[NB. The reader will probably find it easier to understand this page if he or she has already acquainted himself or herself with the contents of Chapters 1 and 2. Chapter 1 concerns the basic building-blocks used in Romano-British place-names and Chapter 2 explains how those building blocks were assembled by the Celts to form compound names.]
Leucomago to Salinis (65)
From the Thames valley to South Wales and the west of England
1 This chapter is concerned with the Ravenna names from Leucomago (45) to Salinis (65) and with Ptolemy’s names in the same region. Leucomago appears to have been at Leckhampstead, lying on a straight line between Cunetzione (46) at Mildenhall and the important crossing of the Thames at Streatley, opposite Goring. The mag element of Leucomago is an inversion-type element meaning ‘hill steep’, so presumably leuc is exactly the same element in the hill-letter l. Leckhampstead, however, does seem an odd place to start a journey to South Wales, so perhaps the compiler of Ravenna did in fact start at Streatley but did not give the name of the Roman post there because he planned to include it later, either as Tamese (76) or as Landini Tamese (75, 76). Cunetzione at Mildenhall is well-known, so one can go straight on to the next name, Punctuobice (47) at Sea Mills on the Avon. Punctuobice is a simplification of Abone traiectus vicus. One can rewrite this as [A] bon [etraie] ctusvicus and then delete the parts in brackets, thus leaving Bonctusvicus. It is this form which changed to Punctuobice. Note that Traiectus is the name immediately after Abone in Iter XIV of the Antonine Itinerary.
2 Ravenna then takes us across the Severn estuary to Ventaslurum (48) at Caerwent and then lists Iupania (49), Metambala (50) and Albinumno (51). These names are not clear, but it would appear that they can only have been the names of places spaced apart along the western side of the Severn estuary. The names Iupania Metambala in Ravenna will originally have been Luba Nemedonbala, where Luba was in fact the then name of the river Wye. Luba will have changed to Luva at some point during the Romano-British period and then later Anglo-Saxon settlers will have taken the initial L to be the letter they knew 'the' Celts had used to signify 'river' (cf. the river Alavna at Stratford-on-Avon, the proper name of the river being taken by Anglo-Saxon settlers to be Avna, now the Avon) so that for them the proper name of the river was Uva, this then being anglicized to Uwa, and this form will be the origin of the modern river-name Wye. The original river-name Luba will have been transferred by the Romans to a fort which they built on or close to the Wye, probably to control a crossing of the river and most probably at Chepstow. Note that Luba Nemedonbala was wrongly divided at some stage of medieval copying to give Lubane Medonbala and Lubane, with the minor changes L → I and b → p, changed to Iupania. Nemedonbala is a topographical compound in the hill-letters n, m and l where the old-style elements med and bal respectively mean 'hill summit' and 'high hill'. The river-letters corresponding to the hill-letters n and l of Nemedonbala are l (for minor rivers) and t, and these both appear in the river-name Lyd with the t changed to d. Nemedonbala was thus the Celtic hill-fort at Lydney. There must have been n/m confusion at some stage of medieval copying and the initial Ne was lost to Luba, so that Nemedonbala became Medambala which, with the minor change of d → t, became Metambala. Albinumno is somewhat obscure, but the Celtic name was probably of the form Alvinundo, combining old-style alv, meaning 'hill slope', and old-style und, meaning 'hill summit'. This unusual name is entirely appropriate for the hill-fort now known as Welshbury Camp, to the northeast of Cinderford in Gloucestershire. This hill-fort comprises an earlier enclosure standing mostly on the slope, on the hillside, and a later enclosure built on the summit of the hill. It is not clear whether Ravenna's name refers to the hill-fort itself or to a Roman fort built somewhere in the vicinity and to which the name of the hill-fort was transferred. Note that the Celtic name Alvinundo may have had an initial B meaning 'high' or C meaning 'steep'. Assuming the name had been Alvinundo the v changed to b (a fairly common change in place-names) and undo changed to unno (cf. Gabaglanda → Amboglanna). Then, as a result of medieval m/n confusion, unno changed to umno.
3 After Albinumno Ravenna lists Isca Augusta (52), Bannio (53), Bremia (54), Alabum (55) and Cicutio (56). Isca Augusta and Bannio present no difficulty – they were respectively at Caerleon and Abergavenny. But earlier writers have assumed that Bremia was the fort at Llanio, on the river Teifi to the north of Lampeter, and that after listing Bremia the compiler of Ravenna then listed names in the reverse direction, coming back down the Usk valley to Cicutio, thought to have been at Y Gaer, just to the west of Brecon. So far as can be seen the compiler of Ravenna did not behave in such an odd manner anywhere else in the British section of the Ravenna Cosmography, so there is no need to believe that he acted in that manner here unless there is some compelling reason for identifying Llanio as Bremia. But the identification rests entirely on the supposed relationship between the name Bremia and the river-name Brefi. However, the Llanio fort stands on the west bank of the Teifi and the Brefi is a minor river coming in from the east to join the Teifi at a point about a mile south of the fort. It is therefore in the highest degree unlikely that that minor river played any role in the naming of the Llanio fort. It seems much safer, much more sensible, to assume that Ravenna lists names going up the Usk valley and that when it reaches Abergavenny it just carries on up the valley, heading west. In this case Bremia, which is just a compound in the hill-letters r and m, the initial br meaning ‘high hill’, will have been the fort at Pen-y-Gaer and Alabum the fort at Llandovery. One will note that there is no name here for the fort at Y Gaer, but there is a very simple reason for that. The river at Y Gaer is called the Yscir, the ysc element being a land-name meaning ‘hill steep’. There must therefore have been on that river a Roman fort with a name including such an element. But Ravenna hasn’t listed such a name earlier, so, following usual practice, it must list it later, and the only possible candidate is Ypocessa (60), which is assumed here to be the same place as Epocessa (59). Now there are in the British section of the Ravenna Cosmography about nine other names having an essa-type ending and in all those cases the part of the name before the ending includes either the hill-letter s or the corresponding river-letter b. The p and c of Ypocessa are the adjectives b, changed to p, meaning ‘high’ and c meaning ‘steep’, so it seems quite clear that the hill-letter s is missing – the name ought to be Yposcessa or Ypocsessa. Whatever the correct spelling it is this name which was transferred by the Romans to the river now called the Yscir, the r of this name most probably being the river-letter r applied by settlers who used the hill-letter m. That only leaves Cicutio to identify. The name is not clear, but it would appear that there is one hill-letter missing after the initial Ci and another after the cu, the name then being a compound of an old-style element followed by a transitional element, as in Gabrocentio (117) and Concanata, the latter being apparently the original Celtic spelling of the Congavata of the Notitia Dignitatum. However, Cicutio is somewhat similar to the river-name Cothi and so it seems reasonable to assume that Cicutio was a fort on the Cothi, presumably the fort at Pumpsaint, the Romans having transferred the name to the river.
4 As soon as one has identified Y Gaer as Ypocessa the names around Magnis (57) at Kenchester fall into place. Between Magnis and Ypocessa is Branogenium (58) which was presumably the fort at Clyro/Hay-on-Wye. The gen element is presumably just an old-style topographical element meaning ‘steep hill’ and Bran may originally have been Brav, this being an old-style element meaning ‘high hill slope’ (it is a banna-type element but using the hill-letter r rather than n). The fort at Clyro is indeed built on a hillside, on the slope of a hill, so there can be no doubt as to the identification. Now Ravenna lists Macatonion (61), and Glebon Colonia (62) at Gloucester. It is quite clear here that Ravenna is using Magnis as a node, so that Macatonion was somewhere between Kenchester and Gloucester. The obvious candidate is Dymock. The macat of Macatonion is an inversion-type name-element meaning ‘hill steep high’. The element exists in other hill-letters, too, for example in the hill-letter l in Lucotion (170). Presumably the letters of macat have simply been rearranged and the t changed to a d, and the c to ck, to yield the modern name Dymock. But note that it is much more likely that the Romans transferred the name to Dymock from the Iron-Age hillfort now called Haffield Camp (SO 723 339), about four kilometres northeast of Dymock. Ravenna does sometimes omit the initial consonant of Celtic place-names, so it is possible that the Celtic name of the hillfort was Demacatonion, meaning 'summit of hill steep, high'. Modern Dymock will then be derived directly from the Demac part of the old name. After Glebon Colonia Ravenna lists Argistillum (63), Vertis (64) at Worcester and Salinis (65) at Droitwich. It is quite clear that Argistillum was between Kenchester and Worcester and not between Gloucester and Worcester, since if the latter had been the case then the order of names in Ravenna would have been Ypocessa, Branogenium, Magnis, Macatonion, Glebon Colonia, Argistillum, Vertis and Salinis. But that is not the order of the names, so the order actually given makes it quite clear that Magnis is used as a node and that Argistillum must have been between Kenchester and Worcester. It was presumably the Roman fort at Stretton Grandison. If Argistillum is a topographical name of the kind discussed in this study, then arg is an inversion-type element meaning ‘hill steep’ and st an inversion-type element meaning ‘hill high’. But note that there may be a slight problem regarding Branogenium in that it may not have been the Roman fort at Clyro/Hay-on-Wye (see the entry for Branogenium in the Alphabetical List). In this case it is best to equate Ravenna’s Branogenium with Ptolemy’s Brannogenium at Leintwardine. What this would mean is that in listing the names around Magnis at Kenchester Ravenna first goes north to Branogenium at Leintwardine, then swings anticlockwise to list Ypocessa at Y Gaer, near Brecon, then swings further anticlockwise to list Macatonion at Dymock and Glebon Colonia at Gloucester, then swings still further anticlockwise to list Argistillum (perhaps Stretton Grandison), Vertis at Worcester and Salinis at Droitwich. There would thus be no name for the fort at Clyro, but this may not be a problem. That fort is generally considered to have been occupied only in the AD60s, and it is possible that the group of Ravenna names around Magnis at Kenchester dates from a later period. Indeed the presence of Colonia after Glebon suggests that the group cannot be earlier in date than AD96, this being the earliest date on which Gloucester can have been given the status of colonia.
5 That completes the list of Ravenna’s names in the region under consideration in the present chapter. Ptolemy lists very few names in this region. He does refer to a Bullaeum in the territory of the Silures, which is presumably the Burrio of the Antonine Itinerary and was therefore at Usk. He also mentions Brannogenium and Mediolanum in the territory of the Ordovices. Mediolanum was at Whitchurch, Brannogenium at Leintwardine. Ptolemy also lists Luentinum and Maridunum, further west, in the territory of the Demetae. Maridunum is no doubt the Muridunum of the Antonine Itinerary and so was at Carmarthen. There is no problem with the dunum element of the name – that is just an inversion-type element meaning ‘summit of hill’, but the m and r combination, no matter whether spellt mar or mur, could be a hill-letter compound as in Moriduno (Hembury) or a river-letter compound as in Ptolemy’s Moricambe estuary. Carmarthen does stand on the river Towy, but since the fort was actually built on the top of a hill it does seem more likely that mar is a hill-letter compound, Ptolemy’s Maridunum then being in fact the same name as Ravenna’s Moriduno at Hembury. But note that Maridunum will originally have been the hillfort on Merlin's Hill, a little east of Carmarthen, the name simply being transferred to the Roman fort/settlement at Carmarthen when this was built. Luentinum is more of a problem and no suggestion can be made here as to its identification. Others have suggested that it was at Pumpsaint or Loughor. No comment can be made here regarding Pumpsaint, other than that the fort there appears actually to have been called Cicutio. Ptolemy’s Luentinum is not likely to have been the same place as the Leucaro of the Antonine Itinerary, which will have been on the river Loughor somewhere in the Fforest/Pontardulais area. The river Loughor itself will have been the Leucarosena, probably earlier the Letucarosena. This is a river-name of the kind having a river-element, here sena, attached as a suffix to a place-name, here Letucaro. The river-letter s in the sena element corresponds to the hill-letter r in Letucaro. The t and ar of Letucarosena have been dropped and the c has undergone the common change to g¸ so that Letucarosena became the Leugosena of Ravenna.
6 The names of the Roman forts discussed in this chapter are shown on the two maps reproduced below. The forms and the numbering of the names are those given by Richmond and Crawford. The identifications of the place-names are:
45 Leucomago = Leckhampstead (fort as yet not discovered)
46 Cunetzione = Mildenhall
47 Punctuobice = Sea Mills
48 Ventaslurum = Caerwent
50 Metambala = Lydney
52 Isca Augusta = Caerleon
53 Bannio = Abergavenny
54 Bremia = Pen-y-Gaer
56 Cicutio = Pumpsaint
57 Magnis = Kenchester
58 Branogenium = Clyro or, more probably, Leintwardine
59 Epocessa = Y Gaer, west of Brecon
61 Macatonion = Dymock
62 Glebon Colonia = Gloucester
63 Argistillum = Stretton Grandison
64 Vertis = Worcester
65 Salinis = Droitwich
Names not shown on the maps:
49 Iupania = apparently Chepstow
51 Albinumno = apparently Welshbury Camp, northeast of Cinderford
55 Alabum = Llandovery
Note that Branogenium is shown on the maps as having been at Clyro, but it may alternatively have been at Leintwardine, hence north of Magnis.
[This page was last modified on 02 April 2022]