Heroic Age Logo */The Heroic Age/, Issue 1, Spring/Summer 1999*

------------------------------------------------------------------------


GILDAS AND THE CITY OF THE LEGIONS


by


P.J.C. Field <mailto:[e-mail]?subject=Gildas>,
University of Wales, Bangor

In Gildas's /De excidio Britanniae/, he says the spiritual life of
Britain had suffered because the partition (/divortium/) of the country
was preventing the citizens (/cives/) from worshipping at the shrines of
the martyrs.^1 <hagcl2.htm#note1> He names three such martyrs, all of
whom he thought had been put to death during the Diocletian persecution
(in effect in Britain from A.D. 303 to 305), although he is hesitant
about the dating. Gildas can be very inaccurate about events occurring
more than a century before his own time, and in this case his hesitation
seems well founded: It is likely that all three died earlier and not
necessarily at the same time (Frere 1987:321 <hagcl2.htm#frere>). The
martyrs were St Alban, whose shrine was at Verulamium, and St Aaron and
St Julius, whose shrine was at the city of the legions (/urbs
legionum/). Strictly speaking, Gildas says only that Alban came from
Verulamium and that Aaron and Julius were citizens of the city of the
legions, but if the shrines had been in different places we can be sure
that he would have said so. He was after all concerned not with the
geographical origins of his three saints but with access to their
relics.^2 <hagcl2.htm#note2>

Bede adds nothing to what Gildas says about Aaron and Julius. After a
very full account of St Alban?s martyrdom, his /Historia/ gives the
other two only a skimpy little tailpiece, based on the passage in Gildas
and echoing its wording (Bede 1.7; Colgrave and Mynors 1991
<hagcl2.htm#colgrave>). Bede used the same passage for the entry about
Aaron and Julius in his /De temporum ratione./ This work, in turn, was
used by various continental chroniclers and martyrologists and by the
compiler of the entry for the two saints in the expanded continental
version of Bede?s /Martyrology,/ put together by Florus of Lyon in the
ninth century, in which their names immediately follow that of St
Alban.^3 <hagcl2.htm#note3> Taken in isolation, the fact that Florus
gave the two saints a feast day (22 June) might suggest that he used an
independent tradition that gave a date for their martyrdom, but, since
St Alban was already on Florus?s list under that date, it is more likely
that he was simply lumping two obscure British saints together with a
better-known compatriot. The two saints also seem to have been known in
Ireland from an early date. An early commentator, in an Irish
martyrology of a generation before Florus, notes against the feast day
of a much better-known Aaron (July 1), the brother of the Old Testament
patriarch Moses, that there was a St. Julius "i nAlbain".^4
<hagcl2.htm#note4>

The identity of Gildas's city of the legions is not immediately obvious.
The phrase /urbs legionum/ is not known to have been used by any other
writer in Roman Britain either as a place-name in the strict sense or as

a recognised description of a place. It is recorded in the Middle Ages,
but almost all the known instances seem to derive from Gildas.^5
<hagcl2.htm#note5> The place Gildas intended must be somewhere in the
part of Britain that he knew, which certainly stretched from the Straits
of Dover in south-east Britain round to Cornwall in the south-west and
Gwynedd in north-west Wales but which may have extended beyond that.
Even the meaning of the phrase is not quite certain. Depending on the
relationship between its two terms, it could mean a legionary camp large
enough to be a kind of city or a city with legionary associations.
Modern scholarship has taken Gildas to have intended the former but the
latter has never been excluded, and we will have to return to it. The
former sense would present fewer problems. The part of Britain that
Gildas knew contained many legionary encampments, but he knew it well
enough to have known the difference between lesser camps and the major
permanent fortresses, of which Britain had three. If his /urbs legionum/
was a legionary camp, it must have been one of the three fortresses,
despite the lack of evidence that any phrase like his was ever used of
any of them (Rivet and Smith 1979:s.v. Deva, Eburacum, Isca, and passim
<hagcl2.htm#rivet>).

Two factors suggest that Gildas meant the fortress at Caerleon-on-Usk.
It was near the centre of his area of interest, and there is solid
evidence of a shrine dedicated to Sts Aaron and Julius there from the
ninth century onwards.^6 <hagcl2.htm#note6> That shrine may even explain
why these two obscure and very British saints are unexpectedly invoked
among the thousands of glosses in Irish and Latin in the marginalia of
the well-known ninth-century manuscript of Priscian's /Institutiones
Grammaticae/ at the monastery of St Gallen in Switzerland. One
explanation would be that Irish monks taking the manuscript to
continental Europe travelled by way of Caerleon and that some event
there that we can no longer guess at caused them to write an invocation
to the local saints into their book.^7 <hagcl2.htm#note7>

In the twelfth century, Geoffrey of Monmouth?s mendacious but hugely
influential /Historia regum Britanniae/ made the Caerleon identification
part of the accepted history of Britain. Geoffrey mentions the City of
the Legions more than once.^8 <hagcl2.htm#note8> (The phrase can be
given initial capitals straight away in this case because it is so
obviously a place-name.) He says that the City of Legions was on the
River Usk, implying that it must have been Caerleon, and in a passage
that echoes what Gildas says about the martyrs, he adds that in King
Arthur?s day the city had two principal churches, one dedicated to St
Aaron and the other to St Julius. Geoffrey knew Caerleon personally, but
this looks like a characteristic stroke of his imagination, creating
another bit of the millennia of pseudo-history he needed out of the
passage in Gildas, the obvious etymology of the place-name Caerleon, and
the local cult of the two martyrs. The two churches in particular look
like inventions for the sake of his story: The evidence of the cult,
such as it is, suggests a single church dedicated to both saints, but
Geoffrey wanted a church of canons as the metropolitan church of
Wales.^9 <hagcl2.htm#note9> and a church of nuns to which Guenivere
could flee from Arthur?s wrath. What Geoffrey said about the two saints
found its way into the work of many other mediaeval historians (Tatlock
1950:235 <hagcl2.htm#tatlock>), but nothing testifies more strongly to
its seductive power than the effect of this part of his book later in
his own century on Gerald of Wales. Gerald (Dimock 1861-1891: Vol. 6
<hagcl2.htm#dimock>; Thorpe 1976 <hagcl2.htm#thorpe1>), too, had seen
Caerleon for himself, but, into the lively details of what he saw, he

interpolated a description of the two lost churches made up of a mosaic
of snippets from Geoffrey?s /Historia/.^10 <hagcl2.htm#note10> He did
this despite reporting in the same chapter the well-known tale of how,
when a local man afflicted with unclean spirits had St John?s Gospel
placed on his lap, the evil spirits all flew away "like so many birds",
but, when Geoffrey?s book was put there instead, the demons would return
and "alight all over his body, and on the book too."

The post-mediaeval period has seen almost universal agreement that
Gildas's /urbs legionum/ must have been Caerleon, although cautious
scholars have sometimes noted for the record that he might have meant
Chester, whose legionary fortress was at or just beyond the north-east
edge of his apparent area of interest. There are one or two passages in
the works of later authors that could be taken as supporting Chester,
notably one in Bede?s /Historia/ where he describes the place as
/civitas legionum/, a phrase very similar to Gildas's in form and
meaning (Colgrave and Mynors 1991, 1:2 <hagcl2.htm#colgrave>). We will
return to this.

David Dumville has pointed out, however, that taking the city of the
legions as Caerleon-on-Usk is at odds with what seems to be implied by
the rest of the literary and archaeological evidence for the period
(Lapidge and Dumville 1984:78, 82 <hagcl2.htm#lapidge>), and much the
same is true of Chester. It is difficult to think why "the citizens"
should have been unable to make their way to either place when Gildas
was writing, unless much more of western England and Wales was then
permanently in Anglo-Saxon hands than any historian has suggested. This
is a serious matter. Whatever mistakes Gildas may make about the history
of Britain in earlier generations, his /De excidio/, with all its
limitations, is much the best historical source for the history of
Britain in his own lifetime, and its testimony is in principle capable
of outweighing all the rest of the evidence put together. Dumville was
right to say, therefore, that this is a problem that urgently demands a
solution: If Gildas means what he has been taken to mean, a great deal
of fifth and sixth-century British history needs rewriting.

The most recent attempt at the problem ingeniously suggests the
citizens' difficulties were with a Saxon settlement in the upper Thames
valley (Higham 1994:103-6 <hagcl2.htm#higham>), but whatever
difficulties a settlement in that area might have created for British
pilgrims coming to Caerleon from the east, it would not have caused the
kind of comprehensive exclusion that Gildas implies from Caerleon, and
still less of course from Chester. There is, however, a simple solution:
that the /urbs legionum/ is the third major permanent legionary fortress
in Britain, York.

That identification, unlike that of Caerleon or Chester, is entirely
consistent with the rest of what is believed about the period, although
no exact date can be given either for the completion of Gildas?s book or
for the time when York fell permanently into Anglo-Saxon hands.^11
<hagcl2.htm#note11> The time when Gildas wrote has been the subject of a
great deal of dispute, but the dates most commonly supported are close
to A.D. 540. York fell to the Anglo-Saxons in the course of the westward
expansion of a settlement in the east of what was later to become the
Kingdom of Deira and, later still, Northumbria. The conquest of York
must have been one of the decisive moments in that expansion and might
even have been what prompted the expanding community to declare itself a
kingdom. If the /urbs legionum/ is York, then the loss of the shrine of

Sts Aaron and Julius might have been very recent when Gildas wrote and
all the more bitter for that.

It may help to test this hypothesis if we consider how Gildas expected
his readers to recognise his "city of the legions". He might have
assumed they would know where the shrine of Sts Aaron and Julius was; if
so, his words would mean "that particular legionary city (of the three
possible candidates) where you know these martyrs are buried." Yet it
would be simpler, more normal, and sustain a parallel (Gildas liked
parallels) with what he appears to be saying about Verulamium if he were
offering his "city of the legions" as a place better-known than its
shrine, so that those readers who did not know where the shrine was
could locate it from the city.^12 <hagcl2.htm#note12> If so, /urbs
legionum/ will be the equivalent of a name, a reasonably well-known and
unambiguous descriptive phrase. Such a phrase should fit only one place,
and, therefore, if it means what it has always been taken to mean, one
and only one of Britain's three legionary fortresses. Surprising though
this may seem, it does so; and the place that it fits is York.

We may take the two words in Gildas's phrase separately. The first,
/urbs,/ is a civil, rather than a military, term, and York, in addition
to being the military headquarters for northern Britain, was a Roman
/colonia/ and, therefore, in Roman administrative terms, a city.
Moreover, it was no ordinary city but a major administrative centre:
Since the Romans divided Britain into two provinces in the first century
A.D., York had been the administrative capital of the north and so the
most important city in Britain after London (Mann 1961:316-20
<hagcl2.htm#mann>; cf. Frere 1987:242-43 <hagcl2.htm#frere>). It could
even be called a metropolitan city, both because it was the chief city
of its province and also in something of the modern sense as well. Also,
in the late empire, /urbs/ was sometimes used to describe cities that
fell into one or both of those categories, in the latter case by
extension from a secondary sense of /urbs/ as /the/ city (i.e., Rome;
Lewis and Short 1886 <hagcl2.htm#lewis>).^13 <hagcl2.htm#note13> When a
writer of the period wanted to speak of an ordinary city as opposed to a
major city, he would be likely, particularly if his Latin was good (and
Gildas?s was excellent), to choose a word with different connotations
from /urbs/. The word /civitas/ was available to describe ordinary, as
opposed to major, cities, although it could also be used to describe all
cities, both major and ordinary together.

There is no evidence that either word was applied in any sense to
Chester or Caerleon under the Roman empire. Neither was a /colonia/ nor
a /municipium/ nor the chief city of a province (however insignificant
or short-lived) nor even of a tribe (Stevens 1937:202-3
<hagcl2.htm#stevens>). Their modern names, which are both cognate with
/castra legionis/ ("the camp of the legion"), suggest that they were
thought of primarily as military centres,^14 <hagcl2.htm#note14> and
there is some evidence to suggest that /castra/ and /civitas/ (in its
inclusive sense) were alternative categories in Britain during the final
years of the empire.^15 <hagcl2.htm#note15>

In Gildas's initial description of Britain, he claimed that it contained
twenty-eight cities, which he called /civitates/.^16 <hagcl2.htm#note16>
He will have been following an official list of some kind, which will
have used /civitas/ inclusively. York would certainly have been on any
list of that kind: There is nothing to suggest that either Chester or
Caerleon was.^17 <hagcl2.htm#note17> Elsewhere, however, Gildas seems to

observe some kind of distinction between /civitas/ as any city and
/urbs/ as a great city, although it is difficult to be sure because on
many occasions his usage looks as if it may be constrained by a literary
or experiential context.^18 <hagcl2.htm#note18> Almost three-quarters of
his uses, for instance, quote or paraphrase the Bible. Two instances of
/urbs/, the passage under consideration here and a reference to Aquileia
as the place where Magnus Maximus was executed, seem likely to reflect
established late Roman usage, though there is no reason to suppose that
Gildas would have disagreed with it. (Aquileia, according to Ausonius
(Green 1992:171 <hagcl2.htm#green>),^19 <hagcl2.htm#note19> was the
ninth most important city in the world.) It is reasonable to suppose,
therefore, that Gildas's /urbs legionum/ embodies a recognised
Romano-British distinction between York as the legionary /city/ in the
full sense and Chester and Caerleon, possibly as lesser cities
(/civitates/ rather than /urbes/) but more probably as mere fortresses
(/castra/).

Under the late empire, when ecclesiastical organisation followed the
structure of imperial administration, York's civil status made it an
episcopal see. In 314 its bishop attended the Council of Arles, in whose
records he was named first among the British delegates (Haddan and
Stubbs 1869-78, 1:6-7 <hagcl2.htm#haddan>).^20 <hagcl2.htm#note20> That
suggests York was the most important episcopal see in Britain at the
time, although that status may have been only a temporary consequence of
the favour of the Emperor Constantine, who had been proclaimed emperor
there eight years before. Any bishop's see, however, must have had
places of Christian worship, where the cult of martyrs would naturally
have been fostered, even when, as at York, no particular dedications are
known. It would be entirely natural if, as the civil structure of the
empire disintegrated further and further, an awareness of York's
religious status reinforced or even determined the use of the word
/urbs/ in a writer like Gildas, to whom religion was even more important
than /Romanitas/. There is no direct evidence of bishops or shrines in
Chester or Caerleon before the ninth century.^21 <hagcl2.htm#note21>

Gildas's second word, /legionum/, also points to York, although less
decisively. The word is plural, and York could claim to be a base for
more than one legion in a way that Caerleon and Chester could not. York
alone, after Roman military dispositions had settled down at the end of
the first century, had had two legions based in it (IX and VI,
successively), and it had also, as was natural for the legionary
fortress nearest to the threat from beyond Hadrian's Wall, been most
favoured with flying visits from emperors with accompanying favourite
legions. The use of the plural may therefore, like the later competition
between the Archbishops of York and Canterbury over the number of
processional crosses each was entitled to have carried before him,
embody a claim to superiority over rivals, reminding hearers that
neither Caerleon nor Chester could properly be called more than /castra
legionis/, "the fortress of the legion."

When all these things have been taken into account, we must surely
conclude that in Gildas's time the shrine of Sts Aaron and Julius was in
York, and that the cult of the two martyrs in ninth-century Caerleon has
to be explained in some other way. Even if the ninth-century cult was
based on authentic relics, it is easy enough to imagine plausible
explanations. There must, for instance, have been later points in the
/divortium/ of Britain when a pagan Anglo-Saxon ruler who controlled
York could have cemented a truce or alliance by giving the relics of the

two saints as a good-will gesture to Britons who cared about them much
more than he did.^22 <hagcl2.htm#note22>

* * *

The problems raised by Gildas's /urbs legionum/ may be further
illuminated by considering later phrases similar to Gildas's, and
particularly such phrases in two other books written before the first
known record of the shrine at Caerleon.

The first of these is Bede's /civitas legionum/, which shows that
Britain's first great historian thought of Chester as a city of more
than one legion. However, since modern archaeology says otherwise, Bede
must have been misled, presumably by inadequate evidence. The status of
his phrase is uncertain, but it seems more likely to be a description
than, like Gildas's phrase, a recognised place-name equivalent. We can
be certain that it is not a name: as it happens, in the very sentence
that includes the phrase /civitas legionum/, Bede gives the British and
the Saxon names for Chester, as /Carlegion/ and /Legacestir/
respectively (and observes that the British name is the more authentic).
Nor is there any reason to suppose the phrase was a place-name
equivalent. Bede, writing two centuries after Gildas, on the wrong side
of the country and on the wrong side of a major cultural divide, was not
well-placed to learn of such things. Moreover his /Historia/, whether
from lack of interest or lack of information or both, shows little
interest in legions or Roman cities; and much of what he does say on
those subjects, including his assertion that Britain had twenty-eight
cities (/civitates/), is simply taken over from Gildas, whom he
recognised as the historian of the Britons (Bede 1.1--the 28 cities,
Colgrave and Mynors 1991 <hagcl2.htm#colgrave>; 1.22--Gildas, Colgrave
and Mynors 1991 <hagcl2.htm#colgrave>). It is difficult to know why he
used the phrase /civitas legionum/ of Chester, but he might have guessed
that it was one of Gildas's twenty-eight cities from its size, that it
was a legionary city from its British name, and that Latin descriptions
of legionary cities put the word for legion in the plural by Gildas's
/urbs legionum/. However that may be, there is no reason to take Bede's
phrase as implying that the shrine of Sts Aaron and Julius was in
Chester.

We should notice at this point that four centuries after Bede, William
of Malmesbury used Gildas?s phrase in describing a rebellion against
Edward the Elder. The rebellion, which took place in A.D. 924, was
clearly at Chester, since Edward died a few days later at Farndon-on-Dee
(Stenton 1985:339 <hagcl2.htm#stenton>). In his account of the incident
in his /Gesta regum/ in 1120, William calls the place /urbs legionum
/(William of Malmesbury 133.1 and passim; Mynors 1998
<hagcl2.htm#mynors>); but since William admired Bede and intended his
book as a continuation of Bede?s /Historia/, he may have coined the
phrase either as a conscious variant on Bede?s phrase, which he uses
quite often elsewhere, or by working independently from similar
information (Gransden 1974:166-85, especially 169
<hagcl2.htm#gransden>). Either way, his phrase is no guide to what
Gildas had meant when he used the same words six hundred years before.

* * *

The other book that demands serious consideration is Nennius's /Historia
Brittonum/,^23 <hagcl2.htm#note23> which was put together about a

century after Bede wrote his /Historia/. The /Historia Brittonum/ has
had a bad press recently, particularly from David Dumville, who has
argued that what it says about fifth and sixth-century Britain is
worthless except as a record of what people in later centuries thought
happened then (Dumville 1986:1-26 <hagcl2.htm#dumville3>; 1977:173-92
<hagcl2.htm#dumville5>).^24 <hagcl2.htm#note24> That is a sweeping
claim, but we need not address the whole of it here, since Dumville
acknowledges that the different sections of the/Historia Brittonum/ need
individual assessment, and we are concerned only with a single section,
the much-discussed list of Arthur's battles in Chapter 56. In that list,
the ninth battle is said to have been fought at the /urbs legionis/
("the city of the legion").^25 <hagcl2.htm#note25> There is general
agreement that the list of battles derives from a Welsh battle-listing
poem, which seems to have rhymed on the names of the battlefields (Jones
1964:3-21).^26 <hagcl2.htm#note26> Omitting some Latin inflections that
would have had no equivalent in the Welsh, two consecutive names rhyme
on /-as/ (/Dubglas/ and /Bassas/), and at least three on /-on/
(/Celidon/, /Guinnion/, and /legion/).^27 <hagcl2.htm#note27> Rhyme
tends to give stability in transmission, so Nennius's rhyming names are
likely to be the closest part of his chapter to the underlying Welsh
poem.

There is a consensus that Nennius's /urbs legionis/ translates an
original Welsh /cair legion/. The Welsh phrase as a whole was current as
a place-name (Rivet and Smith 1979 <hagcl2.htm#rivet>), its individual
words correspond to Nennius's Latin ones,^28 <hagcl2.htm#note28> and
/legion/, as we have seen, is also supported by the apparent rhyme. It
may be added that one important group of manuscripts of the /Historia/
glosses /urbs legionis/ with /brittannice Cair Lion dicitur/ ("it is
called 'Cair Lion' in the British language"; Dumville 1985:104
<hagcl2.htm#dumville3>).^29 <hagcl2.htm#note29> It might be objected
that if Nennius's original read /cair/ ("a fortified place"), he ought
to have translated that by /castra/, a cognate Latin word similar in
form and meaning; but Nennius's own book shows that the sense of /cair/
can vary with its context. What in one place he calls /Cair
Guorthigirn/, in another he calls /arx Guorthigirni/ ("the citadel of
Vortigern"), and his /castellum Guinnion/ ("Guinnion fort") might
represent a /cair Guinnion/ in his original.^30 <hagcl2.htm#note30> When
he tries to identify Gildas's twenty-eight cities, he gives an
introductory sentence in Latin followed by twenty-eight place-names in
Welsh; the introduction calls the cities /civitates/ and every name
begins with /cair/, which here at least must be equivalent to
/civitas/.^31 <hagcl2.htm#note31> The battle-listing poem may well have
implied that its /cair/ was a city too. We may ask why Nennius did not
call the place /civitas legionis/, since he used /civitas/ to introduce
his list of cities, in which both the major British sites called /Cair
Legion/ are included. However, unlike Gildas, Nennius seems to have had
no sense of a distinction between /urbs/ and /civitas/ (He uses both
words for the same place more than once.), and he appears to have a
slight preference for the former.^32 <hagcl2.htm#note32>

It seems certain that Nennius would have understood /cair legion/ to
mean Chester. He names Chester and Caerleon among the twenty-eight
cities of Britain, and calls them /Cair Legion/ and /Cair Legeion guar
Uisc/ ("Caerleon-on-Usk") respectively.^33 <hagcl2.htm#note33> That
implies that a /cair legion/ without /guar Uisc/ or any other
qualification meant Chester to him and the audience he was writing for.
This line of reasoning has made modern scholars almost unanimous that

when Nennius wrote /urbs legionis/, he was saying that there was an
Arthurian battle at Chester; scholarly disagreement has focussed
entirely on whether he was right.^34 <hagcl2.htm#note34> Some, pointing
to the mobility of fifth-century warfare, and to Gildas's assertion that
the Saxons raided right across Britain as far as the western ocean
(Gildas § 24; Winterbottom 1978:27 <hagcl2.htm#winter>), have insisted
that such a battle was at least possible. On the other hand, it has
often been said to be very unlikely. One distinguished authority has
said forthrightly that Chester was no place to meet Saxons in the fifth
century, and therefore concluded that Nennius?s phrase was an "almost
certain" interpolation into his source; another suggested this supposed
British victory was a confused memory of the Battle of Chester in 613 or
616, which was a British defeat (Bromwich 1963:93
<hagcl2.htm#bromwich2>; Jackson 1959:8 <hagcl2.htm#jackson1>). Since the
_urbs legionis_ seemed to be the most straightforwardly identifiable
place in Nennius?s whole list of battlefields, such doubts about his
reliability naturally made scholars very suspicious of what he said
about the other battles. It has been suggested that he invented
battlefield names or supplied them from miscellaneous sources, perhaps
to replace names lost in transmission, or to provide Arthur with a total
number of battles thought desireable for a great warrior,^35
<hagcl2.htm#note35> and although scholars who put these ideas forward
have always made it clear that they were only suggestions, the absence
of serious counter-arguments has given them a status very close to
received ideas.

The whole edifice, however, is shaky. First, although we may agree that
Nennius would have understood /cair legion/ as Chester, his source was
composed by someone else, and these arguments take no account of what
those words might have meant to that someone else. Second, there is some
reason to suppose that Nennius himself may not have been referring to
Chester. The main purpose of his list of Arthur's battles was clearly to
tell readers where the battles were fought. He repeatedly uses a formula
emphasising that the battles were past but the names are present: "The
nth battle was fought in x, which is called. . . ." He names five of the
nine battlefields like that, and names the region in which one of them
was situated in a similar way. All the other battlefield names may have
been current too: he never implies the contrary. It fits in with this
that the essential element of each name is in Welsh, since any other
language would have made the names less comprehensible. Apart from /urbs
legionis/, only one name is wholly in Latin, and it is immediately
followed by a Welsh equivalent: /bellum in silva Celidonis, id est Cat
Coit Celidon/ ("the battle in the Caledonian Wood, that is the battle in
the Caledonian Wood"). Perhaps Nennius was not sure whether his
original's /Cat Coit Celidon/ was a description or a name: translating
and quoting provided for both possibilities.

/Cair legion/, however, presented no such problems. For Nennius and his
circle in nearby Gwynedd, Chester must have been a very important place,
and its Welsh name one of the most familiar in Britain. To provide his
readers with the complete set of comprehensible Welsh names that he
obviously wanted, he had only to reproduce the name his source put in
front of him; but he chose instead to translate it. That demands
explanation, and I suggest that the best explanation is that, knowing
the Welsh phrase would make his readers think he was speaking of
Chester, Nennius translated it into Latin to make them realise that he
was not.

We cannot tell whether the Welsh poem or some other source prompted
Nennius to this apparent refusal to endorse Chester . We can, however,
be confident that he did not feel able to suggest any alternative
location for the Arthurian "city of the legion". Such a location would
have completed his list of place-names, and if he had had it, he would
surely have given it to his readers. Instead he took pains, as he did at
other points where we might guess he had doubts about his source,^36
<hagcl2.htm#note36> to reproduce exactly what his source said, which in
this case meant providing a descriptive phrase that reproduced precisely
the ambiguity of his original. His phrase was much more ambiguous than
Gildas's, because (as we have seen) Nennius apparently did not
distinguish between /urbs/ and /civitas/, and his source's uninflected
/legion/ would have applied equally to those places that had
associations with several legions and to those that had associations
with only one. Nennius will certainly have known that his phrase
/described/, even if it did not name, both Caerleon-on-Usk and Chester,
and presumably that it might have applied to other places that he did
not know of.

It is worth noticing that a refusal to provide the specific name his
list asks for tells strongly against any theory that Nennius invented
battlefield names or supplied them from miscellaneous sources, even
though this present essay itself might seem to provide more grist for
that mill. It might be suggested that Nennius's /urbs legionis/ is a
modification of Gildas's /urbs legionum/, which Nennius, perhaps without
knowing the place referred to, had recognised as a kind of place-name
from the distant past, the very kind of thing he wanted as a battlefield
name. However, if he had decided to introduce Gildas's place into his
list, it is difficult to see why, having altered his main source by
doing so, and having pointlessly altered Gildas's /urbs legionum/ to
/urbs legionis/,^37 <hagcl2.htm#note37> he should have been unwilling to
translate the altered phrase into Welsh to match his other place-names.
In reality, all theories of this sort look unlikely. If Nennius was so
alert to and so scrupulous about the possibility that his source might
mislead his readers that he refused even to reproduce the wording that
his source put in front of him, we have no grounds for thinking that he
carelessly or deliberately corrupted it in more radical ways.

What Nennius did with his source-material, however, is only a beginning.
The most interesting question raised by this passage is what was meant
by whoever composed the equivalent passage in Nennius?s source. That is
a much more difficult matter. Nennius?s scrupulosity or otherwise with
his sources does not tell us anything about how those sources treated
their own source-material; and if inadequate sources could have misled a
great historian like Bede, they certainly could have misled Nennius and
a source whose origins lay in oral tradition. Nevertheless, /could/ is
not /did/, and evidence has been accumulating to suggest that Nennius
and his sources may deserve to be taken more seriously than they have
been in recent years (Koch 1996:246-251 <hagcl2.htm#koch>). It is worth
considering, therefore, without presuppositions, what place the composer
of an early battle-listing poem might have understood as "the city of
the legion" and apparently referred to in such a way as to make a
ninth-century enquirer think of that place as a legionary city that was
not Chester.

The most important factor in giving an answer to this question is that
the names in Nennius?s list are not, as far as we can tell, his own
deductions from his source, but are directly taken from it, where they

provided its rhyme-words. If Nennius?s source gave him real names, or at
least, in the case of "The Battle of the Caledonian Wood", a recognised,
established, and unambiguous phrase equivalent to a name, its /cair
legion/ should be a name or a name-equivalent of the same kind. But
although Britain had a great many legionary encampments, their sheer
number ensured that very few of them took their names from that fact
alone. Their number would have made such names ambiguous, and so nearly
all British place-names with legionary associations incorporate
qualifying elements, usually as prefixes, as in modern /Manchester/ and
/Winchester/. The process can be seen at work in Nennius?s name
/castellum Guinnion/ for another of Arthur?s battles. Some of the few
apparent exceptions to this process actually have quite different
meanings, which would have been obvious in the early stages of their
development; such as /Caerleon/ in Morval and /Carlyon/, both in
Cornwall, whose second element derives not from Latin /legion-/
("legion"), but from the plural of Cornish */legh/ ("flat stone",
"slab"; Padel 1985:52 <hagcl2.htm#padel>).

The most important exceptions are of course /Chester/ and /Caerleon
/(-on-Usk), where in both cases the importance of long-established
legionary fortresses ensured first that the place could be called simply
"the" fortress of the legion, and second that that name could entirely
displace an earlier native British name. The place-name record shows, as
one would expect, that this took time: a common description had to turn
into a recognised name-equivalent, then become a secondary name, and
finally displace the original name. At York, this process was never
completed, so its modern name is a descendant of Roman /Eburacum/ by way
of British /(Caer) Ebrauc/, Anglo-Saxon /Eoforwic/, and Norse /Jórvík
/(Rivet and Smith 1979 <hagcl2.htm#rivet>; Smith 1937:275-80
<hagcl2.htm#smith>); but Gildas's evidence shows that it was begun.
Gildas shows that in the fifth century a name-equivalent existed that
identified York as a legionary city, and that a competent author could
use that name-equivalent in a book addressed to the whole of Britain,
expecting his phrase to be recognised by his entire audience.

The overall pattern of warfare in Britain in the late fifth and early
sixth centuries suggests that even if, as Gildas said, the invading
Anglo-Saxons raided right across Britain, battles were more likely to be
fought in the east than in the west, and specifically were much more
likely near York than near Chester or Caerleon-on-Usk. In terms of
antecedent historical probability, York (as we have seen) is a highly
probable place for a late fifth-century battle between Anglo-Saxon and
British forces, whereas Chester and Caerleon are very unlikely. Since
the place-name evidence shows that a phrase like Nennius's could apply
to all three of those places, it follows that, whatever Nennius himself
thought the phrase meant, what he said about Arthur's ninth battle is
much more likely to have been true of York than of Chester or
Caerleon-on-Usk. We must therefore assume that if Nennius's source was
reliable, his /urbs legionis/ is Gildas's /urbs legionum/ too; and that
if, conversely, Nennius?s /urbs legionis/ is Gildas?s /urbs legionum/,
the principal factor that has led scholars to believe Nennius?s list of
battles to be unreliable no longer exists.

Notes <hagcl2.htm>

References <hagcl2.htm#bib> Next <hati.htm>

Return to Table of Contents <hatoc.htm>

Return to Homepage

<http://member.aol.com/heroicage1/homepage.html>