Gildas' picture of Roman conquest and rule is quite legendary. We do not even have to suggest that a real memory of the historical Roman conquest underlay it. Celtic nations had a habit of picturing their first origins as conquests from overseas. The notion that Britain had been peopled by successive inflows from across the ocean is found in Nennius, in Geoffrey of Monmouth, and in bardic/triadic tradition^[1], that is, in every source we have. Even Arthurian legend has an element of transmarine origin in the arrival of the Grail Family from Palestine across the ocean, which includes demonstrably Celtic material about Bran^[2], a typical legendary ocean-crosser. In Ireland, not only the successive invasions of the Book of Conquest, but also separate legends about the royal houses of Leinster (from Labraid Loingsech) and Munster (from Eogan Mor) told of invasions from over the ocean^[3]. There is no reason to see any of these tribes as other than native - in the sense that they had been settled in their land for hundreds of years, whatever their ultimate origin - and by the same token, whether or not any memory of the real Roman conquest of Britain had been preserved, the presence of a British aristocracy claiming Roman descent would ipso facto suggest to any traditional Celtic version of history that, like Labraid Loingsech and Eogan Mor, it had come from abroad.

To understand Gildas' picture of history and its relevance to historical fact, it is important to be clear on this: the fact that the Romans had in fact come to Britain from abroad is quite incidental to the tale he tells. Had they been aboriginal, the chances are that he would have told the same story. The fact that Celtic language and culture may be presumed to have come to the British Isles from abroad has misled many scholars into giving undue attention to such legends as the Book of Conquests as historical sources; but the Book of Conquests, which took shape centuries and perhaps millennia after Celtic aristocracies had acclimatized themselves to Ireland and could be regarded as native, has nothing to say about Irish prehistory, despite centuries of earnest endeavour to prove the contrary. And as no detail of Gildas' account of the conquest has anything whatever to do with historical truth, the same must be held to go for him. He would have treated the Romans as conquerors from abroad whether or not they were.

Part of this unhistorical picture is the startling fact that Gildas believed the Picts of north-east Scotland to have only entered Britain's far north after Maximus took the Romans off. Indeed, it is only after the second of two post-Maximus Roman rescue expeditions has set a permanent border (in the shape of the Wall) to Pictish expansion, that the Picts actually seize the land north of it. The Latin is clear enough: omnem aquilonalem extremamque terrae partem pro indigenis muro tenus capessunt, they lust-after/approach/seize (capessunt) the northernmost and furthest parts of the land as far as the wall. Only the two words pro indigenis, (19.1) are slightly unclear (“as if they had been native” or “in the face of the natives”); they probably mean that the Picts claim to be indigenous though they are not. This certainly does not affect our understanding of the sentence: the Picts had only seized the north of the land after the last Roman expedition. The verb capesso stands for taking, or designing to take, something that is intensely lusted for, desired, wanted; it describes well a greedy, impoverished, eager barbarian horde invading a richer and more civilized territory. And what the Picts of Gildas lust after is to devastate and settle the island of Britain, a purpose to which they clung in spite of successive reverses and which he regards as their more solito, their usual habit (22.1). Despite two repulses at Roman hands, they manage to capture omnem aquilonalem extremamque terrae partem... muro tenus, all the northern and furthest part of the land, right up to the Wall. It is the aquilonalem partem that is lusted-after/approached/seized, the object of the verb: and that it is only lusted-after/approached/seized after the Romans built the stone Wall, can only mean that the whole island, right up to John O'Groats, was originally inhabited by the British, and governed and defended by the Romans. The Picts had arrived after the Romans left, and the two Roman expeditions had only put two successive borders to their southwards expansion^[4].

Nothing shows more clearly Gildas' ignorance of the real history of Roman Britain. The Picts were nothing else than a confederation of north British tribes never conquered by Rome, which the Roman historian Ammianus Marcellinus divided into Dicalidones - the "two Caledonians" - and Verturiones (probably connected with the ancient place-name Fortrenn.). They lived in East Scotland from prehistory, and the Caledonians (who left their name to Dunkeld in the Grampians) were of course known to Tacitus. Yet Gildas sees no difference between them and their frequent allies the Scots (from Ireland), calling both transmarinis gentibus. Transmarinus is his word for people or things from outside the great island, and he says that, while the Scots came from circio (a word which implies a surrounding position), the Picts came from aquilo, the north. What this means is not clear: he might refer to any of the islands north of Britain, the Hebrides, the Orkneys, the Shetlands, even Iceland, or share the later British belief that they were late arrivals from Scytia. The one thing he clearly states is that they are fifth-century invaders - and anyone who had read Roman accounts of Britain would know that that was false. The minimum conclusion to be drawn is that Gildas has no knowledge of actual Roman British history from, say, Carausius to Magnus Clemens Maximus, when the Picts were the biggest foreign policy issue.

History begins to emerge from legend in his account a few years, perhaps a few decades, before the Rescript of Honorius. Magnus Maximus already has the role of last Roman Emperor in Britain and ender of the Roman golden age, typical of all later Welsh legend; the parts of his account that Gildas takes from Sulpicius Severus must be understood as Roman grafts on a British (Welsh) trunk, rather than the reverse, and the fact that the Picts reach Britain immediately after his doomed adventure only underlines that his role in Gildas' story is that of "weakener of Britain", legendary and not historical, part of a complex of pseudo-historical ideas of which this "Pictish invasion" is another. It is not even clear that real memories of Magnus Clemens Maximus underlie, as is generally assumed, the legends gathered around his name; the name and the details taken from Sulpicius Severus may have been, as we will see, misattributed to a story told of a slightly earlier usurper, Magnentius.

There is, however, no doubt that the memory of military usurpers assaulting the rest of Romania from Britain, only to be defeated, underlies the legend: without the idea of a great army leaving Britain and being lost for ever, as the hosts of Magnentius and Magnus Clemens Maximus were when the armies of legitimate emperors destroyed them, there could be no legend of Maximus at all (no such story of great armies lost abroad is known in Ireland so far as I am aware). This is still a part of the British legendary picture, but it is a part that connects with genuine history. It preludes, through a few more partly legendary stages, to the undeniably historical Rescript of Honorius, and many of its obviously legendary elements depend on misunderstandings of actual fact.

We can now see that Gildas starts from the premise that his nation, the British, used in the past to possess the whole island of Britain, a united country first conquered and then defended by the Romans. We only hear of Picts and Scots - let alone Saxons - after Maximus takes the troops to the continent, leaving Britain incapable of defending itself; but then we hear of them immediately. As soon as the Romans are gone, the British suffer savage Pictish and Scottish invasions.

Indeed, the role of the Romans is still the same: they defend the island, even after they are gone. The British, overwhelmed by Pictish hordes, send to Rome for help twice, and are rescued. The first time, the Romans come, swiftly deal with Picts and Scots (Gildas says that a single legion had been enough to drive them off, which has an amusing but surely unintended echo of Agricola's assessment that one legion would be enough to conquer Ireland) and order the British to build a wall to hold back further Pictish inroads (15.2). But the useless turf wall built by incompetent British hands does nothing to hold the enemy back, and they come like an avalanche.

The second and last Roman expedition is an epic episode; so determined is Gildas to play it up that I suspect he may have based himself on a lost battle-song. His tone is of awestruck gratitude, his language aggressively poetic and suggestive, his rhetoric high-flown even by his standards (ch.17): the Romans come uolatus ceu aquilarum, equitum in terra, nautarum in mari cursus accelerantes - "in flight as if of eagles, speeding the course of horsemen across land, of sailors across the sea" - and there is a lot more of this sort of thing.  He even manages to suggest that the sky was not closed to the Roman army. Mention of horsemen and sailors, in the genitive plural, is preceded by one of eagles in the same case. Though aquilarum is the predicate of a different subject, uolatus rather than cursus, those two words also are in the same case and from the same word-group, the comparatively rare fourth declension, and are both constructed on verbal roots. The effect is to suggest that the Roman army had an Air Force^[5].

After the success of this expedition, the Romans set up a permanent border. Just advising the British to build a wall after expedition number one (15.2) had not worked; therefore, unwilling to be summoned a third time, they themselves organize the construction of a second wall (18.2), built of stone rather than turf, obviously the Wall of Hadrian. Both the Walls of Antonine and of Hadrian have therefore been built at the direction of the Romans - and, in the case of the stone one, under their direct orders - to keep the Picts out.

Now Gildas obviously knows nothing of how the two Walls had come to be. He believes they were built in the fifth rather than the second century, and that the turf Wall of Antonine was earlier than the stone Wall of Hadrian. Every modern schoolboy knows he is wrong, of course; but why did he think so?

Well: for a start, he believed the Walls to have been built to hold the Picts back - and the Picts, in his system, had only reached Britain after the end of Roman power. Only when Britain was deprived of "Roman" fighting men - those eagle-like, invincible heroes of ancient legends, of whom one swoop could scatter enemies like chaff - could this tribe from some more or less dimly perceived northern land could it. So his problem was: given that - as he thought - the Picts had only entered the island after the Romans had left, how come that the Romans had left two walls across the neck of Britain, marking what he regarded as stages in the barbarians' forwards advance?

However, this question only makes sense if it depends on a clear statement that "the Romans built the two Walls to protect Britain from the Picts". Now, the stone wall of Hadrian showed its Roman origin in its building technique, lost to Britain then and for centuries to come; but there was little to tell the turf Wall of Antonine from the many earthworks that criss-crossed the island. Yet Gildas knew that its building had had something to do with the Roman army, though the fact that it was built of turf rather than stone misled him into thinking that the actual building had been carried out by Britons. Therefore, if he thought that "the Romans built the two Walls to protect Britain from the Picts" he must have had other reasons.

This kind of misunderstanding is based on verbal precision coupled with absolute ignorance of background facts. His source has to be a single statement - probably a single sentence - taken out of context, almost certainly from a written source^[6]. Gildas knows only this, that The Romans built the Walls to protect Britain from the Picts, but he had no idea of chronology or context. Yet he regarded this information as reliable enough to incorporate in his picture of history. It is not only possible but easy to see that Gildas was working out a clear process of historical deduction from evidence. First: Gildas knew that the British built turf walls, whereas Romans built in stone. Therefore the turf wall was built by Britons, not by Romans. Second: Gildas also knew that the turf Wall, as well as the stone one, were in some way a result of Roman activity. Third: the stone Wall lies farther back than the turf one, suggesting loss of territory. Fourth: Gildas believed that the Picts had only entered the island of Britain after the Romans had left it. Therefore the turf Wall was built first; once the barbarians had penetrated deeper, the Romans, unable to undo the results of British fecklessness, directed the building of the second, using their own methods of stone building.

If we accept that Gildas knew that the Roman army was somehow responsible for both Walls, this is an entirely reasonable reconstruction. That the Wall of Antonine was built of turf must have puzzled Gildas' contemporaries no end: if the Romans always built of stone, why would they have made this important structure of turf? The difference between stone building and turf building was literally visible in his time and for many centuries to come, in the old, crumbling, but ever impressive stone remnants of the past, city walls, villas, civic buildings, so unlike the wooden or half-timbered buildings, however stately, raised by Britons in places such as Wroxeter^[7]. And the fact that the stone fortification, stronger and more impressive than the turf one, also lay further back, clearly suggests a loss of territory. It is perhaps wrong to treat Gildas as "not a historian". While The Ruin is certainly a polemical tract rather than a historical study, the amount of interpretation and argument from historical premises this shows certifies that a first-class historical mind is at work; the solutions are wrong, but they are brilliant. The only question is whether they are Gildas' own.

I think they are. The emphasis on the unmilitary British nature is wholly Gildasian; and Gildas' statement that both Walls were built after the end of Roman rule is a freak. No other historian except Bede (and Geoffrey of Monmouth) uses this datum; even Nennius, who had read both Bede and Gildas^[8], disregards it. And Bede, who copied Gildas word for word, can be shown to have worked hard to integrate Gildas' narrative with the chronological schemes of Roman historians; with wholly unfortunate results. The best-known aspect of this effort is his long and unhappy wrestle with the date of the diplomatic Letter to Agitius, which fostered a disastrous misdating of the adventus Saxonum, the arrival of the Saxons to Britain, to 449, still to be found in some textbooks. As John Morris^[9] saw, this depends on Bede's word-for-word acceptance of Gildas. Gildas had understood the letter to refer to a Pictish invasion, whereas in fact it spoke of the great Saxon revolt dated by another source^[10] to 442. The letter is dated to after the third consulate of the Roman general Aetius (the Agitius of the title), which was in 446; therefore Bede came to understand that the Britons were still fighting the Picts at that late date - and the earliest date he could confidently offer for his own people's arrival in Britain was 449, which, as we will see, was wrong by about eighteen years^[11].

Likewise, Bede found himself faced with Gildas' clear statement that the Picts and Scots were transmarinis gentibus. Now Gildas and Bede each had a part of the truth: Gildas was quite right to qualify the Scots as transmarini; Bede was equally right in stating that the Picts had always lived on the island of Britain, beyond the Firths of Forth and Clyde. But faced with Gildas' clear statement that the Picts were late transmarine invaders, Bede did not want to contradict him. "I term these races extraneous", says he - though it is not he but Gildas who so terms them - "not because they came from outside Britain", which is exactly what Gildas meant, "but because their lands were sundered from that of the Britons; for two sea estuaries lay between them" - not a very good reason (since if the sea had been their favoured way of invasion, why all those Walls?), but the best that he could think of. (We notice, by the way, that Bede regarded the Scots as native to modern west Scotland - the one thing in which he was wrong, and Gildas right.)

His account of the building of the two Walls is of a piece with this. Bede clearly had a source other than Gildas, which attributed Hadrian's Wall to the African emperor Septimius Severus (193-211), since he corrects the "some" who "imagine" that Severus built a wall^[12]: he had built an earthwork - and Bede describes somewhat pedantically the difference between the two. Because Severus was in fact repsonsible for extensive renovation of a decayed Wall, later Roman historiography sometimes attributed the Wall to him rather than tohis predecessor Hadrian. The reason why Bede corrects his Roman source, however, is, without a doubt, that his other source is Gildas, who claimed that the first Wall was the turf-built one, and that it was the result of a Roman expedition to rescue Britain from barbarians - such as Severus had in fact launched, and in whose course he had died. Since Gildas says that the stone Wall was not built till some time after the end of direct Roman control, whatever defence Severus built cannot have been a wall. Then, following Gildas more closely, quoting him, in fact, word for word, Bede astonishingly repeats the building of the turf Wall, this time by the incompetent British, finally following it up with the stone Wall built under Roman supervision. The result is that Bede is now stuck with three Walls - Severus' phantom earthwork, the British turf Wall, and the Roman stone one: one too many.

How did that happen? I suggest that, copying out Gildas, Bede, who had already decided to date the turf Wall to the time of Severus - because his other source certainly affirmed that Severus had built something - found himself faced with the alarming fact that Gildas had made the building of both Walls depend on two successive Roman rescue expeditions after to the end of Roman rule in Britain - which Bede well knew to date to 410 and centuries after Severus. Bede could not, without the best cause, tamper with Gildas, a primary source, a holy and wise man, a great writer (whose literary greatness the brilliant Northumbrian was perhaps better able to value than most) and most importantly a true prophet; since Bede, like every writer before and after him, had misunderstood the great man to predict the island's conquest by the English. A Saint, inspired to see the future, carrying such a burden of truth - how could Bede, with his religious respect for the great men of the past, play fast and loose with his witness?

So Gildas' account of the Walls was not only mistaken, but a fertile generator of further errors in the greatest of the scholars who followed him. Nevertheless, just as Gildas did know that the turf Wall of Antonine - physically indistinguishable from a British-built "dyke" - was in fact a Roman work, so too there is nothing in the Gildasian legend to deny that he was in receipt of some dim but real memory of Roman intervention in the last days of the western empire. His notion of two Roman expeditions from the Continent to shore up crumbling British defences against the Pictish danger echo known facts of the last decades of Roman power. I admit that the province's history is obscure; but the fact is that we know of two, and no more than two, major such campaigns, that of Theodosius the elder in 367 and that of (or ordered by) Stilicho in about 399. These were measures over and above the ordinary defence of the realm, taken when regular border forces had failed - in one case, due to treason. Information about them may have reached Gildas through the same channels as the Roman origin of the two Walls.

Gildas dates the two Roman expeditions after the usurpation of Maximus; in historical fact, that of Theodosius was earlier. Of course, Gildas' time-frame is as reliable as quicksand, but there is a fascinating coincidence: Theodosius' mission followed the fall of another usurper with a similar name and British backing, Magnentius. Magnentius is in many ways more similar to the Maximus/Maximianus/Maxen of legend than the historical Magnus Clemens Maximus, and I think it likely that Gildas, or some of his predecessors, may have confused the two pretenders. Both their careers started in Britain; both overthrew and killed Christian emperors (Constans and Gratianus); both were defeated and killed by armies from the Eastern Roman Empire; but only one, Magnentius, was born in Britain. Like Gildas' Maximus, but unlike Magnus Clemens Maximus, he was not of Roman blood^[13]. And his usurpation was more closely identified with Britain than that of Maximus, who took more interest in Gaul and Italy; when it collapsed, the imperial secretary Paulus Catena unleashed a ferocious repression of British Roman noblemen, whose contempt for legality and justice horrified Ammianus Marcellinus - clearly, Catena believed that all the British aristocracy had backed Magnentius.

These events seem to have seriously affected public order and loyalty in Britain. We cannot be sure of the events, because our only source – Ammianus – is tainted, but there are any amount of reasons to believe that all Britain, both the Roman province and the borderlands, were disaffected to the point of revolt by 367 or so. In Ammianus’ suspiciously vague account, Valentinianus – an emperor of which the historian has little good to say elsewhere – receives a growing series of Job’s messengers from Britain, indicating that Britain was being starved^[14] by a barbarian conspiracy (qui Britannias indicabat a barbarica conspiratione ad ultimam uexatas inopiam), that Nectaridus the admiral had been killed, and that General (dux) Fullofaudes had been deceived (circumuentum) by enemy tricks. Severus comes domesticorum, a very high official, was dispatched, but swiftly recalled. Iovinus, who followed him, reported that he needed a powerful army (exercitus ualidi), and, as no such thing was available, he too was recalled. The fact that Iovinus reported that an army was needed suggests that nobody realized that on the Continent, and it follows that Iovinus, and Severus before him, had been sent without troops, to take charge of provinces and troops already present. In other words, whatever Job’s messages may have been received by Valentinianus (who was, at the time, near Trier, and surely well in touch), he had not been informed, as some historians have wrongly guessed, of any major military reverse. It was only upon inspecting the situation on the ground that Jovinus realized that an army would be needed.

What actually happened in Britain in 367 is a piece of lost history. The elderly general Theodosius, brought out of retirement for the mission, took a small task force made up of largely barbarian detachments, Batavi, Heruli and so on – decidedly not the ualidus exercitus demanded by his predecessor – to Britain, entered London, and did a certain amount of pursuing bandits (there is a clear hint in the text that quite a bit of the bandits’ loot, once recovered, stuck to his soldiers’ hands^[15]). According to Ammianus, he found out from prisoners and deserters that there were so many tribal bands (uariarum gentium plebem) at large that only stealth and trickery could hope to overcome them; but as his next step was to persuade Roman troops, with the promise of an amnesty, to join his ranks, it looks as though many of these “tribal raiders” were in fact Roman soldiers. Ammianus has the cheek of telling us that some of them had gone on libero commeatu, official leave – some leave, if it takes the promise of amnesty to end it!

But his greatest deed, according to Ammianus, was to “disappear” one Valentinus and a few of his leading supporters: a political assassination with the stated purpose of avoiding the publicity of a treason trial. Nor, indeed, did he want his own name, or the army’s, associated with it: he apparently nominated a new head of the British civil service, one Dulcitius, expressly so that Dulcitius, rather than him, should take care of the vengeance (uindicta) against Valentinus. According to Ammianus, this gentleman and all his friends had to go because – in a beautiful Latin circumlocution – ad res perniciosas consurgebat et nouas, he was rising up to strange and vicious matters. Clearly, with the Admiral dead, and apparently no head of the civil service, Valentinus (originally an exile from Pannonia, but one of great reputation and perhaps wealth) had become a big man in Britain. His and his friends’ sudden and mysterious disappearance clearly served as a strong incentive to yet unreclaimed troops and British Roman citizens, especially since Theodosius – unlike Paulus Catena – made it clear that he did not want to indulge in a wholesale purge - de coniuratibus quaestionibus agitari prohibuit, ne formidine sparsa per multos reuiuscerent prouinciarum turbines, he forbade public investigation of the conspirators, to prevent fear spreading through many places and reviving the storms of the provinces.

This gives the game away: the threat that Theodosius was eager not to revive, was not a barbarian invasion, but the prouinciarum [i.e. Britanniarum] turbines - Magnentius’ entirely Roman rebellion. It is perhaps no coincidence that his little task force was made entirely of Continental barbarians; did he (or his notoriously paranoid master Valentinianus) not want to trust Roman citizens? By the same token, there was no major collapse of law and order. Archaeologists have noticed that there is little of what we would expect from a major security breakdown: no burnt or plundered villas, no evidence for war or slaughter, and no great wave of buried treasures. Law and order may have weakened, bandits may have been having a field day – though many of those bandits may have been Roman soldiers - but there had been no vicious invasion and no collapse. What Theodosius had to do was not to destroy barbarian invaders threatening the island, but to destroy the reasons for the disaffection of Roman soldiers and provincials, negatively – by “taking out” the nobleman who had been making hay out of their disaffection – and positively – by restoring Roman law and order, but with a clear message that this was not going to lead again to Catena-style bloodbaths.

It is not clear that the Picts were any real danger. In the South, Theodosius only hunted down bandits; in the North he "recovered a province that had fallen into enemy control" and organized it ex novo, re-naming it Valentia: to be treated as new-conquered territory. This, however, must have been the far north, between the Walls and York, which had never been very Romanized, and numbers almost no villae. Barbarian settlement in this debatable land amounts at best to nibbling at Roman Britain’s edges, and, as we know that a large Roman force still existed for Valentinus and Theodosius to contend over, it may not have threatened the properly Romanized territories; it is not impossible that it was consensual, with Valentinus and his party surrendering unimportant border marches to highlander friends while they prepared their revolt. At any rate, the closer we look at events, the stronger is the impression that they reflect a widespread British dissatisfaction, both Roman and Highland, with Roman government, closely connected with Magnentius’ revolt and Catena’s repression, and worsened by years of neglect.^[16]

Another problem Theodosius had to solve, according to Ammianus, was the treachery of advance scout units called (according to what editor you follow) arcani (secret units) or areani (area troops). This was no sudden blow; Pictish raids had been getting worse for years, and the treachery of the areani/arcani was, according to Ammianus, long-prepared. It may be that the defeat of local lad Magnentius and the ferocity of Paulus Catena had sapped the loyalty of troops that, being certainly local (scouts and intelligence troops must necessarily be familiar with terrain and people) must have had close connections with local lords and functionaries. This assumes a good deal about the little-known social structure of provincial Britain, but I think it is reasonable: some at least of Catena's many victims must have had highland connections, and their deaths may have left the arcani/areani either thirsty for revenge against the emperor in distant Constantinople, or at least masterless, and therefore open to corruption from over-Wall barbarians.

It is therefore possible that the Gildasian legend of the usurpation of Magnus Maximus and the two Roman expeditions that followed may have something to do with memories of the usurpation of Magnentius, which seems to have been popular in Britannia, and the expeditions of Theodosius and Stilicho. It may of course be that Gildas (or someone before him) created a set of two Roman expeditions without any previous memory, to account for the two Walls in the north. But it is more likely that he wanted to explain the two Walls in terms of two already known Roman expeditions; for it may be argued that Gildas' first Wall, the turf one, does not demand, from his point of view, any Roman intervention at all. It looked "British", and he claims that it was actually built by British labour; so why involve the Romans in its building in any way? - unless he already knew, on other grounds, that it was Roman in origin. And finally, it does seem quite coincidental that he should be aware of two separate late-Roman expeditions, the first of them following the collapse of a British-born and British-based usurper which had left Britain open to Pictish invasion, and both of them aimed mainly at the Picts, when every single other datum he gives - bar those borrowed from Church historians such as Rufinus and Sulpicius Severus - is wildly wrong.

If Gildas' legendary version of Magnus Maximus is in any way connected with Magnentius, this tells us something about the transmission of tradition in British/Welsh/Breton culture. Every later item of writing on the subject consistently identified the historical Magnus Clemens Maximus with the usurper of legend, and Magnentius was quite forgotten. If the mistaken identification was Gildas', then it was very influential; indeed, the point hardly changes even if an earlier writer had made the same mistake and Gildas had only copied him. The point is that a definite identification with a Roman pretender known to written history was handed down the generations, from the age of Gildas to that of Geoffrey of Monmouth and the Triads; and, as we will see, the story did not stop developing with Gildas, but threw out a major new feature unknown to him - a duplication. In other words, the tradition we are talking about is fundamentally a written one, which kept some touch with strands of Latin learning preserved from the classical age; since otherwise it would be unthinkable that the same legendary figure should be identified with the same historical one, and, for that matter, should not be identified with a more suitable one (Magnentius). This is a written error, not an oral one; a mistaken entry in the records.

What is however certain is that Gildas knew of a sequence of three successive Pictish invasions, of which the first two are stopped by Roman intervention. In spite of the coincidence with the two Roman expeditions, the sequence is entirely legendary in nature, being built on the legendary idea that the Picts had only started invading the island of Britain after Maximus - the legendary Maximus - had left her defenceless. This is probably the first recorded instance of that favourite Welsh mnemonic form, the triad, since it can easily be demonstrated that this part of Gildas' legend is built on triads. A second is easy to find: "three peoples who invaded Britain: Picts, Scots, Saxons; and none of them went away". A closely similar Triad is actually found in the medieval collections of Triads of the isle of Britain (Trioedd Ynis Prydein), no.36 in the celebrated modern edition by the Welsh scholar Rachel Bromwich^[17]. The Romans, as far as Gildas is concerned, are excluded, since if one thing is clear, it is that they did leave (with Maximus), whereas Picts, Irish and Saxons all held parts of the island in his time. Also, he regarded the Romans as legitimate rulers accepted by the islanders, whereas none of the other groups had any right to be in Britain.

I am proposing to identify two sources for Gildas’ legendary account of the end of Roman power in Britain: the legend proper, built on triads, which I shall call A, and the historical allusions from which Gildas drew the idea of two major expeditions and two Walls, which I shall call B. B does not seem to amount to a coherent text, but must be understood as a number of ideas and data coming perhaps from written text, or perhaps from observation; the latter might include Gildas' knowledge of the forts of the Saxon Shore, and of the difference between the two Walls, both of which are visible to the naked eye and he may have observed himself^[18].

As for the origin and date of A: those armies of north Britain which, in A’s picture, were being overwhelmed by the Picts, were, to A, British, not Roman. Following him, Gildas regarded the forces overwhelmed or corrupted in these episodes as British; apparently, there was not enough force in B to contradict this. The Roman forces of the two expeditions - which, ex hypothesi, we must identify with those of Theodosius and Stilicho - were splendid helpful strangers, allies rather than fellow-citizens. This shows a mentality which has nothing in common, for instance, with that of St.Patrick. To the Saint, British and Roman nationality coincide, to the extent that the worst insult he could hurl at Coroticus was "a ciuis not of the holy Romans, but of the demons". This does not offer us an absolute date - we don't have that until we have dated St.Patrick himself - but certainly a relative one: the legend belongs to a later stage of British cultural history, which, consciously or unconsciously, projected onto the time of the Rescript the political situation of its day, in which the ancient defences of the Wall and of eastern and western shores must indeed have been welded into the general political structure of a Britain separate from Rome.

The existence of a triadic structure at the very core of the narration - the number of peoples who invaded Britain, the number of Pictish invasions - surely indicates that, whatever more or less misunderstood historical memories may be present, the narrative of A is already Welsh in character, built according to the mnemonic constructive principles of Welsh legend. This agrees with what we have found about Gildas' legends of Roman conquest and of Magnus Maximus, in which his narrative is basically not Roman, and, especially in the legend of Magnus, virtually identical with all later Welsh tradition. For all his gorgeous Latin and his Catholic Christianity, Gildas' historical framework, as well as the ideology that underpins it, is clearly Welsh: the first written document of Welsh culture in history. Gildas is, in everything outside religion and language, no longer a Roman, but already a Welshman. And even his love for words and learning - is that not a famous Celtic trait, found with equal virtuosity in the poems of Taliesin and Neirin?

Notes

^[1]BROMWICH, Trioedd, 228 (White book of Rhydderch, folio 600).

^[2]HELAINE NEWSTEAD, Bran the blessed in Arthurian tradition, New York 1939.

^[3]BYRNE op.cit. pp.11, 131-6, 199-201

^[4]This is not a new discovery. NEIL WRIGHT - Gildas’ geographical perspective: some problems, in DUMVILLE & LAPIDGE op.cit.- reviews the views of previous scholars, including a few unconvincing attempts to rewrite Gildas by surprisingly eminent figures - and covers several points not dealt with here.

^[5]In fact, one wonders whether a battle poem about this Roman expedition did not exist and attribute them the power of flight. It would of course be a barbaric rather than a classical Latin production; more importantly, it would see the Romans as a great people from long ago, rather than, as Gildas did, a contemporary empire; hence the attribution of such super-powers as flight and invincibility. If that was the case, Gildas may have felt that the statement had to be toned down and turned into a mere poetical simile, since he knew that the Romans were the Byzantine Empire, and that in spite of its great power the one thing its soldiers could not do was fly. On the other hand, the image of the flight of eagles may simply have been inspired by the Roman ensign.

^[6]It is not impossible that the statement may have come from verbal rather than written communication, but if - to translate from jargon to English - someone had told St.Gildas that “the Romans had built both the Walls to keep the Picts out of Britain”, it is hard to see how Gildas himself would not go back to this authority and ask him about the background to the story, rather than go through the ingenious but rather desperate process of reasoning tinged with guesswork that I describe. It seems to me that the Saint’s elaborate efforts at reconstruction imply that he simply had nobody reliable to go to.

^[7]R.WHITE & PHILIP BARKER, Wroxeter: life and death of a Roman city, Stroud 1998, esp. ch.7.

^[8]As I intend to show in the rest of this work.

^[9]MORRIS, Age of Arthur 39-41 - one of those cases in which Morris is so thunderingly and emphatically right one feels like applauding. Patrick Sims-Williams, no romancer, reaches the same conclusion: “Bede’s chronology of the adventus and mons Badonicus is simply a valiant attempt to interpret Gildas, and has no independent value whatsoever. Had Bede known… the Gallic Chronicle of 452, and he trusted Gildas’ Aetius synchronism less, the course of English historiography from the Anglo-Saxon Chronicle to the present would have looked quite different.” SIMS-WILLIAMS, the settlement of England in Bede and the Chronicle, in Anglo-Saxon England 12 (1983), 21.

^[10]The Gaulish Chronicle of 452.

^[11]We will also find out that this misdating helped him in another difficult area, dating Hengist to after 449, and connecting him with the known dates of Aethelberht of Kent, supposedly his grandson or great-grandson. If Hengist had come to Britain some time in the 420s or 430s, the genealogical rungs would have been impossibly long; as Bede made them, they were just about tolerable - though they still made the kings of Kent remarkably long-lived.

^[12]Here, Nennius is evidently in receipt, not of Bede, but of his source, since he calmly attributes the Wall to Severus - although it is the legendary Severus of the legend of the Seven Emperors.

^[13]Though not by descent: he was the son of a German in the service of Constantine the Great. Gildas cannot have known this, but it is yet another instance in which the racist theory underlying his argument was at odds with historical reality: Gildas held that "Maximus" was a tyrant because he was descended from tyrannical British blood. This is true neither of Magnus Clemens Maximus, who was Spanish, nor of the German-descended Magnentius. It is however distantly possible that Zosimus, to whom we are indebted for Maximus' nationality and who is a notorious bungler, may have called him (in Greek) Hiber, Iberian, Spanish, as a misreading for (Latin) Hibernus, Hibernian, Irish; although, if that were the case, it would be curious that all the anti-Maximus literature evidently produced during and after his usurpation made no known use of this racially extremely despised origin.

^[14]The accounts of this episode to be found in most textbooks are so far from what Ammianus Marcellinus actually wrote as to make me doubt whether my Loebs edition contains a monstrous series of misprints and mistranslations - since each and every scholar seems to have read something different from what my eyes did. To begin with, what he is saying here is clearly, not that Britain was militarily conquered by the barbarians, but that she was being starved by them; which is a very different picture.

^[15] 27.8.8: eisdem restituta omni praeter partem exiguam impensam militibus fessis: “having returned all things to [the victims of bandit raids] except for a tiny part handed over to the weary soldiers…” Now why on Earth would Ammianus mention this partem exiguam, if it were really so exiguam? It sounds remarkably like an excuse.

^[16]In which the emperor Julian (the Apostate) was the guiltiest party: long aware of growing barbarian encroachment in the island, he had made almost no effort to reverse it. SNYDER op.cit., 10: in 360AD, Julian - already emperor in the West - "was informed that the Picts and the Scots had broken a truce and were raiding the northern frontier lands of Britain. This was spreading alarm throughout the British provinces, and the morale of the army was said to be very low. Julian decided not to go himself, but to send his magister equitum Lupicinus to Britain with four units of his field army. Lupicinus waited out the winter in London, and was soon after recalled by Julian and arrested under trumped-up charges. While Julian was making his own play for power, the situation in Britain was growing steadily worse." - account taken from Flavius Lupicinus 6, in The Prosopography of the later Roman Empire, Cambridge 1971, 1.520-21, and Ammianus Marcellinus 20.1. Compare this with Julian's willingness to take exaggerated risks against the Germans in Gaul, and against the Persians in Mesopotamia: evidently, this young man, scarcely out of his teens and nourished exclusively on books, was more willing to defend the Roman Empire where he had illustrious literary precedents (Caesar, Alexander the Great) to guide him - to ultimate disaster. (One of the more peculiar phenomena in history is how many modern historians idolize this owlish enemy of Christianity; anachronistically bent on restoring an imagined golden age of Rome and Athens, insisting on behaving like an unhistorical mixture of Plato and Cato - in other words, reflecting the idolatrous bent of more than one historian towards classical antiquity).

^[17]The three terms of Triad 36 are actually the Coraniaid, the Gwyddyl Ffichti, and the Saxons; "and none of them went back". The Coraniaid were a legendary tribe of invading wizards from Arabia, who are part of the cycle of legends about two contending dragons that will meet later, and it is important to notice that they are directly connected with the legendary history of Britain which begins with the legendary king Lludd and ends with Vortigern and Ambrosius. The Gwyddyl Ffichti, literally "the Irish Picts", are an uncomprehending repetition of the ancient formula "the Scots and Picts", found in early documents as standard and closely allied enemies of Roman and post-Roman Britain; these Triads were written down at a time when the Picts had long ceased being an independent people, while the Irish origins of the kingdom of Scotland might still be remembered, and consequently the one people had been assimilated to the other, in legend as they were in fact. The Saxons need no comment.  The idea that "not one" of these groups of invaders "went back", however, harks directly back to the Gildasian idea of Britain as a unified single kingdom, originally populated only by Britons, from Land's End to John O'Groats, but later invaded and settled by Scots, Picts, and Saxons. The fact that the Coraniaid - who, like all later invaders, would not go back - had been completely exterminated by Lludd (or, according to the Triad, Casswallawn) may have served as a beacon of hope for the future.

^[18]Welsh and Breton sources agree that Gildas was the son of a king of Strathclyde called Cauus or Caunus or Caw. If he actually came from Strathclyde, the northernmost of the British post-Roman kingdoms, this would lend some point to his views on such matters as the Walls and other recognizably northern legends, as well as to his devoted love of the Latin language and of Roman ways - a love which seems to have grown in lands where Rome never set foot.

History of Britain, 407-597 is copyright © 2002, Fabio P. Barbieri. Used with permission.

Comments to: Fabio P. Barbieri