Ptolemy’s Celtic tribes: Part 2


[Part 1 of ‘Ptolemy’s Celtic tribes’ is directly accessible using the menu entry shown above. It includes a few notes relating to the structure of Celtic/Romano-British place-names and then discusses the Iceni, the Silures, the Deceangli, the Damnoni, the Dumnoni, the Venicones, the Parisi, the Brigantes and the Coritani. Part 2, given on the present page, discusses the Trinovantes, the Catuvellauni, the Canti, the Dobunni, the Cornovii, the Durotriges, the Atrebates, the Belgae, the Regni, the Selgoves, the Votadini, the Novantae, the Demetae and the Ordovices.]


10     The Trinovantes, the Catuvellauni and the tribes of Kent

10.1     It seems easiest to consider the Trinovantes, the Catuvellauni and the tribes of Kent together as a group, and to do that by looking at the chronological develoment of place-names in southeast England, both north and south of the Thames estuary. Leaving aside Sulloniacis (Brockley Hill), which may perhaps include the hill-letter s, the earliest place-name elements in that region are old-style elements in the hill-letter m. One sees such an element in Camuloduno (Colchester), where cam means ‘steep hill’. The m is probably also old-style in Verulamium (St Albans) and Lemanis (Lympne). But those m-people clearly lived over a much wider area since we see the corresponding river-letter r in the modern river-names Dour, Stour and Darent. Note, too, that the mago of the Noviomago of Iter II  of the Antonine Itinerary was probably originally magno (as in the case of Ravenna’s Noviomagno at Maiden Castle), where the m is the hill-letter m. Romano-British Noviomag(n)o (in Iter II of the Antonine Itinerary) is thought by some to have been the settlement at Crayford, which is not far from the Darent, though Celtic Magno may have been somewhere else in the vicinity, adjacent a steep hill, which is the meaning of the ‘gn’ element of Magno.

10.2     At some point people who used the hill-letter r  settled at Brabis, the Celtic precursor of Ravenna’s Durobrabis at Rochester. 

10.3     Later, much of the region north of the Thames was settled by people who used the hill-letter l1 .  They took over Colchester (now called Camulo) and may have taken over Brockley Hill (if the later Celtic name was Suldodinacis) or may actually have founded it (if the later Celtic name was Culdodinacis). They also founded London (later Londinium) and occupied the old hillfort at Wandlebury (later Libonde), southeast of Cambridge.

10.4     Then at some point the land to the south of the inner end of the Thames estuary was settled by n-people. They took over a settlement (now called Magno) near Crayford. They may have founded Vagniacis (Springhead) if the first part of this name was originally the old-style element Bagn meaning ‘high steep hill’. In addition, if the Med of the river-name Medway is a Celtic river-compound, as seems quite possible, then the m will be the river-letter m applied to major rivers by those who used the hill-letter n2

10.5     At some later point people using the hill-letter l2 settled in the area between the Darent and the Medway. We see their river-letter t in the river-name Darent and, changed to d, in the river-names Medway and Eden. The Eden element of the modern place-name Edenbridge is conventionally considered to be derived from an Old English personal name Eadhelm, but it might be possible that the river-name is in fact a straightforward Celtic river-name in which the d is a modified river-letter t corresponding to the hill-letter l2. There is, for example, a river Eden in Fife very close to Levioxava at Abernethy, the l in this name being l2. There is also a river Eden at Carlisle, the d in this name apparently being a modified river-letter t corresponding to the l in the Lag element of Lagubalium (Carlisle), this l  being l2. In addition much of what is now South Yorkshire seems to have been inhabited by people using the hill-letter l2 (as explained in paragraph 7 above) and in that area there is a river Don. Finally there is a Dean Water in Angus, the d in this name being a modified river-letter t corresponding to the hill-letter l2 in Ugrulentum at Cardean. Thus unless there is clear proof that the Kent river-name Eden is derived from the personal name Eadhelm it is safer to take it to be a straightforward Celtic river-name originally coined by people who used the hill-letter l2.

10.6   We have no tribal names for the people who lived in Kent and used the hill-letter m, r or l2.  But the n-people of northwest Kent will have been the original Canti. They will not have been called Canti when they arrived in Britain because when they arrived they coined old-style place-names, whereas the tribal name Canti  is based on a transitional place-name (see paragraph 10.10 below) and is thus later in date. The l-people living north of the Thames were Ptolemy’s Trinovantes.

10.7   At some point prior to 130BC the Canti expanded north of the Thames, taking control of London, which was from that time on called Londinion. They expanded further, up the valley of the river Lee (river-letter l corresponding to hill-letter n2) and appear to have taken over Wandlebury (now Libonde) and Bicester (now Brinavis). But at some point, presumably when the centre of gravity of the tribe had shifted north of the Thames, the Canti living north of that river were given a new name, the Catuvellauni, this tribal name being based on an inversion-type topographical place-name (see paragraph 10.9 below), thus later in date than Canti. At a later date the Catuvellauni expanded eastwards, displacing some Trinovantes, who moved to Kent and took over a swathe of land stretching from the northeast coast of Kent right round southeast Kent as far as the river East Rother. We see this movement of the Trinovantes, who used the hill-letter l1 and river-letter t, in the place-names Regulbio (Reculver), possibly Levo (the Celtic predecessor of Durolevo in the Sittingbourne area) and in Lemanis, as well as in the river-names Stour, Dour and Rother (this point being explained in more detail in paragraph 2.6 of Chapter 23).

10.8   The tribal name Trinovantes is based on a river-name, as are the tribal names Novantae and Taexali. The river-name will have been of the form Trinobandion, this being a river-name of the kind comprising a river-prefix, here a compound of the river-letters t and r with an in ending, attached to a land-name, here bandion, this element using the hill-letter n1, meaning ‘high hill summit’ and referring to a place up on the top of high ground close to the river Pant/Blackwater (the name Pant will be derived from bandion). The people of that area will have been called the Trinobandes. With the common change d→t we obtain the form Trinobantes, which form is actually used by Caesar and Tacitus, and with the further common change b→v  we obtain the better-known form Trinovantes.

10.9   The tribal name Catuvellauni, like most British Celtic tribal names, is based on a topographical place-name. The place-name was most probably of the form Vencatuvelcatunion, with two elements, Vencat and velcat, having exactly the same meaning, namely ‘slope of hill steep high’. The name indicates that the place concerned had belonged to the Trinovantes but was taken over by the Catuvellauni. The people of the tribe will have been called the Vencatuvelcatuni. Then, with loss or omission of Ven and the second t, and the change lc→ll, we obtain the form Catuvellauni. The place Vencatuvelcatunion may have been the hillfort/settlement at Langdon Hills in Essex (part of modern Basildon) and was perhaps occupied by the Catuvellauni during their eastwards expansion around 100 BC which caused some Trinovantes to move to Kent (see paragraph 10.6 above). The hillfort/settlement stands on a slope, with a 10 metre difference between its highest and lowest points (Lock, G. and Ralston, I., 2017). Another possible location for Vencatuvelcatunion would be Braughing. The enclosure at Gatesbury Wood, just south of Braughing, stands on a slope, so if that enclosure was occupied in the first half of the 1st century BC it may have been Vencatuvelcatunion. The nearby enclosure, possible oppidum, between modern Braughing and the river Rib, also appears to be built on a steep, high slope, so here we have a third candidate for Vencatuvelcatunion.

10.10   The tribal name Canti appears to be based on a topographical place-name Cantion, this comprising the transitional element Cant meaning ‘steep hill high’. One sees the element as cunet in Cunetio. The place-name will have been coined sometime around 125BC and will have been the name of a place in northwest Kent, probably west of the Darent. Later, during the Roman period, the name Canti was applied to all the people of Kent, but this was probably already the case in the late pre-Roman period. The Catuvellauni appear to have taken control of the r-people of northeast Kent after Caesar returned to the Continent in 54BC (see Chapter 23, 2.9.2).  Later, when they absorbed the Trinovantes in Essex, they presumably also gained control of the Trinovantes living in southeast Kent. Thus, assuming the l-people who lived between the Darent and the Medway had also been brought under Catuvellaunian control, it is probable that all the inhabitants of Kent were officially Canti regardless of their original tribal affiliation. They were all subjects of the Catuvellaunian state, with its capital at Camulodunum,  Colchester.



11     The Dobunni

When we look at place-names in the region thought to have been occupied by the Dobunni we see a preponderance of names in which the hill-letter r is the only or the latest hill-letter in the name. We thus see Cironium/Corinium (originally at Bagendon, the name later being transferred to Cirencester), Vertis (Worcester), Ariconio (Romano-British town east of Weston-under-Penyard, but most probably earlier the name of the hill-fort in Chase Wood, just south of Ross-on-Wye), Argistillum (somewhere between Kenchester and Worcester, perhaps the Roman fort at Stretton Grandison, but the name will earlier have been that of a Celtic settlement somewhere in the vicinity) and Verlucione (Sandy Lane, southeast of Chippenham). There are a few place-names in this region which do not include the hill-letter r, for example Glebon/Clevo (Gloucester, but probably earlier the name of the hillfort at Crickley Hill, near Birdlip), which uses the hill-letter l1 and Macatonion (Dymock) which uses the hill-letter m.  But the preponderance of names in the hill-letter r seems to make it quite clear that the Dobunni  used the hill-letter r. Brinavis (Bicester) and Cornovio (Celtic precursor of Durocornovio near Wanborough, but most probably earlier the name of Liddington Castle or Uffington Castle) are interesting in that it would appear that both places were founded by people who used the hill-letter r but were taken over by people who used the hill-letter n2, presumably the Catuvellauni, though it may be that Cornovio was taken over by Atrebates (discussed below) moving up from the south. Note that it is possible that the r-people living in and around Liddington Castle or Uffington Castle moved northwest to the area around the Wrekin when the n-people took over Liddington Castle or Uffington Castle. Those r-people may then have given the name Biriconion to the hill-fort on the Wrekin (evidently replacing an earlier name since the hill-fort was built at an earlier date), where Biric is a transitional element (meaning ‘high hill steep’) and thus later in date than the old-style element Cor  of Cornovio. When the Romans created two civitates in the west of England, the southerly one being centred on Cirencester and being the civitas of the Dobunni, and the northerly one being centred on Wroxeter, they may have called the northern civitas the civitas of the Cornovii because they knew that the people living in the Wrekin/Wroxeter area had migrated from the Liddington Castle or Uffington Castle area, the hillfort here being called Cornovio in the late pre-Roman period, after it had been taken over by n-people. The Romans will have transferred the name of the hill-fort on the Wrekin to their fortress (and later town) at Wroxeter, the initial B  of Biriconion being changed to V to give the better-known name Viriconio.  It would thus appear that the Cornovii also used the hill-letter r. Indeed they may have been part of the Dobunni  tribe in the pre-Roman period. In this case it would appear that the two names Brinavis and Cornovio indicate that at some date the Dobunni came under pressure from their neighbours to the east, the Catuvellauni, and those to the southeast, the Atrebates.

11.1   Dobunni appears to be a tribal name like most of the British Celtic tribal names, i.e. it is based on a topographical place-name, presumably the name of the place where the tribal chieftain lived. That topographical place-name was probably somewhat of the form Derobundion, where Der means ‘summit of hill’ and bund means ‘high hill summit’. The chronologically later of the two elements, Der, of course uses the hill-letter r because that was the hill-letter used by the Dobunni. The people of Derobundion will have been called the Derobundi, which form, with loss or omission of er and the fairly common change nd → nn (cf. Gabaglanda Amboglanno), yields the well-known form Dobunni. Derobundion will have been a hillfort on the summit of a high hill, but as there were many hillforts in Gloucestershire, and in the neighbouring parts of adjoining counties, the writer is unable to identify any particular hillfort as having been Derobundion.


12     The Durotriges and the Atrebates

12.1     Turning now to the south of England, west of Kent, it would appear that the two commonest latest hill-letters in place-names are l1 and n2. The map below indicates all place-names in southern England in which l1 is the only or the latest hill-letter of a name (leaving aside Giano Eltabo at Nanstallon, which may have been Celtic Glavo Eltabo, and also Anderelion, most probably a hill-fort somewhere in the catchment area of the river reaching the sea at Pevensey).




The place-names are as follows:

10 Bindogladia

11 Axula

12 Meledio

13 Leucomago

20 Clavinio

21 Dolocindo

22 Lindinis


23 Uxella

24 Bolbelagunio/Borbeladunio



31 Bernilis

32 Vilatis/Vilacis

33 Elconionemedo

34 Ucselis


One sees Bindogladia at Weatherby Castle, Axula (a place-name included within the river-name Traxula) at an as yet unidentified location on the river Axe, Meledio (probable original form of Ravenna’s Melezo) on the summit of Melbury Hill and Leucomago at Leckhampstead in Berkshire. Further west one sees Clavinio at Charterhouse, Dolocindo at Westbury Camp, near Westbury-sub-Mendip, Lindinis at Dundon Hill, Uxella at Cannington Park, southwest of Combwich, and Bolbelagunio/Borbeladunio (possible Celtic forms of Bolvelaunio) at Wiveliscombe. Further west still one sees Bernilis (Celtic form of Vernilis) at Roborough Castle just north of Barnstaple, Vilatis/Vilacis (Celtic form of Pilais) at Berry Castle, to the east of Bideford, Elconionemedo at Launceston and Ucselis (Celtic form of Uxelis) at Liskeard. Not shown on the map is Caleva/Calleva at Silchester. This name appears to have been Celtic Calveva, where the l of the old-style element Calv (meaning 'steep hill slope') is l1. The name was transferred to a new Celtic settlement at Silchester on the same site as the later Romano-British Calleva from a hillfort at Pond Farm, about 1 kilometre northwest of Calleva.

12.2     The names in which nappears to be the only or the latest hill-letter are shown on the map below.



The place-names are as follows:

1 Venta

2 Claducendum

3 Cunetio

4 Anicetis

5 Sorbilodoni

6 Noviomagno

7 Cantia

8 Moriduno

9 Milidunum

10 Conda/Conva

11 Carnis

12 Ardaoneon

13 Navimagno Regentium

14 Landini Tamese

15 Cornovio


One sees Venta at Winchester, Claducendum (probable Celtic form of Clausentum) at Meonstoke, Conda/Conva (possible original forms of Ravenna’s Onna) at Horsebridge, Carnis (probable original form of Ravenna’s Armis) at Butser Hill, Ardaoneon at Harting Beacon, Navimagno Regentium (probable original form of Ravenna’s Navimago Regentium) at Chichester and Cunetio at Mildenhall. Further west one sees Anicetis at Castle Ditches near Ansty in Wiltshire, Sorbilodoni (probable Celtic form of Sorbiodoni) at Badbury Rings and  Noviomagno at Maiden Castle, south of Dorchester. Still further west one sees Cantia (Celtic form of Canza) at Henton in Somerset, Moriduno at Hembury Hill and Milidunum, apparently the hill-fort on Whitechapel Moss to the east of South Molton. But note that Moriduno and Milidunum may be names in which the latest hill-letter is m, the hill-letter used by the Dumnoni (see 'Ptolemy’s Celtic tribes: Part 1’, 5.4).

12.3     It is clear that the centre of gravity of the n-names is further east than that of the l-names and that the n-names include Venta (Winchester), later the civitas capital of the Belgae.  We also know from two inscriptions, RIB 1672 and 1673, that Lindinis (apparently at Dundon Hill) was in the civitas of the Durotriges. Lindinis is a compound in the hill-letters l1 and n2. In light of the above points it seems reasonable to conclude that the hill-letter l1 was used by the Durotriges and that n2 was used by the Atrebates (this term is used broadly to include the Belgae and the Regni). The gn of Noviomagno  at Maiden Castle in Dorset is an old-style element in the hill-letter n2 and so is earlier in date than Venta. Indeed the place-name Noviomagno contains no indication that Maiden Castle was ever occupied by the Durotriges. It indicates that the Atrebates took the hill-fort over from earlier settlers who used the hill-letter m. This is borne out by the name, Frome, of the nearby river, the river-letters r and m of this name corresponding to the hill-letters m and n of the magno part of Noviomagno. But there were Durotriges to the north. We see this in the names Bindogladia (Weatherby Castle) and Meledio (probable Celtic form of Ravenna’s Melezo at the top of Melbury Hill). And of course we see the corresponding river-letter t in the river-name Stour. But to go back to the hill-letter n2 there are two old-style elements in place-names on the eastern side of the Severn estuary, namely Lindinis (Dundon Hill) and Dolocindo (Westbury Camp, though the n in this name might be n1) and on the western side of the estuary we see Alvinundo (probable Celtic form of Ravenna’s Albinumno at Welshbury Camp near Cinderford), Banvio/Gobanvio (Celtic forms of Bannio/Gobannio at Abergavenny), Magnis (at Kenchester, though it was probably originally the name of the nearby hill-fort in Credenhill Park Wood) and Bravogenium (probable Celtic form of Branogenium at Leintwardine). In all of these names west of the Severn estuary the hill-letter n2 appears to be used in an old-style manner, as in Lindinis and Dolocindo on the eastern side of the estuary. This rather speaks for a landing of n2-people around the Severn estuary quite separate from the landing of the Atrebates on the Hampshire/West Sussex coast. And further south we see the transitional name Cantia (probable Celtic form of Ravenna’s Canza) apparently at Henton in Somerset and the later inversion-type element dun in Moriduno (Hembury Hill) and Milidunum (apparently the hill-fort a little east of South Molton). These three names may indicate southern expansion of the n2-people on the eastern side of the Severn estuary rather than a westwards expansion of the Atrebates

12.4     But in the Roman re-organisation of government the old territory of the Atrebates was split in two, a northern civitas of the Atrebates centred on Caleva/Calleva at Silchester and a more southerly civitas of the Belgae centred on Venta at Winchester. We have no precise information as to the location of the boundary between those two civitates. We do not therefore know, for example, whether Anicetis at Castle Ditches near Ansty and Saranus at Old Sarum were in the civitas of the Atrebates or that of the Belgae. But there are a few place-names in the hill-letter n2 which were presumably included in the civitas of the Atrebates. These are Cunetio at Mildenhall, Landini Tamese apparently at Streatley on the river Thames, and Cornovio, apparently the hill-fort known as Liddington Castle or that known as Uffington Castle, though this was later replaced by the new town called Durocornovio in the south-eastern outskirts of modern Swindon. 

12.5     The Regni present a slight problem. It is suggested elsewhere on this website that Ravenna’s Navimago (Regentium) was really Navimagno (like the Noviomagno at Maiden Castle in Dorset) where magno is an old-style compound in the hill-letters m and n2. The name Magno will have been that of the hill-fort known as The Trundle. In addition we have Ardaoneon at Harting Beacon. The Celtic name appears to have been Cardadonecon, where the n will be n2In addition it is suggested elsewhere on this website that Ravenna’s Armis (at Butser Hill) was probably Carnis, where the n is the hill-letter n2.  But otherwise, leaving aside Claducendum (apparently the Celtic form of Clausentum) at Old Winchester Hill, which may have been a settlement of the Atrebates and later perhaps included in the civitas of the Belgae, there would appear to be no place-names along the south coast including the hill-letter n2.  If, then, much of Sussex was in the territory of the Regni, one has the impression of a Regni elite living in the Chichester area and ruling a population largely made up of descendants of earlier settlers. But so far as one can see that ruling elite used the hill-letter n2, seen in Navimagno, Cardadonecon and Carnis.

12.6     But the population distribution in southern England was much more complicated than is suggested by Ptolemy’s reference to five tribes, the Atrebates, the Belgae, the Regni, the Durotriges and the Dumnoni. We see, for example, and leaving aside names in west Cornwall, place-names in which the only or the latest hill-letter is s, as in Isca at Exeter, Iscalis on or close to the Somerset Axe and probably Masona, most likely on the North Devon coast, or as in Omirededertis (Celtic form of Omiretedertis) at Ham Hill, Morionio at Norton Camp, Vertevia at Tavistock, Saranus (Celtic form of Aranus) at Old Sarum, Brige, most probably Tatchbury hill-fort to the southwest of Southampton, and Racstomessa (Celtic form of Raxtomessa) at Mount Caburn near Lewes, or even l2, as in Lelamon (Celtic form of Melamoni) on the West Dart or East Dart river and Locs (Celtic form of Lox, a place-name included within the river-name Velox) somewhere on or close to the river Brit. But note that whilst Morionio and Saranus have been treated elsewhere on this website as names in which the letter n is just part of the name-ending, it is possible that the n is the hill-letter n2, indicating that all three places were taken over by n2-people from earlier people who used the hill-letter r. The above names in the hill-letters and r indicate that not all of the earlier inhabitants of that region were wiped out or displaced on the arrival of the Durotriges, and later of the Atrebates, though whether they retained some form of independence or were subject to the rule of the Durotriges or Atrebates is not something one can deduce from the place-names themselves. Nonetheless, the fact that the Romans chose Isca at Exeter as the administrative centre of the Dumnoni, who used the hill-letter m, might suggest that the s-people in Devon were considered to live on Dumnonian territory.

12.7   We see in Durotriges another case of a tribal name based on a topographical place-name, the place-name apparently having been of the form Dulodrigion (Dul = ‘summit of hill’; drig = ‘summit of hill steep’). This name indicates that the Durotriges, who used the hill-letter l1, had taken over a hillfort previously occupied by a people who used the hill-letter r. The people ruled from Dulodrigion will have been called the Dulodriges. Then, with simple changes – the r/l  interchange and d → t  Dulodriges became Durotriges. Dulodrigion was probably somewhere on the eastern side of the Severn estuary, where there is a concentration of place-names in the hill-letter l1. One sees the Dul element as Dol in Dolocindo, this Celtic name apparently referring to Westbury Camp, just west of Westbury-sub-Mendip.

12.8   The Atrebates appear to have brought their name with them when they came to Britain from the Continent. The origin of the name is thus to be sought on the Continent, not in Britain. Nonetheless, since it is a name which developed in the same way as the tribal names Trinovantes, Novantae, Ordovices and Taexali  it will be discussed briefly here. The tribal name Atrebates is based on a river-name of the kind comprising a river-prefix, here a compound of the river-letters t and r, attached to an old-style land-name comprising one or more hill-letters. The river-name will thus have been somewhat of the form Atrebandion, where the n is n1. In names of this kind the only or the latest river-letter in the river-prefix is the river-letter used by the tribe concerned. Now we know that the Atrebates used the hill-letter n2, so the corresponding river-letter, l or m, should be the latest river-letter in the river-prefix. The compound river-name must therefore have been of the form Matrebandion or Latrebandion.  This is borne out by the place-name Nemetacum (probably Nemedacum originally), said to be the name of a hillfort occupied by the Atrebates and located near modern Arras in the Pas-de-Calais region of France. The hill-letter n2 of Nemet/Nemed is used in the inversion-type manner, i.e. it is placed at the front of the existing place-name. One would therefore expect the corresponding river-letter, l or m, to appear at the front of the river-prefix. The people of Matrebandion/Latrebandion would be called the Matrebandes/Latrebandes. This form, with loss or omission of initial L/M and the hill-letter n1, together with the common change  dt, yields the tribal name Atrebates.

12.9  The Continental tribal name Belgae was presumably just adopted by the Romans and applied by them to the inhabitants of a new civitas of the province of Britain, the civitas having its capital at Venta (Winchester). Perhaps the Romans knew that those inhabitants, or their forefathers, had migrated to Britain from that part of the Continent which was inhabited by various tribes to whom the Romans applied the collective name Belgae.

 12.10   The tribal name Regni was presumably invented by the Romans and applied by them to the inhabitants of the kingdom of Cogidubnos, with its capital at Chichester, the kingdom apparently becoming a civitas of the province of Britain on the death of the king.


13   The Selgoves

Ptolemy’s information indicates that the Selgoves  lived east of the Novantae, south of the Damnoni and north of the Brigantes. This information is more appropriate for the region around the inner end of the Solway Firth than it is for the western part of the Tweed basin. The name Selgoves  is clearly based on a topographical place-name in the hill-letters s and l, both of which are seen in Ucselodunum (Uxelodunum) at Castlesteads and apparently in Eburocaslum at Broomholm on the river Esk. In addition one sees the river-letter b (corresponding to the hill-letter s) changed to n  in the river-names Anava (the Annan) and Novius (the Nith). But the chronological order of the hill-letters in that region is first s and then l1, so the name of the tribal centre will have been of the form Seglovion (with gl rather than lg, gl being an old-style element meaning 'steep hill'), the people will have been called the Segloves and they used the hill-letter l1. Ravenna does actually list a Segloes (the v of the ending is missing) under his diversa loca. It would appear that Ravenna has simply misread a tribal name on a map as a place-name.


14   The Votadini

The tribal name Votadini seems quite clearly to be based on a topographical place-name, but the hill-letters are missing. The Vot part of the name means ‘slope of (hill) high’ and the ad part means ‘(hill) summit’, ini just being a name ending as in Ravenna’s Landini. Now Ravenna gives us a river-name Lenda, apparently referring to a tributary of the Tweed, though Lenda is actually a land-name in the hill-letters l and n. And just as Lindum Colonia was close to the river Till (a tributary joining the Witham at Lincoln), so it seems reasonable to conclude that Lenda was close to the river Till (a tributary of the Tweed) in Northumberland. Yeavering, long considered to have been the tribal centre of the Votadini, is only some five kilometres from the river Till, so it is quite possible that the name of the tribal centre included the same hill-letter combination as Lenda. This would yield a place-name of the form Voltandinion. The people of the tribe would be called the Voltandini and this form, with omission of the hill-letters l and n, yields the tribal name Votadini. The Votadini  used the hill-letter l1 and the river-letter t. One sees this river-letter in many river-names in that region, as in Intraum and Antrum (one the Leader, the other the Ettrick) and, changed to d, in Adron (the Whiteadder), the Duabs (the Tweed) and probably in Vividin (the Teviot). Vividin is most probably just a scribal error for Tividin or Dividin. Note that the name Voltandinion would be unusual in having one element referring to the slope of a hill and another to the summit, but in this it is similar to Alvinundo (apparently the Celtic form of Ravenna’s Albinumno) at Welshbury Camp, near Cinderford in Gloucestershire. That hillfort has two enclosures, one on the slope of a hill, the other on the summit. At Yeavering we see a hillfort on the summit of Yeavering Bell and a settlement on the northern slope of the hill. And just a little to the southwest we see a hillfort on the summit of Great Hetha and settlements on the western slope of the hill. The name Voltandinion  would thus be appropriate for either location.


15   The Novantae

We start with Ptolemy’s name for the Water of Luce (in Dumfries and Galloway), the Abravannus. It seems quite clear that this river-name was earlier Abrabandion, a river-name of the kind having a river-prefix, here a compound of the river-letters b and r, attached to a place-name, here BandionBandion is a place-name in the hill-letter n1 and will refer to a hillfort founded in the 3rd or even 4th century BC. The Band element means ‘high hill summit’, so the hillfort will have been on the summit of a high hill close to the river. There is an earthwork close to the river at the village of Glenluce, though it is thought to be of medieval date (see the Canmore website of Historic Environment Scotland). There is also a hillfort a little to the north on Cruise Back Fell (NGR: NX 179 622), the hill dropping down to the east bank of the river. The Water of Luce is the main river nearest to Ptolemy’s Novantarum peninsula, presumably the Rhinns of Galloway, and to his Novantarum promontory, presumably a promontory somewhere on the Rhinns of Galloway (Ptolemy gives the peninsula and promontory the same coordinates). It is helpful here to consider the tribal name Trinovantes, which will be derived from a river-name of the form Trinobandion, this being substantially the same name as Abrabandion, there being merely a slight difference in the river-prefix. The place called Bandion  in the case of the Essex river will have been a hillfort or settlement up on the top of high ground close to the river Pant/Blackwater (Pant will be derived from Bandion). But the important point to note is that the river-prefix Tr  in Trinobandion  comes complete with an in ending. One sees this ending in a number of river-names. It is seen as an after the Tris element of Trisantona and as en after the Derb element of Derbentio. We can therefore be quite confident that the Celtic form of Abravannus  had actually been Abranobandion, with an an ending after the river-prefix Abr. The people who lived around the river would then be called the Abranobandae, which form, with omission of initial Abra and the common changes b → v and d → t, yields the tribal name Novantae. Note that the later river-letter in the river-prefix Abra is r, indicating that the Novantae used the hill-letter m and corresponding river-letter r.


16   The Demetae

The Demet of the tribal name Demetae is a straightforward inversion-type place-name element meaning ‘summit of hill high’. The name indicates that the Demetae used the hill-letter m. One sees this m in Maridunum, the hillfort on Merlin’s Hill, a little to the east of Carmarthen, and indeed Ptolemy assigns Maridunum to the Demetae. The m in Maridunum will thus be the hill-letter m used in the inversion-type manner. The only other name we have for the territory of the Demetae  is Ptolemy’s Luentinum. Unfortunately it is not clear what the Celtic form of this name had been and there is the further complication that we sometimes see an l/m interchange in Celtic names, as in Mugi → Lugi and, in the opposite direction, Lelamon → Melamoni. But at least if the tribal name is correct we can conclude that the Demetae used the hill-letter m and the corresponding river-letter r.


17   The Ordovices

The tribal name Ordovices appears similar to the names Trinovantes, Novantae, Taexali and Atrebates in being based on a river-name of the kind comprising a river-prefix, including one or more river-letters, attached to a land-name, i.e. a place-name including one or more hill-letters. The river-letters in the case of the name Ordovices are r and t (changed to d). Note that these river-letters correspond to the hill-letters m and l in the place-name Mediolanum (Whitchurch), said by Ptolemy to be in the territory of the Ordovices. We can thus see that the Ordovices used the hill-letter l1 and river-letter t. In the case of the four other tribal names mentioned above the land-name element in the compound river-name is an old-style element, so this is presumably also true of the name Ordovices. The element vic was thus most probably bics originally (cf. Becsa at Cockleroy in West Lothian). The hill-letter s in bics corresponds to the river-letter b seen in the river-name Sabrina, the Severn, which river flows through the territory of the Ordovices. The river-name upon which the tribal name Ordovices is based will thus have had a form such as Ortobicsion, the people of that region then being called the Ortobicses. This form, with td, b v and loss or omission of the first s, gives us the tribal name Ordovices.  (For a more detailed account of the development of the tribal name Ordovices see the entry for Mediolano in the Alphabetical List). 


18     In conclusion, following what is said above and in Part 1, we can see that of the Celtic tribes mentioned by Ptolemy:

the Deceangli used the hill-letter s;

the Demetae, the Dumnoni/Damnoni and the Novantae used the hill-letter m;

the Iceni, the Dobunni and the Cornovii  used the hill-letter r;

the Coritani, the Durotriges, the Ordovices, the Selgoves, the Silures,  the Trinovantes  and the Votadini used the hill-letter l1;

the Atrebates, the Belgae, the Brigantes, the Canti, the Catuvellauni, the Regni and the Venicones used the hill-letter n2; and

the Parisi  used the hill-letter l2.


19   Note that Ptolemy's Celtic tribes north of the Antonine Wall are discussed in detail in 'The Celtic "Picts"', which may be accessed from the Main menu above. Leaving aside the Damnoni and the Venicones, who are discussed above, of the Celtic tribes mentioned by Ptolemy:

the Smertae used the hill-letter s;

the Lugi used the hill-letter m;

the Caereni, the Cerones and the Decantae used the hill-letter r;

the Caledoni  and the Taexali  used the hill-letter l1; and

the Carnonacae, the Cornavi, the Epidi and the Vacomagi used the hill-letter n2.


20   Although not relevant for the present page it may be noted for completeness that the Boresti, referred to by Tacitus, and the Verturiones, mentioned by Ammianus Marcellinus, both used the hill-letter r. For the Boresti see 'The Boresti and Mons Graupius', which may be accessed from the Main menu above. 



[This page was last modified on 01 April 2022]


[NB    Regarding the Deceangli attention is drawn to the note at the beginning of ‘Ptolemy’s Celtic tribes: Part 1’].