Gildas’s 28 Cities: Indirect Evidence for the Kingdom of Lindinis
Gildas said early in his famous book De Excidio “Britain has 28 cities”. His so saying is evidence that there was a proto-medieval Kingdom of Lindinis in and around Somerset.
Just as the absence of incontrovertible contemporary evidence leaves a gap in the fence of intelligent perception of history through which the ultra-sceptics can justify their belief that King Arthur did not exist, so a similar absence of direct evidence could enable sceptics to doubt the existence of his Kingdom of Lindinis.
Importantly, what Gildas said is not exactly that Britain has 28 cities but that she has “two groups of ten and two groups of four” cities (civitates). Commentators have casually dismissed this wording as “verbose”, when in fact it is precise. Roman Britain in the fourth century was governed as four Provinces; and it is probable that Gildas was copying a fourth century document – geography was altogether marginal to his writing, and besides, he was not a geographer. It is possible to identify Britain’s 28 Roman cities by identifying ten cities each in two of her four Provinces, and four each in the other two Provinces.
The word civitas had a legal meaning. It meant both a civilian local capital and the surrounding territory governed therefrom. It did not mean an urban settlement without that governmental function, which no matter how populous was an urbs or vicus. Nor did it mean a legionary fortress or a native town or any other place in areas under military rule. Neither Chester (a large Legionary Fortress) nor Caernarfon (a military base) was a civitas, though both were quite thriving towns 200 years after the Roman legions had departed. A civitas was a territory – a governance unit with a designated civilian urban settlement as its administrative capital.
Within each civitas, mileposts gave the distance to its capital – except on major highways. Mileposts in the east Midlands reveal two data of Britain-wide relevance. Smallish areas centred on Buxton Spa and Lincoln were both civitates. It follows that the category of place to which these two towns belonged was included in the list of civitates. Buxton had special status as a Spa (Aquae…); the only other one in Britain was Bath. Lincoln was a colonia, a higher status than a normal civitas. There were five coloniae in Britain: the others were Colchester, London, York, and Gloucester. Deducting five coloniae and two spas from the required total of 28 civitates, it is therefore necessary to identify 21 ordinary civitates to make the total numbers up – ordinary, meaning a territory at least as large as a modern county and governed by civilian Roman citizens from a Roman city legally designated as its capital.
Britain’s northernmost civilian Province was Britannia Secunda, whose provincial capital was York. Historians of Roman Britain have identified three places in the Province that apparently had the governance functions and civilian authority of a civitas capital in the fourth century: Corbridge (near Hexham, Northumberland); Aldborough (near Ripon, Yorkshire); and Carlisle. Together with York, these constitute one of Gildas’s two groups of four.
The group that can next most readily be identified is the ten civitates in the Province of Maxima Caesariensis, the south-east of the island. Along with London and Colchester, six are known with certainty: Silchester (between Reading and Basingstoke); Winchester; Chichester; Canterbury; St. Albans; and Caistor St. Edmund (near Norwich). Another can be identified on circumstantial evidence: Widford (near Chelmsford). The areas of ordinary civitates, as distinct from coloniae and spas, were usually based on areas controlled by tribal kings before the Roman conquest. The tribe in Essex and east Hertfordshire were well known to the pre-conquest Romans, who called them Trinovantes. It is inconceivable that they would not have had their own civitas, separate both from that of their rivals the Catuvellauni based in St. Albans and from that of the Iceni to the north, based in Caistor St. Edmund. Widford, Caesaromagus in Roman times, is the only candidate for its capital.
Nine civitates in Maxima Caesariensis are thus determined. Wacher indicates (in The Towns of Roman Britain, 1997) that the probable location of the tenth is Water Newton, near Peterborough. There was not a separate civitas in the area during the early part of the Roman occupation, but Wacher and others show that a civitas of the Fens is likely to have been carved out following works of fenland drainage. It is indeed quite possible that this was one of a few new civitates created by dividing older ones as part of Diocletian’s local government reforms of 296CE, whose most prominent effect in Britain was to create the division of Britannia for civilian governance into four Provinces.
With clarity on the two Provinces of Secunda and Maxima, we can go on to also identify with a large measure of confidence the civitates of the Province that was between them, Flavia Caesariensis. Buxton Spa and Lincoln Colonia were both in this Province, and undoubtedly Leicester was the capital of an ordinary civitas in the interior. One more civitas must be identified to bring the Provincial total up to the four that Gildas’s words require. The remaining civitas was almost certainly one based at Brough-on-Humber, 11 miles up the Humber estuary from Hull. When the Romans took control of the area that is now East Yorkshire, the British tribe that lived there were the distinctive Parisi. It is highly probable, in view of the patterns in the rest of lowland Britain, that they governed themselves in their own civitas and that, as Wacher suggests, the civitas capital was Brough.
The remaining Province was Britannia Prima, western Britannia south of the Mersey. Gildas’s count requires it to have had ten civitates. Two have been mentioned above as places with special status, the small civitas territories for Bath Spa and Gloucester Colonia. Half a dozen more are definitely known: Exeter, the civitas for the people of the far south-west whom the Romans called Dumnonii; Dorchester, for the Durotriges; Cirencester, for the Dobunni; Wroxeter; Caerwent, for the Silures; and Lampeter, for the Demetae. Lindinis, it is clear to most commentators, became a civitas with its capital at Ilchester (Roman Lindinis City) in Diocletian’s reforms, formally in 296 though in reality not till several years later. Like Water Newton, Lindinis was carved out of a pre-existing larger civitas, in its case that of the Durotriges. Its status is most convincingly evidenced by two inscriptions on Hadrian’s Wall.
As with the Trinovantes and the Parisi, the best indication for Prima’s remaining one civitas, for whose existence the historical record is less sure than for the other nine, is to locate an area with a tribal identity. It has to be in the civilian zone of Roman administration; places in militarily controlled upland Wales cannot be considered as possible locations for the missing civil government-status civitas.
Milestone evidence points to Caerhun, on the Conwy estuary in north Wales, as a possibility. Archaeological evidence, however, demolishes that hypothesis. Caerhun might conceivably have been a civitas capital when milestones with distances to Caerhun were erected in c129CE and c204CE – though it might not, for there is no clarity that it was a civil district even then. But Caerhun was burnt and left derelict soon after 204, and only sporadically occupied thereafter, even by the army. It cannot have been a civitas capital in the fourth century, not even on an official document of legal rather than practical meaning.
The tribal area that does have credibility for having been made into a civitas by Diocletian is in and around Herefordshire. The evidence that this was indeed done is thin, yet favourable. One piece is a Roman milestone that was found at Kenchester. The vital letter carved on it, revealing the initial letter of the name of the civitas Kenchester was in, could be a D; if so, Kenchester was in the civitas of the Dobunni. But Professor Collingwood, and others, reads the letter as a B. Could Kenchester have been in a civitas whose name began with B? Bring in ‘Exhibit B’, the Ravenna Cosmography. As indicated in the discussion of Corbridge above, places in Britain listed in the Ravenna Cosmography in the form of a place name (noun) followed by that of a tribe in the genitive plural, as for example Isca Dumnoniorum (Exeter), were civitas capitals – unless they were in a military district. The one place in Prima named there in this format, in addition to its definitely known civitas capitals, is Brano Genium – Leintwardine (near Ludlow, but over the county boundary into Herefordshire).
If ‘genium’ here is the common noun, it translates as ‘of the tribes’. Brano Genium is first recorded by Ptolemy. The apparent meaning is ‘city of the tribes [devoted] to Bran’. It also implies that, for Ptolemy, ‘to Bran’ specified the tribe(s) to which it belonged. In truth, perhaps many British tribal warrior-kings could have identified themselves as ‘devoted to Bran’: Bran was a major Celtic war god. But only the Herefordshire tribes did so to the adventurous visiting foreign geographer Ptolemy; and the name stuck.
Such a description indicates a nodal place, as a place chosen for a civitas capital would be likely to be. Certainly the site of the hillfort above Brano Genium still bears his name: Brandon (i.e. Bran’s dun, Bran’s Castle). It could be, then, that the name of the putative civitas based at Brano Genium was a Latinization of ‘Bran’s civitas’: beginning with a B.
The full list of Gildas’s 28 civitates, in his two groups of ten and two groups of four, can be completed with Leintwardine, and tabulated:-
Latin Name Nearest modern place Nearest modern town
(as given on the (where different)
Ordnance Survey map)
Coriosopitum Corbridge Hexham
Isurium Brigantum Aldborough Ripon
Caesaromagus Widford Chelmsford
Calleva Atrebatum Silchester Basingstoke
Durobrivae Water Newton Peterborough
Durovernum Cantiacorum Canterbury
Londinium Augusta London
Noviomagus Regnorum Chichester
Venta Belgarum Winchester
Venta Icenorum Caistor St. Edmund Norwich
Verulamium St. Albans
Aquae Arnemetiae Buxton
Petuaria Brough-on-Humber Hull
Ratae Corieltavorum Leicester
Aquae Sulis Bath
Bravonium Leintwardine Ludlow
Corinium Dobunnorum Cirencester
Isca Dumnoniorum Exeter
Venta Silurum Caerwent Chepstow
Viroconium Cornoviorum Wroxeter Shrewsbury
So what happened to all these civitates during the fifth and sixth centuries, after Roman Emperor Honorius had memorably pulled his occupying troops out and told them in 409 to take care of their own defences?
Most became kingdoms. Each of the known early English kingdoms, the ones that Gildas called “our enemies in the east”, can be correlated with a prior Roman civitas. Without any recorded battles, all of the civitates of Flavia became English-ruled kingdoms. Petuaria became Deira; Lindum, Lindsey; the civitas of the Corieltavi developed into (Old) Mercia; and, with English rule not coming until much later, the hillcountry governed by Aquae Arnemetiae became the client kingdom of Mercia identified in the Mercian Tribal Hidage of c672 as Pecsaetna (the People of the Peak District).
Further south, in Maxima, the English kingdom of East Anglia was heir to the civitas of the Iceni. The kingdom of the East Saxons, which came to be called Essex, apparently combined the small territories of London and Colchester with the normal, two county size, civitas of the Trinovantes. The kingdoms of Kent and Sussex inherited the Canterbury and Chichester civitates, respectively, and the kingdom of the Gewissae in the upper Thames valley grew out of the Silchester civitas. The Romano-British, on the other hand, retained control of much of the civitas of the Catuvellauni until 571, though probably not St. Albans itself. Their territory had sufficient autonomy still a century later to be recorded in the Mercian Tribal Hidage as a separate entity, Cilternsaetna.
The Water Newton civitas does not feature in recorded fifth or sixth century history, but it too shows up in a distinctive colour in the Mercian Tribal Hidage of 672. The Hidage registers two small territories in the Fens, apparently, like Cilternsaetna, as client to but not part of Mercia: North and South Gyrwa.
There was no such English intrusion into Secunda or Prima Provinces. In Secunda, the British kingdoms that emerged into the historical record begin as successor states to the Roman civitates – but with the difference that Hadrian’s Wall ceased to be a political barrier. So the kingdom of Rheged included the lands north of the Wall known today as Dumfries-and-Galloway and maybe much of Ayrshire, as well as the Carlisle civitas. Likewise the kingdom of the Gododdin ruled the east coast going north from the Wall as far as Edinburgh, as well as the Corbridge civitas in Durham and south Northumberland. The Aldborough civitas and the York colonia combined into the Kingdom of Elmet.
In Prima, later history attests abundantly to the kingdom of Kernow, heir to the Roman civitas of Dumnonia and continuing to be called Dumnonia in Latin documents. Similarly the Silurian civitas became the Kingdom of Gwent; the Demetian civitas, the Kingdom of Dyfed; the Wroxeter civitas, Powys; the territory of the Cirencester civitas along with Gloucester and Bath re-emerges into the historical record in the late sixth century as the Kingdom of the Hwicce.
In the areas that in the fourth century were the civilian-governed parts of Prima, there is one other kingdom flourishing in the sixth century: Ergyng. Its territory was between the Wye and the Severn, including Herefordshire and the Forest of Dean. Ergyng is the British name for the Roman industrial town near Ross-on-Wye called Ariconium in Latin, nowadays sporting the characterful name Weston-under-Penyard. The kingdom’s territory is that of the suspected Brano Genium civitas. Its name suggests that perhaps its rulers chose a hillfort near Ariconium as their principal kaer (castle) in the fifth century, which became as important as Brandon, the hillfort above Leintwardine, and Magnis (Kenchester) from which the area’s later Saxon name, Magonsaetna, was derived.
In all, 25 of Gildas’s 28 civitates surface as identifiable fifth and sixth century kingdoms. Five of the seven little special-status civitates merge with their principal civilian neighbour, the two exceptions being those in Flavia, the only Province in which no war fighting for kingdoms is recorded. Of the 21 ordinary civitates, 16 re-emerge by the sixth century as early-medieval kingdoms, either British or English; one as divided into two (the Gyrwa); and one – that of the Catuvellauni – as a British kingdom that had almost certainly lost part of its territory to an English neighbour.
The only civitates absent from this consistent pattern are those whose capitals were Dorchester, Winchester, and Ilchester. Of the governance of the Dorchester civitas, nothing is known. According to White (Britannia Prima), archaeology shows that Bokerly Dyke guarded its eastern frontier against Winchester, that the local elite moved from Roman Dorchester to the adjacent hillfort of Poundbury, and that their domestic architecture showed early English influences. The conjecture that this civitas, like at least six others in Prima, became an independent British kingdom ruled primarily from Poundbury, is logical, but not proved.
Archaeology suggests, again according to White, that the Winchester civitas became English-ruled during the fifth century. References in later English sources imply that the area was Jutish-controlled; this is supported by the designation in the Tribal Hidage of a medium-sized area, listed after Wihtgara (the Isle of Wight), named Ohtgaga (i.e., surely, Jute-gaga: Jutes’ Land), and a larger territory in the same general area as the unidentified but perhaps related Noxgaga. So the Winchester civitas perhaps became divided into three kingdoms, Wihtgara, Ohtgaga, and Noxgaga, all under Jutish rule.
So what of Lindinis? Every other ordinary civitas that remained under British governance became a kingdom, as also did most of those that became English-ruled. Every kingdom in the area of civilian governance during Roman times apparently corresponded directly to a civitas (after adjusting for the Hadrian’s Wall effect). It would be weird indeed if governance in the Lindinis territory had not developed similarly. Having identified Gildas’s 28 civitates, and the proto-medieval kingdoms that emerged from every one of the others that remained under British control, the assumption that there was likewise a Kingdom of Lindinis as the fifth-century successor state to the Lindinis Civitas is the only rational possibility.
It is a step beyond this to say that Lindinis was King Arthur’s kingdom, or that it became mistily remembered as the Kingdom of Lyonesse of later legends. On other posts here, I am showing why I hold these two propositions to be true.
 “bis denis bisque quaternis ciuitatibus” (De Excidio, chapter 3)
 Civitates whose status as such is not disputed, including Aldborough and Carlisle, are identified by C.E. Stevens (1937. Gildas and the Civitates of Britain: English Historical Review 206: 193-203), by J. Wacher (1997. The Towns of Roman Britain: Routledge), and on the Ordnance Survey map of Roman Britain.
Corbridge’s status is attested by its presence in the Ravenna Cosmography in the format regularly used there and elsewhere for civitas capitals, the place name being followed by that of the appropriate tribe in the genitive case, Corie Lopocarium; and by its absence from the Notitia Dignitatum, which lists military units and their stations including those guarding nearby parts of Hadrian’s Wall.
 Caesaromagus is “the one British place-name of a type which is not uncommon in Gaul, in which a Celtic suffix is combined with an attribute of the emperor (Julio-, Caesaro-, Augusto-). In Gaul such names…… are …… almost invariably tribal capitals” (C.E. Stevens, 1937, p198).
 A fifth Province of Valentia was created, for the area north of Hadrian’s Wall. It was wholly under military rule.
 R. White (Britannia Prima: 2007, p198) does not put Brough-on-Humber in Flavia Caesariensis. He draws boundaries of the four Provinces on the hypothesis that the commanders of each Military Province invited ‘Saxon’ mercenaries from a different Continental homeland, differences revealed by the distribution of varying types of Germanic brooches. However, boundaries placing East Yorkshire in Flavia, and Norfolk in Maxima, improve the brooch evidence for White’s hypothesis compared with his boundaries. The coast of Flavia as amended makes military sense, being all of, and solely, the part of the east coast unfortified in the 4th century: from south of Bridlington to the Wash.
 The two areas of Britannia south of Hadrian’s Wall that were under military rule in the fourth century were north and mid-Wales, and Lancashire (at least between the Mersey and the Lune).
 See P.K. Baillie Reynolds, 1938. Excavations on the Site of the Roman Fort of Kanovium at Caerhun, Caernarvonshire. Cardiff: Kanovium Excavation Committee.
 There are, for example, a Roman ‘Saxon Shore’ fort of Brancaster on the north coast of Norfolk; and a village of Branscombe on the coast of east Devon.
 Leintwardine is sometimes known as Bravonium, reserving Brano Genium for Brandon, the hillfort on the hill overlooking the Roman city.
 The ending –saetna indicates a territory with British Celtic inhabitants, never settled by English immigrants, but under English suzerainty.
 It approximately comprised the areas that became Essex and Middlesex.
 Much later, in the second half of the sixth century, inspired by King Urien and his son Owain, the historical poems of Taliesin celebrate Rheged’s successes in expanding eastwards to conquer much of North Yorkshire and almost overrun English Bernicia on the Northumberland and Durham coast.
 I have used the word ‘English’ to mean any or all of Britain’s Germanic newcomers. Jutes were one of the fifth century’s groups of English warriors and immigrants, as also were Angles, Saxons, and Frisians.