Rivet and Smith thought it possible that Pexa should perhaps be Pecti, meaning 'the Picts', or Pectia, meaning 'Pictland'.


But Pexa appears to be a topographical name, albeit one which has been slightly modified. It will have been Becsa originally, where becs is an old-style topographical element in the hill-letter s, the element meaning 'high steep hill'. The Becs element is seen again in the succeeding Ravenna name, Begesse, which will have been Celtic Becsesse. These are old-style names in the hill-letter s and in all probability refer to hillforts, as in the case of Lecimocsava (Ravenna’s Levioxava) at Castle Law, Abernethy, and also Marcotacson (Ravenna’s Marcotaxon) at Castle Craig, south of Pairney, both in Perth and Kinross. It thus seems reasonably clear that Celtic Becsesse and Becsa  were respectively the hillforts on Bowden Hill (southwest of Linlithgow) and Cockleroy (south of Linlithgow), the esse ending of Becsesse indicating that the hill-fort was at the top of a steep slope and overlooked a river, as is indeed the case here, the river being of course the Avon. Note that the hill-letter s of Becs corresponds to the river-letter b, which is changed to v in the modern river-name Avon. Note, further, that the two hillforts are just a little over one kilometre apart. It is probable that Cockleroy, on the highest summit in that area, is the earlier of the two hillforts. It probably became too cramped at some stage and as there was no room for expansion its occupants decided to build a second hillfort on the nearby and slightly lower summit of Bowden Hill, applying to the second fort the name Becsa of their first fort, but with an essa-type ending to distinguish the two forts from one another.

The Romans transferred the names of both hillforts (and presumably the inhabitants of the forts as well) to new fortlets which they built as part of the preparations for the construction of the Antonine Wall across the Forth-Clyde isthmus. It would appear that the name Becsa, modified to Pexa, was applied to a fortlet at Mumrills, and that Becsesse, modified to Begesse, was applied to a fortlet at Seabegs. For further discussion of Roman Pexa and Begesse in the context of the Antonine frontier see Chapter 22 (The Antonine Wall) of the Home menu.


Note 2:

Example of b → p


[The entry for Pexa was last modified on 06 August 2020]




(Celt) (Rav)       (Mod)
Vilatis Pilais (7)       Bideford
 or         (Devon)



Rivet and Smith, like Richmond and Crawford, make no suggestion as to the derivation or meaning of this name.


The writer’s reasons for identifying Bideford as Pilais are given in Chapter 10 of the Home menu (Roman place-names in southwest England). But the name appears to have been transferred to Bideford by the Romans from the Celtic hill-fort now known as Berry Castle at SS 495 223, just southwest of Huntshaw. The hillfort stands on the slope of a high hill, so the Celtic name may have been Vilatis, where vilat is an inversion-type element meaning ‘slope of hill high’. But the hill is fairly steep, especially south of the hill-fort, so the Celtic name may equally well have been Vilacis, where vilac is an inversion-type element meaning ‘slope of hill steep’. The Celtic name will have undergone common changes - initial V changing to B and then B changing to P, and intervocalic t or c being dropped – to leave Pilais. Note that Bideford appears to be the destination of Margary road 493 coming over from Crediton via Burrington Moor. Presumably the Romans built a fort at Bideford to guard a harbour or control a river-crossing.




(Celt) (Rav) (Ptol)     (Mod)
Binnatis Pinnatis (211) Pinnata Castra     Inchtuthil
          (Perth & Kinross)

Note 1:

Rivet and Smith see Pinnata Castra as being a wholly Latin name meaning ‘winged camp’. Richmond preferred ‘camp with merlons’, a suggestion discounted by Rivet and Smith on the ground that such a name would not distinguish this fort from many others. Rivet and Smith thought it possible that the name might be related to Latin pinnae, meaning ‘pearl-bearing fan-mussels’, presumably on the assumption that there might have been many such mussels in a river close to the fort. Rivet & Smith were not of course thinking of the river Tay since they locate Pinnata Castra much further north, probably on the river Findhorn.


But Pinnatis is more likely to be a Celtic topographical name – indeed most Roman forts in Britan had Celtic topographical names. In that case the Celtic form of the name is most likely to have been Binnatis (or possibly Bannatis - see the entry for Bannatia). This is a compound of the old-style element Bin meaning ‘high hill’ and the inversion-type element nat meaning ‘hill high’. The structure of the name (old-style element followed by an inversion-type element) is unusual, but there are a few other names of this kind, for example Omirededertis (Celtic form of Ravenna's Omiretedertis). Celtic Binnatis may have been a few kilometres upstream from Inchtuthil at Dunkeld, at a point where the valley of the river Tay is very narrow, so that a settlement at Dunkeld would permit its occupants to control all movement along the valley of the Tay, arguably the best route from the south into the central Highlands and on from there to the Spey valley and the region around the inner end of the Moray Firth. A settlement at Dunkeld would thus have significant strategic value, a point which would not have been lost on the Romans, who were apparently planning the conquest of the far north of Britain, the only part of the island still outside their control. This identification of Celtic Binnatis is of course speculation, but wherever Binnatis was the Romans will simply have transferred the name to their new fortress at Inchtuthil, at some point the initial B changing to P, in itself a fairly common change in Romano-British place-names. The writer’s reasons for identifying Inchtuthil as Roman Pinnatis are given in Chapter 16: Roman place-names in Scotland, in particular in paragraph 5.2.


Note 2:

Example of b p


[The entry for Pinnatis was last modified on 27 February 2021]













(Celt) (Rav)         (Mod)
Burocoronavis Purocoronavis (6)         Bude
(a river)           (Cornwall)



Richmond and Crawford see Durocornavis as being the correct form, the name meaning ‘fortress of the Cornavii tribe’. Rivet and Smith take essentially the same view but give the form Durocornovium.

It seems clear from the order of names in Ravenna that Purocoronavis was away down in the southwest of England, in Cornwall or west Devon. This is a long way from the Duro-names in southern and eastern England, so it seems unlikely that Purocoronavis was ever a name like Duroaverno or Durocobrivis. In addition the change D → P is very unusual (indeed no other example comes to mind). It thus seems very unlikely that Purocoronavis was ever DurocoronavisIt is much more likely that the initial P of Purocoronavis was originally a B, Burocoronavis then being a river-name of the kind having a river-element, here a compound of the river-letters b and r, added as a prefix to a land-name, here Coronavis, apparently a compound in the hill-letters r and n, where the old-style element Cor means ‘steep hill’. Note that if this area was later settled by l –people they might have added their river-letter t  to the end of the river-element, so that the latter would then become Burit. It is then possible that with the deletion of intervocalic r and the change of t to d Burit would become Bud – this may be the origin of the town-name Bude. The river at Bude is called the Neet or the Strat, and it is possible that Neet is also derived from Burit, though here the development would have been different. The initial B may have changed to V and the V may have switched to N. The intervocalic r would again be omitted and Nuit may have become Neet. As to Coronavis this will have been the name of a Celtic hillfort or settlement adjacent a steep slope close to the river. There are in fact two Celtic settlements close to the river Neet. The first is the settlement on Stamford Hill, just north of Stratton. The second is further northeast, next to Hunthill Wood. In the case of both settlements the drop down to the river is steep and so either settlement may have been Coronavis. But if the n of Coronavis was originally a v, then the old-style element Corov, meaning ‘steep hill slope’ would be appropriate for the settlement at Hunthill Wood but apparently not for that at Stamford Hill. The Romans will simply have transferred the river-name Burocoronavis (appearing with initial P in the Ravenna form of the name) to a fort which they built at Bude, the fort being built here presumably because it could easily be supplied by sea.

But note that there is some inconsistency in maps as to which river in that region is actually called Neet. If the river flowing up the western side of the steep hill crowned by the Ashbury Camp hillfort (NGR: SX 228 975: just west of Week St Mary) is actually the Neet then that hillfort will have been Coronavis and the river the Burocoronavis.


[The entry for Purocoronavis was last modified on 03 April 2021]




(Celt) (Rav) (Ptol) (AI)   (Mod)
Rate Ratecorion (92) Rate Ratas   Leicester
      (Iter VI)    
      (Iter VIII)    


This name is generally regarded as being derived from hypothetical British ratis, taken to mean 'earthen rampart, fortification, fort', though Rivet and Smith point out that whilst there is evidence for a pre-Roman settlement at Leicester there is none for a pre-Roman fortification. Rivet and Smith refer to another hypothetical British ratis, taken to mean 'fern, bracken', which in their view might possibly explain some names with the element rat.


There is high ground on both sides of the river Soar at Leicester, so the name might include the inversion-type topographical element rat meaning 'hill high'. It would appear that Ravenna's Corda (171) was also at Leicester and this name probably includes the old-style element cord meaning 'steep hill summit', the name probably referring to a location up on the high ground coming into the modern city from the southeast. It is conceivable that Corda lost its initial C (as happened with several other Ravenna names) and that orda changed to rate. Note that Ravenna's form Ratecorion acquired its cori element from the next-following name in Ravenna, Eltavori - this was earlier Corieltavori (the name appearing on a tegula which was found at Cave's Inn). The name Eltavori is also known in the Eltanori form and it is possible that Ptolemy's tribal name Coritani is merely a Roman simplification of Corieltanori, i.e. Cori [el] tan [or] i → Coritani. This suggests that the tribal centre was at Lutterworth in the pre-Roman period and was moved to Leicester by the Romans.


[The entry for Rate was last modified on 11 February 2021]




(Celt) (Rav)       (Mod)
Rascatonion Ravatonium (209)       Stracathro


Note 1:

This has been taken to be a river-name in the past. Williams related the name to a Sanskrit root sravat meaning 'stream'. Another possibility would be the hypothetical root sruta, taken to mean 'flowing'.


But Ravatonium appears to be a topographical name, though one which has been modified. The Celtic name will have been Rascatonion, where ascat is an inversion-type element meaning 'hill steep high', the element referring to the steep, high river embankment at Stracathro, assuming that the Celtic settlement was on the same site as the later Roman fort. The asc part of the name survived in the modern river-name North Esk. At some point the was dropped and Rac changed to Rav. One sees the same c → v change in Leciodanum → LeviodanumLeciocsava → Levioxava and Lacobrinda Lavobrinta


Note 2:

Example of c → v


[The entry for Ravatonium was last modified on 02 January 2021]




(Celt) (Rav)       (Mod)
Bravonia Ravonia (123)       Elslack
          (North Yorkshire)

Note 1:

Rivet and Smith, following Richmond and Crawford, see this name as a corrupt form of Bravoniacum at Kirkby Thore.


However, the order of names in Ravenna appears to indicate that Ravonia was between Camulodono at Skipton and Bresnetenaci at Ribchester, so Elslack seems the obvious candidate. This fort appears to stand on the lower slopes of a hill which rises to a height of about 250 metres above the fort itself. The Celtic name was thus most probably Bravonia, where Brav is an old-style element meaning 'high hill slope'. The fort is now called Burwen Castle, though whether Burwen is derived from Bravonia is not clear.


Note 2:

Example of missing initial letter - B




(Celt)       (ND) (Mod)
Regulvio       Regulbio Reculver
 or         (Kent)



Rivet and Smith see this name as a compound of a hypothetical British prefix ro, taken to mean 'great', and hypothetical gulbio, taken to mean 'beak' or, metaphorically, 'headland'.


The Roman fort called Regulbio stood on raised ground at the northeastern corner of what was then mainland Kent, Thanet being an island at that time. Much of the fort has been eroded by the sea, so that there is now a cliff immediately north of what is left of the fort. Presumably there was also a cliff in the Roman period - that cliff has simply moved south as land has been eroded. The g in the name will then just be the adjectival c meaning 'steep' (c/g interchange is common in Romano-British names). The elements of the name are thus most likely to be reg and l , where reg means 'hill steep' and l  is just the hill-letter l. The b cannot be the adjectival b meaning 'high' since that b comes before its hill-letter, not after. The b may be original or may be a modified v, the bio/vio part of the name just being an ending, which may or may not have had meaning.


[The entry for Regulbio was last modified on 19 September 2019]




 (Celt)   (Ptol)      (Mod)
 Rerigonion   Rerigonium      Girvan
(a river)         (South Ayrshire)



Ptolemy places Rerigonium in the territory of the Novantae.


Rerigonium is generally regarded as a latinisation of British ro-rigonio, where ro is a hypothetical prefix, taken to mean 'great', and rig or rigon a hypothetical British term taken to mean 'king', the name as a whole thus being taken to mean 'very royal (place)'.


The above seems a very unlikely meaning for a Celtic place-name. It seems much safer to see Rerigonion as a river-name of the kind having a river-letter, here r, added as a prefix to a land-name, here Rigonion, where Rig is an inversion-type element meaning ‘hill steep’. The river-name thus suggests that the Novantae, who appear to have used the hill-letter m and river-letter r, had extended the territory under their control to include the southern part of modern Ayrshire. That the then inhabitants of Ayrshire used the hill-letter r is borne out by Vindogara (Celtic Bindogara) at Irvine, where the old-style element gar also uses the hill-letter r.  Rigonion may have been the hillfort on Dow Hill, immediately south of Girvan. The Novantae may then have seen the Water of Girvan as the river at Rigonion and so called it the Rerigonion. But note that there may have been a hillfort upriver at Doonans Hill (NS394 028), a little south of Straiton. Opinions seem to be divided as to whether there was a hillfort there at all, but if there was then the hillfort stood at the top of a steep slope immediately adjacent the river. The hillfort may thus have been called Rigonion and the river the Rerigonion. Ptolemy’s Rerigonium will have been a Roman fort somewhere close to the river and simply took its name from the river. That fort may have been built at Girvan to defend a harbour, though no trace of the fort has yet been found. And of course Ptolemy’s Rerigonius bay will have been the bay at the mouth of the river Rerigonion, the Water of Girvan.

Note that Rerigonium is conventionally identified as a place on Loch Ryan. However, the Novantae (who appear to have used the hill-letter m and the corresponding river-letter r), seem already to have been in control of the Luce valley and the Rhinns of Galloway in the period prior to about 130BC  (when old-style names were being coined), so it seems unlikely that people using the hill-letter r  lived around Loch Ryan as late as the 1st century BC, which is when the inversion-type name Rigonion is most likely to have been coined. We need, therefore, to locate Rerigonium somewhere north of Loch Ryan.


[The entry for Rerigonium was last modified on 12 January 2021]




(Celt)   (Ptol)     (Mod)
Rigodunum   Rigodunum     North Rigton
          (North Yorkshire)


The first element is generally thought to be derived from hypothetical rig or rigon, taken to mean 'king', and the second is thought to be Celtic dunon, from hypothetical dunos, taken to have meant 'hill' originally and later to have taken on the meaning 'fort'. The name as a whole would then have a meaning such as 'king-fort' or 'royal fort'. 


Rigodunum, however, appears to be a topographical compound comprising the inversion-type elements rig meaning 'hill steep' and dunum meaning 'summit of hill', so the name tells us that the Roman fort (assuming that is what Rigodunum was) was at the top of a steep hill or slope. The order of names in Ptolemy appears to indicate that Rigodunum was at North Rigton, to the southwest of Harrogate. The modern village stands on a steep slope and appears to extend on to the top of the hill, which is where one would expect to find a place called Rigodunum.









(Celt) (Rav)       (Mod)
Rumabo Rumabo (190)       Cramond
(a river)         (Edinburgh) 



Rivet and Smith see Rumabo as being the same name as Ptolemy's river Abi, which they identify as the Yorkshire Ouse, with its estuary, the Humber. They suggest that the compiler of Ravenna misread Flum Abo on a map as a habitation name and further misread Fl as R.


Rumabo is, however, a straightforward river-compound in the river-letters r  and m with a bo ending. Normally the river-letter m appears to be applied to major rivers, so it is a little surprising to see it used in the then name of the River Almond. But there is a special case where a river has two headwaters. The river-letter may be applied to one or both of the headwaters and the river-letter m  to the main river. One sees this in the rivers at Cardean. The river-letter l  was applied to one of the headwaters, namely that stretch of the modern Isla upstream from Cardean, whereas the river-letter  was applied to the main river, the stretch of the modern Isla downstream from Cardean - this river was called the Matovion. The river Almond also has two headwaters so in this case, too, the river-letter was applied to the main river. The Romans simply transferred the river-name Rumabo to the (Flavian) fort which they built either at Cramond or further upriver, though so far as the present writer is aware no trace of that fort has yet been found. Note that the river-letters r and m correspond to the hill-letters m and n, both of which appear in the mond element of the modern name Cramond. It would thus appear that the land-name of the location was of the form mond  at one time and that at some later date new settlers added their hill-letter l  to yield a form somewhat like Lamond or Almond, this name then being transferred to the river by the Romans during the Antonine period. Note, however, that it is much more likely that the place called Mond, later Lamond or Almond, was not Cramond itself but the hillfort on Craigie Hill, on the north side of the river Almond at a point just north of Edinburgh airport. During the Antonine period the Romans will have transferred the name and presumably the inhabitants of the hillfort to Cramond. Flavian Rumabo was most probably at Cramond, too, but not necessarily on the same site as the later, Antonine fort.


[The entry for Rumabo was last modified on 22 January 2020]




(Celt) (Rav)   (AI)   (Mod)
Borto Veratino (87)   Rutunio   (where Margary road 6a
(a river)     (Iter II)   crossed the river Roden, Shropshire)



Ekwall and Jackson suggest a derivation from hypothetical Indo-European rev, taken to mean 'to move swiftly'. Rivet and Smith then suggest the river-name would have been Rutuna and the derived place-name Rutunion.


Veratino is a river-compound in the river-letters b (changed to v), r and t. The Romans simply transferred the name of the river to a fort or posting station built close to the river, in this case at the point where the Roman road from Whitchurch to Wroxeter crossed the river Roden in Shropshire. Note that the form Borto suggested above is merely the river-prefix of the composite river-name (such names are discussed in Home/Chapter 19, 11) of the river Roden. For a more detailed discussion see the entry for Mediolano.


[The entry for Rutunio was last modified on 01 May 2021]







(Celt) (Rav) (Ptol) (AI) (ND) (Mod)
Turupis Rutupis Rutupie Ritupis Rutupis, Richborough
(a river)  (73)   (Iter II) Rittupis (Kent) 


Note 1:

This name has caused scholars some difficulty in the past. Williams associated the Rut part of the name with Welsh rhwd, now meaning 'rust' but earlier 'filth' in general. On the basis of that suggestion Richmond and Crawford suggested for the name a meaning 'mud-flats, muddy creek(s)'. Hamp considered the initial Rutu to be derived from hypothetical rutu meaning 'mud', the name then meaning 'muddy (estuary, waters, shallows)'.


But the name is a river-name and it is likely that the Celtic name was in fact Turupis, the river-letters t and r  then being in the same order as in the modern river-name Stour, in Ravenna's river-name Durbis (where the t is changed to d) and the modern river-name Dour (the river at Dover), and in the same order as the corresponding hill-letters l and m in the place-name Lemanis (Lympne). Richborough, Dover and Lympne are fairly close together, so it is likely that they were controlled by the same group of Celts at any given historical time. The Romans will then have rearranged Turupis  to Rutupis, just as they rearranged Durbis to Dubris.


Turupis is then a straightforward river-compound in the river-letters t and r with a pis ending seen as bis in Durbis. The river-name was simply transferred by the Romans to the fort/harbour which they built at Richborough. The letter s in the river-name Stour will be the river-letter s corresponding to the hill-letter r of Averno, the late Celtic name of the hillfort now known as Bigbury Camp, just west of Canterbury. This Celtic name was incorporated in the name of Romano-British Canterbury, Duroaverno.


Note 2:

Example of bp