Caerwent, Camelot and the magical coming-and-going lake

This page uses images which re-size to maintain a constant appearance across all browsers and resolutions. Please allow scripts or ActiveX controls to run if asked.

Geoffrey of Monmouth, writing in AD1136, says that Arthur held court at Caerleon in Gwent, South Wales, for several years. Chrétien De Troyes circa 1180 sees Camelot, Arthur's fortress of legend, as separate from Caerleon but strongly implies that they were close together - Arthur's court is said to have moved on from Caerleon to Camelot as if Camelot was some kind of "next stop".

In Le Morte d'Arthur, published in 1485, Malory says that Camelot was "called in English, Winchester", which is of course the name of a cathedral city in Hampshire which was the capital of Wessex, and of Saxon England, from the late 7^th to late 11^th C. But Caxton, in the introduction to Le Morte d'Arthur, states that Camelot was in Wales. Since printed books were such a rare occurrence at the time, I presume that Caxton had actually read the book for which he was providing the introduction, and he knew that when he said that Camelot was in Wales, he was saying "The place which Malory calls Winchester was in Wales". This has led some scholars to propose that when Malory equated Camelot with Winchester he had either confused Venta Silurum (the Latin name for Caerwent, a village in Gwent a few miles from Caerleon) with Venta Belgarum (the Latin name for Winchester), or he meant "translated into English, its name would be Winchester". "Winchester" and "Caerwent" actually mean exactly the same thing, one in Anglo-Saxon and Brittonic, one in Welsh and Brittonic - Wenta-chester and Caer-wenta, "fortified place with market".

Indeed, Caerwent might have been called by both names at the time, just as Edinburgh is called both Edinburgh and Dunedin, one in English, one in Gaelic. K Matthews supports the idea that the names "Winchester" and "Caerwent" were used bilingually and interchangeably. He sees it as the other way round, saying that "Kaer Guent" could refer to any of three British towns whose names included the root "Venta" - Caerwent, Winchester and Caistor St Edmund - and that ancient references to Caerwent may therefore refer to the modern Winchester. But the Winchester in Hampshire was in decline during the putative Arthurian period, while Caerwent was at that time one of the last great strongholds of Rome in Britain. The ancient Welsh folk-romance Culhwch ac Olwen (believed to be 11^th C or earlier although the oldest surviving copy dates from 1325) links Arthur with characters known to be associated with Gwent (although it relocates them to Cornwall - of which more anon) and Caxton states that Malory's "Winchester" is in Wales. I suggest, therefore, that Matthews is wrong and it's more likely that the "Winchester" being referred to is Caerwent.

Malory, who may or may not be working from an older tradition, has Arthur marrying Guenever at a church of St Stephen at Camelot, and burying twelve defeated rebel kings in the churchyard there. We know that Venta Silurum (later Caerwent) almost certainly had a church and churchyard already in operation by the late 5^th C, and that a church on that site would later be known as St Stephen's. It doesn't really matter whether it was called St Stephen's in Arthur's day or not, so long as it was called St Stephen's when the stories were being composed, and the authors of those stories weren't aware of any earlier name.

The theory then is that the church where Arthur was married was the original church at Venta Silurum, and that the town was Arthur's home base. It belonged properly to Caradog Strong-Arm, who according to the tales was one of Arthur's main supporters, and in the late 5^th C Caradog donated the whole walled town to St Tathyw to house a monastery. If Caradog donated Venta Silurum to the church following Arthur's death then Arthur must have died in or by the late 5^th C. The 9^th C historian-monk Nennius has Arthur dying in the early to mid 6^th C, but there is other evidence in favour of the earlier date so the issue is not clear-cut. In any case Geoffrey of Monmouth places Arthur's court in the Caerleon area for some years, not permanently, so Caradog might have passed the walled enclave and church at Venta Silurum on to St Tathyw after Arthur moved his base elsewhere, rather than following his death.

Whether or not Caerwent was where the putative real Arthur really lived, there are good reasons to think that it's where early Mediaeval scholars and story-tellers thought he lived. If it was, why would it have been called Camelot? About a mile and a half nor'-nor'-west of Venta Silurum is an Iron Age hill fort believed to have been used by Caradog - the rebel general one - and nowadays called Llanmelin, but its original name was first Llan-y-Gelli ("church of the grove") and then Caer Melyn. "Camelot" could be a corruption of that, especially if it ever got called Caer Melyn Guent (Melyn's fort near the market-town), which could easily end up as something like Caimeluent. And "Merlin Ambrosius" is normally translated as "Emrys from Carmarthen" - but perhaps it was really "Emrys from Caer Melyn".

In addition, the early Arthurian tale Culhwch ac Olwen places Arthur's court at a place called Celliwig in Cornwall. There are at least two or three places in Cornwall with similar-sounding names, but the people who are mentioned as being at the court at Celliwig are ones associated with the Caerwent/Caldicot area, such as Bishop Bydwini (an early bishop of Llandaff whose name is preserved at Bedwin Sands, offshore from the village of Rogiet about a mile west of Caldicot) and Caradog Freichfras. It's highly possible, therefore, that originally Arthur's court was said to be at Llan-y-Gelli and it got conflated with a similarly-named fortress in Cornwall while the story was still being orally-transmitted.

This does not preclude Arthur from having had a presence in Cornwall as well, and indeed if he did have that just makes it more likely that a base in Gwent could have got confused with a similarly-named location in Cornwall. It's the fact that the tales associate Celliwig with Bydwini and Caradog which suggests that it was at Llan-y-Gelli, rather than any problem with Arthur being in Cornwall per se.

From the mouth of the Troggy, the stream which flows past Caerwent and which was formerly more of a young river, to the start of Cornwall is about ninety miles by water, but the Sutton Hoo ship - built about a hundred years after Arthur's time and not showing signs of being particularly new or experimental technology - has been estimated as having a sailing speed of ten knots, i.e. about 11.5mph (see Edwin and Joyce Gifford, The Sailing Characteristics of Saxon Ships). Even allowing for embarking and disembarking, and assuming a favourable wind, if the ships a century earlier were of similar design to the Sutton Hoo one Arthur could have sailed from Venta Silurum to the near end of Cornwall in less than half a day. He could without much difficulty have set out from Caerwent on Tuesday morning, attended a conference and banquet in north Cornwall and been back in Caerwent by Wednesday evening.

Some historians think Camelot might have been Chester because there are reasons to think that the "Round Table" was a Roman amphitheatre and the monk Gildas said that there was a martyr's shrine at Camelot, and Chester had a vast amphitheatre with a martyr's shrine actually inside it. But Venta Silurum too had an amphitheatre, and a church of St Stephen alongside it (whether or not it was actually called that at the time, because what matters is that it would have been called St Stephen's when the stories were being written); and there was a very large amphitheatre at Caerleon (called Isca, "place on the river Usk", in the Roman period) eight and a half miles away - about ninety yards long including the seating, as against Caerwent's amphitheatre's forty-five yards or Chester's hundred and ten yards.

I wonder, I really do, whether the young JK Rowling visited Caerleon, only about fourteen miles from her home, and whether the amphitheatre there contributed to the development of the Quidditch pitch. But I suppose one stadium looks much like another.

Penny Mayes at Geograph reports that this large amphitheatre at Caerleon was indeed "Once known locally as King Arthur's Round Table" - although in fact it's slightly oval. In AD303 Isca was the scene of the execution of the first two named British Christian martyrs, the preachers Julius and Aaron, and of a bloody large-scale massacre of Christians in general. This ties in with the idea that there was a martyrs' shrine at "Camelot" - both Julius and Aaron had churches erected locally in their honour.

According to A Canter Through Caerleon, Isca, a.k.a. Isca Silurum, was in its day a considerable city, twice the size of Venta Silurum both in acreage and in having a combined military and civilian population of about six thousand (although with walls only twenty foot high to Caerwent's thirty), and was so well-appointed and cosmopolitan that it was known as "the second Rome". It included a large complex of stone-built barracks which are now the only ones still (partially) surviving in Europe, and a bath-house the size of a cathedral, as well as a riverside port that was one of the largest in Roman Europe, and the massive amphitheatre which could seat six thousand, the whole population of the town. Initially a frontline garrison, it became a civilian city around a regimental base depot in the 2^nd century, when the focus of action shifted to the North of England.

By the end of the 3^rd century, however, the Legions had gone and by the late 4^th century the fort had been largely dismantled and the town had been ransacked and was in severe decline, with cattle penned in the courtyard of the once grand baths. By the end of the 4^th C the only remaining Roman military presence in Wales was at Cardiff and Caerwent.

Caerwent seems never to have been as grand as Caerleon at its height - its residential areas tended towards wooden tribesmen's houses rather than grand stone villas - but it endured better. According to A Canter Through Caerleon, "Caerwent, it seems, was the western bastion of a final re-organisation of [Roman] defences in Britain in AD399. A few remnants of the proud force founded by Augustus over 400 years before [i.e. the 2^nd Augustan] may have participated, but the fortress at Caerleon no longer had a part to play."

If, indeed, Caerwent was Arthur's home base nearly a century later, he could easily have split some administrative functions between the two neighbouring towns, linked as they were by good fast Roman roads. He could have had his palace and church and his residential and commercial districts at Venta Silurum (Caerwent), whose name even means "fortified market town", and used Isca (Caerleon) and its amphitheatre essentially as a conference centre cum parliament. The main body of his army, including the cavalry, would be based in the old Roman barracks at Isca but there would probably be a garrison in the Iron Age hill-fort at Caer Melyn overlooking Venta Silurum.

Caer Melyn commands a line-of-sight to Venta Silurum. Its line-of-sight to Isca is interrupted by the ridge called Kemeys Craig, but at Kemeys Craig there are, in addition to the ruins of a Mediaeval wooden castle, important Roman and Iron Age earthworks on the crest of the ridge just where you'd expect there to be a beacon-relay between Caer Melyn and Caerleon.

There is another earthwork half a mile north and higher up the ridge, but even though it is about 35yds higher it doesn't command such a good view of Caer Melyn, because the hill just north of Penyworlod Farm very slightly cuts into the line-of-sight. The southern earthwork is at the ideal position from which to relay beacon signals from Caer Melyn to Caerleon, being directly in line between them and having line-of-sight to both. Move it a hundred yards north and it would lose sight of Caer Melyn; a hundred yards south and the hill south of Bulmore would cut off the line-of-sight to Caerleon. Fifty yards east or west and it would no longer be on the crest of the ridge at all but clinging to its side. That there is an earthwork at exactly that spot suggests that it was put there because it had line-of-sight to both Caer Melyn (a British fortress) and Caerleon (a Roman town), and tends to confirm the existence of a beacon relay from Caer Melyn overlooking Caerwent, to Caerleon.

In the event of an attack at Venta Silurum, a garrison at Caer Melyn could alert the main force at Caerleon by lighting a beacon. Even if the day was so foggy that a beacon could not be seen, a rider or even a runner could be sent the seven and a half miles to Isca. If the attack came from the east, from Saxon territory - which is the most likely - they would reach Venta Silurum before Caer Melyn and would have no chance of crossing a mile and a half of rather boggy ground and climbing a hill to seize the garrison in time to stop them from sending a message to Isca, one way or another, so any force attacking Venta Silurum, with its thirty-foot walls, would have a window of opportunity of between one and three hours before the cavalry arrived.

Arthur must have had a cavalry barracks somewhere nearby, if his court was indeed at Caerwent, for historians are agreed that if Arthur existed at all he was more cavalry general than king. According to A Canter Through Caerleon the 2^nd Augustan Legion, who were based at Isca/Caerleon, included a unit of a hundred and fifty cavalrymen, and there are some indications that in the early days of Isca the First Thracian Regiment of Horse, a force of five hundred, was stationed there. A hundred and fifty cavalrymen, or five hundred, would have to have substantially more horses than men to allow for lameness, colic, pregnancy etc., plus there would presumably be packhorses as well, plus given the famously high quality of Welsh ponies, even in those days, and the presence of a large oval amphitheatre, I'd be very surprised if there wasn't stabling for at least a few teams of chariot ponies as well. Isca therefore would certainly have provided enough stabling for a small unit of cavalry, and possibly for a rather large one.

A friend who has made a study of the period tells me that most of the battles attributed to Arthur were in northern England and southern Scotland, too far to reach easily from Caerwent and suggesting a base at Chester or Carlisle. Since Geoffrey of Monmouth places Arthur's court near Caerleon for some years, not permanently, he might have started out at Caerwent because that was where the remnant of the old Roman government was, and afterwards have moved further north. In addition, prior to the Industrial Revolution war was often seasonal and stopped when the soldiers had to go home to get the harvest in: it would be perfectly normal for a commander of the period to have a winter base, and Arthur may well have wintered in Gwent and then moved his base of operations to Chester from March to September.

Chris Gidlow, one of the historians who think that Camelot itself was at Chester, is quoted as saying that "... we know that one of Arthur’s two main battles was fought at a town referred to as the City of the Legions. There were only two places with this title. One was St Albans, but the location of the other has remained a mystery." This is either ill-informed or economical with the truth, for a town on the river Usk, clearly Caerleon, is actually called "Cair Legeion" in the 9^th C Historia Brittanum; and although this could be read as either "City of the Legions" or "Fort of the Legions" Isca at its height was definitely a city, with a population of around six thousand. This certainly raises the possibility that Caerleon was not only the military and administrative zone of Camelot but also the site of one of Arthur's battles.

There are also historians who think that the Round Table was near Stirling Castle, and this idea dates back at least to the 14^th C, being mentioned by the Scots poet John Barbour in 1375, by William of Worcester in 1478 and by Sir David Lindsay in 1529. An archaeological feature near Stirling Castle and known as the Round Table has been shown to be possibly old enough to be genuine, although of course it may have been called that because it looked like Arthur's legendary Round Table, rather than because it actually was it.

This raises the possibility that Arthur had several bases, and a Round Table at every one of them - or that there were several Arthurs all with the same modus operandi. "Arthur" probably just means "bear". Roman armies were known as the Eagles after the standards they carried; the Picts may have fought under a boar standard; the invading Saxon army followed a white horse banner and two war-leaders called Stallion and Horse (Hengist and Horsa); and the Vikings would later fight under the Raven Banner. Just as successive French Crown Princes were all called the Dauphin (dolphin), there could have been a series of British war-leaders all ceremonially titled The Bear and all fulfilling similar rôles.

But if there was one Arthur, with one main base of operations, then Gwent seems a more likely candidate. The sources which say that Arthur's base was at Caerleon, or just along the road from Caerleon, or maybe just across the Severn from Caerleon but associated with historical figures known to be associated with Gwent, pre-date John Barbour by two to three hundred years. Caerleon was in its day one of the largest Roman towns in Britain and had one of the largest ports in the whole Roman Empire, but we know that by the putative Arthurian period it had fallen into some disrepair and was no longer secure, whereas the smaller but more heavily fortified satellite town at Caerwent was still a Romano-British stronghold. It would be an obvious candidate for the rôle of headquarters of the anti-Saxon resistance, even if the earliest surviving writings about Arthur hadn't said that it was.

If Camelot was based at Caerwent but also involved the neighbouring fortresses of Caer Melyn and Caerleon, it would be hard to invent a more defensible position. From Chepstow to Newport you have a strip of low-lying land between the high hills and the water, Gwent Iscoed, Gwent below the Wood, fifteen miles long but only four miles wide. Within it you have Caerwent and Caerleon, not quite nine miles apart, each too far from the hills to the north to be fired down on from above, and too far from the water to the south to be easily raided by sea. In between them, on the southern edge of the hills, is Caer Melyn, in a pole position for lighting signal beacons and with a line of site to Venta Silurum and to a relay station at Kemeys Craig, two miles east of Caerleon.

The waterway is the Severn, too wide to cross very swiftly - anybody coming across the water from England has to come by ferry - yet too narrow for ships to suddenly appear from over the horizon. Any attack by water would have to come a long way parallel to the shore, and be visible all the way.

Any attack on Caerwent through the low-lying part of Gwent could be seen from Caer Melyn; any low-lying attack on Caerleon could be seen from Kemeys Craig; any attack from the hills could probably be seen coming from both Caer Melyn and Kemeys Craig. Neither Caer Melyn nor the outpost at Kemeys Craig could be put out of action before they could light a beacon or send a messenger, because they are both a mile or two from the town they overlook, separated from it by a lake or marsh and at the top of a very steep slope. An attempt to attack Caerwent and Caerleon simultaneously would mean half your forces had to go through the mountainous heart of Wales and then come round from the other end, rendering coordination between the two forces almost impossible, and an attack on only one of them would result in forces being swiftly summoned from the other, leaving any attacker crushed between hammer and anvil.

Curiously, if Caerwent was indeed the central component of Camelot then the populist idea of Arthur as a king in a Mediaeval castle isn't as far off the mark as it seems. With its square layout and compact shape, its 30ft-high curtain walls dotted with round towers and its lake at the south wall's foot, Dark Age Caerwent must have looked very much like an early Norman castle - except that inside the walls, in place of a keep it had a basilica.

The 9^th century compilation of local history and regional origin-myths called the Historia Brittanum was written in or around AD829, probably but not definitely by Nennius. In it the author among many other things lists twenty-eight cities in Britain one of which was called Caer Calemion, which could also be a corruption of Caer Melyn. If so it's evidence that the names of Caer Melyn and Caerwent had become confused with each other, for there's no way the fort at Caer Melyn itself could qualify as a city. It's only about the size of one and a half football pitches: you could get the three thousand people of Dark Ages Caerwent into it in an emergency but they wouldn't have room to do anything much except play knucklebones and sleep a lot - there wouldn't be room to carry on the commercial and social functions of a city.

If the name of the city of Caer Calemion is indeed derived from Caer Melyn, then, it must actually be Caerwent, a mile and a half south-east of Caer Melyn, which is meant. It would show that Nennius was aware of Caerwent, even if under a confused name, and this is important both to the history of the area and to John Nettleship's own personal history.

One of the lists in the Historia Brittanum was a list of Wonders of Britain. Unlike pre-existing lists of Wonders of the World, these were not man-made constructs but remarkable natural and supernatural features of the land itself. There are several slightly different surviving versions of the book, many of which include a list of twenty wonders, but not always exactly the same ones. Dr Andy Evans, author of the Wonders of Britain website, has sorted them into a combined list of twenty-six wonders, many of which are in the west of England and in Wales, since Nennius was based in the west.

Most of the wonders can fairly easily be matched to modern sites. For example Wonder N° 3,

pretty clearly refers to the Roman hydrotherapy pool fed by hot springs at Bath, in what was then the territory of a Saxon nation called the Hwicce.

Of these wonders, one of the most important and yet most seemingly obscure is a lake or llyn which is n° 6 in versions of the list produced in mainland Britain, but is omitted from the Irish versions. The "Teared" referred to in the text below is the Severn Bore. The Bristol Channel funnels in very sharply from a triangular section of coast into the Severn Estuary which has a maximum range between high and low tides of 49ft (the second highest in the world, exceeded only by the Bay of Fundy), with the result that when the tide is especially high it punches a wall of water up to 9ft high twenty miles up-river against the current.

"There is another wonder: it is the confluence of Linn Liuan; the mouth of that river flows into the Severn, and when both the Severn is flooded to The Teared, and the sea is flooded similarly into the aforementioned mouth of the river, both it is received into the lake/pool of the mouth in the mode of a whirlpool and the sea does not advance up. And a bank/shore exists near the river, and so long as the Severn is flooded to The Teared that bank/shore is not covered, and when the sea and Severn ebbs, at that time lake Liuan vomits all that it has devoured from the sea and both that bank/shore is covered and in the likeness of a mountain in one wave it spews and bursts. And if there was the army of the whole region, in the midst of where it is, and it directed its face against the wave, even the army the wave carries off through the force, by fluid full clothes. If, on the other hand, the backs of the army were turned against it, the same wave doesn't harm, and when the sea may have ebbed, then the entire bank, which the wave covers, backwards is bared and the sea recedes from it."

Geoffrey of Monmouth in his History of the Kings of Britain, written circa 1136, also describes the lake although it's not clear whether he had simply lifted his description from Nennius, or is working from a shared older tradition. In a Welsh version of the History, Arthur and a companion have a conversation which translates thus:

"And he told him also that there was another lake in Wales near the Severn, which the men of that country called Llyn Llivan; and that lake, when the sea flowed, received water into it, and swallowed it as though it had been a mountain, until it overflowed its banks; and if it chanced that any stood with their faces towards the lake, and any of the spray of the water touched their clothes, it was hard for them to avoid being drawn into the lake; but if their backs were towards it, how near soever they might stand to its edge, it would have no effect upon them."

Culhwch and Olwen, one of the earliest known Arthurian tales, deals in part with the rescue of Mabon ap Modron (Son, son of Mother), a Celtic hero cum demi-god and, in this story, a cousin of Arthur's, who was stolen from his mother at three days old and then kept imprisoned in cruel bonds almost since the dawn of time: so long ago that nobody can be found who remembers him or where he was taken.
Arthur sends Cei and Bedwyr, along with Gwrhyr who "knows all languages and is familiar with those of the birds and the beasts" to find someone who knows where Mabon is being held. They are passed from hoof to claw along a succession of increasingly ancient creatures - the Ousel of Cilgwri; the Stag of Rhedynfre; the Eagle of Gwernabwy - before fetching up with the oldest creature in Britain, the great Salmon of Llyn Llyw, who takes them on his back to Gloucester where Mabon is being held. Llyn Llew is here described as being a pool on the Severn (Hafren) and the story suggests that Gloucester, which itself is on the Severn, was reasonably accessible by water from Llyn Llew.

Elsewhere in the story Arthur and his men pursue the giant boar Twrch Trwyth from Ireland, across Wales towards the Severn, and meet up with him between Llyn Lliwan - almost certainly the same as Llyn Liuan and probably the same lake as Llyn Llyw, since spelling was an inexact science at the time - and the mouth of the Wye. Because the names are given in that order - Llyn Lliwan, then the Wye - and they were coming from the west, the implication is that they came to Llyn Lliwan first, and it was both west of the Wye and close to it. The mouth of the Wye lies just south of Chepstow (hence, "Wyedean") and about four miles east up the Severn from Caldicot, therefore Llyn Liwan was either at or close to Caldicot. John himself preferred that name for it, "Llyn Liwan", presumably because it was the middle ground between Llyn Liuan and Llyn Llyw, as well as because Llyn Liwan or Lliwan is the version of the lake which was the most clearly located west of and close to the Wye.

According to Celtnet Llyn Llew means "lake of leadership" but it seems to be more commonly rendered as "the shining [or clear, glossy etc.] lake" from a Welsh root lliw. "Lake of the flood/tide" from a root lliant is also possible, especially as it is described as connected to the Severn Bore, which used to begin as far downstream as Caldicot. The great salmon is said to swim upriver to Gloucester every day and as such he may personify the Bore, which makes it possible that Llyn Liuan, Lliwan or Livan started as Llyn Liant and means "lake where the Severn Bore is born".

So, according to the various descriptions Llyn Lliwan was in Wales, a shortish distance west of the Wye, on a river which flowed into the Severn, and close to the point where it did so. It had a whirlpool, probably close to the point at which the river connected it to the Severn, and it had the remarkable characteristic that as the tide rose the incoming water seemed to be swallowed up by the pool, and the water level in the pool and beyond it didn't seem to rise; but then as the tide turned the missing water came back in a rush, raising the water-level to cover a bank at the edge of the lake. To me the detail about the water pulling you in if you faced it but not if you faced away sounds like a sucking undertow: it is easier, I think, to resist an undertow which is flowing from front to back relative to you, as when facing away from a body of water which is trying to suck you back towards it, than from back to front, because of the way your knees bend.

The precise point at which the Mouth of the Severn becomes the Bristol Channel is unclear, but if Llyn Lliwan is both close to the Severn and in Wales, it has to be somewhere between Chepstow and Swanbridge (the point opposite Weston-Super-Mare, after which the channel opens out and becomes unambiguously ocean) - and the closer to the Chepstow end, the more likely it was to be considered "on the Severn". If it is considered to be the next significant coastal landmark west of the Wye, as Culhwch ac Olwen implies, then really it must be between Chepstow at the mouth of the Wye, and Newport/Caerleon at the mouth of the Usk. Knowing this, John with his interest in local history looked at the surrounding area and identified a possible suspect in the Troggy Brook which flows past Caerwent.

Although nowadays just a stream, archaeological and geological evidence shows that the Troggy was once a navigable river which carried boats to or near Caerwent, although its original bed has silted up almost to nothing - perhaps with help from an Anglo-Saxon land-reclamation scheme. It's associated with a small, ruined fort called Cas Troggi, which is somewhere near the start of the brook and is about four miles north-west of Caerwent or two miles north-west of Gray Hill, on the far side of Wentwood. Strictly speaking, although John just referred to the stream as "the Troggy", it's Castrogi Brook in its upper reaches, Caerwent Brook where it flows past the village and the Nedern Brook at the distal end where it flows into the Severn. In ancient books it is called the Taroci, like the Tarot cards, although the origin of the name is believed to be the Welsh Trochydd, "plunger", referring to a headlong mountain stream.

Although the Troggy is considered to be a stream, now, and not a lake, it floods and spreads into a pool during wet weather, and since "wet" is pretty much par for the course in Wale that means that except in very dry years, it actually forms a shallow lake between Caerwent and Caldicot, a larger village a couple of miles to the south, for more than half the year. If I understood John's comments on the 'phone correctly this seasonal lake also comes and goes much more abruptly than one would expect if it were simply being filled by the rain.

Instead of the ground rising smoothly from the current course of the brook, the Troggy sits in a depression around three hundred yards across, with defined boundaries which rise steeply to about the ten metre contour and then slope back more gently. You can see this on the map, where many field-boundaries follow the contour pretty closely, showing that there is a discontinuity in the ground at or near the contour - although if the Ordnance Survey map is accurate then the actual boundaries of the lake were probably at eight or nine metres, as many field boundaries follow the ten metre contour closely but lie just below/inside it.

Archaeological evidence shows ancient dwellings clustering along the edge of the contour but avoiding the area inside it, suggesting that this marks the boundary of a lake, or of an area of marsh around a lake - something too soggy to build on, at any rate - and there is an alluvial deposit, the silt left by an old river, which follows the course of the Troggy but spreads out to cover all or most of the area bounded by the contour. The walls of Venta Silurum are aligned to follow the edge of the alluvium.

When water levels are particularly low, not only does the lake disappear but so does a large middle stretch of the brook itself, from The Cwm just above Llanmelin, two miles north-west of Caerwent, all the way down to the junction with the Crick Brook about a mile and a half downstream of Caerwent. Yet the water still flows above and below the dry stretch - showing that as well as the visible, overground stream-bed the water is following a roughly parallel course underground.

Knowing that Gwent is on karst (soluble and partially dissolved bedrock, in this case limestone, forming fissures and pockets into which water drains) and consequently is riddled with caves, and that early archaeologists had spoken of iron mooring-rings still surviving in the south wall of Venta Silurum, John reasoned that this pool on the Troggy had been a permanent feature in the Dark Ages when the water table was higher and the Troggy less silted-up, and that it was connected to a subterranean reservoir or reservoirs by several narrow fissures which were acting as siphons. The fractures and hydraulics of karst are so complex that it is impossible fully to predict how such a system will behave, but it seemed likely that what was going on was that as the tide rose and water poured into the lake it flowed through into this hidden reservoir, but when the tide fell the fissure became a siphon which caused the excess water in the reservoir to be emptied back out into the lake until the two bodies of water were at the same level, even when that level was below the upper position of the fissure.

If (A) is the normal water-level, an unusually high tide can raise the level of the lake as far as (B) at which point any additional water coming in flows through the fissure into the underground reservoir, until it equalises with the level in the exterior lake - which takes a lot of water, because the underground lake is proportionately large.

If the level is raised to (C) the channel fills with water and an airtight siphon is formed, so that when the water level in the lake starts to drop again, water from the underground reservoir will continue to flow from the reservoir to the lake for as long as the level in the lake is still tending to fall below the level in the reservoir, even after the level has fallen below (B). The drop in the water level in the lake pulls on the water in the airtight tube: since little or no air can get into the fissure the water flowing down from the apex of the siphon on the lake side tries to leave a vacuum behind it and water is sucked up to fill that vacuum, over the apex and down into the lake as if through a drinking straw.

If the water level in the lake reaches (C) and forms an airtight siphon, and then the water in the lake starts to drop with the receding tide before the level in the reservoir has reached (C), the level in the lake will initially drop very fast, because as well as water flowing out above ground and back to the sea, there will still be water flowing through the fissure into the reservoir. Once the level in the lake drops below whatever the current level in the reservoir is, the siphon will kick in, the flow will reverse and water will pour back from the reservoir into the lake.

Spring water or rain water entering the reservoir or reservoirs through other channels may also overflow through the siphon into the lake, enhancing the outpouring into the lake as the tide drops, and causing the water level to rise suddenly at other times.

Llyn Lliwan reborn from the Troggy in flood 17/03/2007, looking south-west along the M48, photographed by Eric Woods. The village of Crick is at right foreground, Caldicot is at centre left and a silver slice of the Severn can be glimpsed at top left. Caerwent is just off-screen at centre right. The flood-water has risen almost to the presumed original limits of Llyn Lliwan, but if it had filled up all the way the triangular bite out of the water near the left of the picture would be surrounded by water to form an island.

Aerial view of the area from Caerwent to Sudbrook, showing the course of the Troggy and the position of the Whirlyholes as shown on the 1775 Caldicot Lordship Sale map and the 1777 Aram map of Sudbrook, and the likely outline of Llyn Liwan based on the 10m contour-line. Sudbrook Camp is at bottom right, and a dotted blue line ending at Sudbrook shows the original outfall of the Troggy.

█The course of Caerwent Brook when present.
█ Position of historic Whirlyholes derived from old maps and accounts.
█ Additional swallow holes which have appeared in recent years.
█ 10m contour-line shows likely boundary of Llyn Lliwan.
█ Line of sight of photograph by John (above).
█ Line of sight of photograph by Mike Simms (below).
█ Boundary wall of Venta Silurum.

There had been in this lake at least three whirlpools, known locally as Whirlyholes and suggested by John to have marked the "swallow holes" at which water was being sucked into or ejected from the underground system: one of them, now dry, is marked by a deep hollow. These could be quite dangerous: a woman born circa 1910 told John that her uncle had worked for a farmer whose son had been lost down a Whirlyhole and never found, although this may be a tale which has grown in the telling, since another informant spoke of a boy who drowned in one of the Whirlyholes after becoming caught in a barbed-wire fence. You can see some of John's own thoughts about the Whirlyholes and Llyn Lliwan, with photographs and maps, on the Caerwent Historic Trust website.

The Nedern now flows into the Severn Estuary two miles due south of Caerwent at Caldicot Pill - a pill (Old English pyll) being the outlet of a tidal stream - but originally its outlet was nearly a mile to the east, at Sudbrook. At Sudbrook, too, there was once a Whirlyhole, now partially tucked underneath a railway line on the edge of the fields of what is now Southbrook Farm.

Unfortunately the whole system was wrecked in 1879 when engineers working on the new Severn Tunnel broke through into a water-filled passage 170 feet underground and 400 yards inland, in the vicinity of the Sudbrook Whirlyhole. Water from this Great Spring, which continues to flow at a rate of 23 million gallons a day, is used by a local paper mill and a brewery and the remainder emerges from an outfall at Sudbrook, bypassing the Troggy. You can read about this event in detail in issue #421 of the Belfry Bulletin, the newsletter of the Bristol Exploration Club.

Part of the bed of the Troggy was also concreted to seal the swallow holes, and for a mile above Caerwent its original, meandering course was straightjacketed into a ruler-straight ditch slightly to the south of its natural bed. The driving of the Severn Tunnel, the breaching of the Great Spring and the concreting over of channels within the rock and of the stream itself has killed off most of the water supply to the Troggy and left Llyn Lliwan as a shallow seasonal flood among the grass.

Nevertheless John's identification of Llyn Lliwan seems very convincing. To be both in Wales and on the Severn, and within the range of the Severn Bore, Nennius's wonder-lake has to have been somewhere within a few miles of Caerwent, and there is no other known candidate in the area. There seems no doubt at all that there was once a narrow but deep and convoluted lake surrounding what is now the course of the Troggy Brook and extending from Caerwent to Caldicot and nearby Sudbrook, in Wales and on the Severn as the Historia states, and with at least three whirlpools in it. The geological evidence confirms that at least the lower reaches of that lake were almost certainly tidal during the Dark Ages, at least as far up as the Whirlyhole which is now just north of the M48, and the Troggy definitely has a subterranean component into which it partially disappears in dry weather.

There is, moreover, a traditional salmon-fishing site at Caldicot (not a common thing in the area), meaning that the tidal lake on the Troggy was probably also well-known for salmon and was a likely site for legend to locate a giant specimen.

That the lake was attached to a subterranean siphon is more conjectural, since such systems are believed to be rare, but the abrupt comings and goings of the magical, seasonal appearing and disappearing pool at Caerwent do suggest it, and if so then the lake would have been able to absorb an incoming high tide without seeming to rise, and then spit water back out as the tide fell, exactly as described in the Historia. Along with Dr Andy J Evans (the author of the Wonders of Britain site) and a student of tidal mechanics named Steven Perry, John co-wrote a paper on "Linn Liuan / Llyn Llyw: The Wondrous Lake of the Historia Brittanum's de Mirabilibus Britanniae and Culhwch ac Olwen", which appeared in the periodical Folklore in August 2008.

There was also another connection which even John didn't spot, but I did. He speculated about possible derivations of the name "Nedern Brook" but missed the possibility that it might have been "Nodons' Brook". Nodons was a local god of the hunt and fishing. A variant of the Irish Nuadha of the Silver Hand, he was also known as Nudd or Lludd and was, perhaps, the origin of the Fisher King of Mediaeval mythology, a rare male manifestation of the Sovereignty of Britain, whose protracted suffering and unhealing wound mirrors the suffering of the land itself. There was a major temple dedicated to Nodons at Lydney Park on the Severn, twelve miles upriver from Caldicot, and one of the images there shows a figure, presumably Nodons, in the act of catching a large salmon. If the Nedern Brook = Nodons' Brook, and the coming-and-going-pool on the Troggy = Llyn Llyw, then the lake associated in folklore with a giant, magical, ancient salmon was on a waterway named for a local deity associated with salmon - as well as being upstream from a salmon fishery.

The sole (not salmon!) problem with the identification of the Troggy as Llyn Llyw/Lliwan/Liuan/Llivan is that there is no surviving record of the Troggy or anything in the area being called anything like it - although if indeed Caerwent was the origin of Camelot, it is at least interesting that there was a lake near Caerwent which may have been called (among other variants) Llyn (=Lake) Llivan and the French Arthurian romances connect Arthur with a "Lady of the Lake" called Vivian. In addition, just east of Caerwent the Troggy crosses the fields of Slough Farm, formerly called Slow Howse. This has been interpreted (Joseph Alfred Bradney, History of Monmouthshire) as a corruption of Llwch, a pool or wet place, and that would certainly fit with the bogginess of the area: but if it's possible for the "S" to be a later addition I don't see why it shouldn't be 'sLlyw House.

Also, if the water-level was up to what is now about the nine metre mark all along that coast there may have been a lake alongside Isca/Caerleon, depending on how silted up, or not, the area around the Usk was, and it's interesting that the amphitheatre and Roman fort are sited just above the ten metre mark. A recent archaeological dig, carried out after John's death, found evidence of a very substantial Roman dockside - one of the largest ever found - close by Caerleon.

I can find no evidence of there being whirlpools or whirlyholes at Caerleon and the lake there would have been nearly half a mile across for much of its length, much wider than at Caerwent, which argues against it emptying and filling very abruptly even if it were on a siphon; and it seems to me unlikely that a very major port would be sited on a waterway which did so. Nor so far as I know is there evidence of a massive spring there, whereas we know there is one which used to feed the Troggy system. It's very unlikely therefore that a lake at Caerleon was the magical coming-and-going lake, but a lake at Caerleon would very likely be called Llyn Legion/Leion, and if there were two lakes nine miles apart, one called Llyn Leion and one called Llyn Llyw, the names may have got conflated.

It's also possible that the lake between Caldicot and Caerwent was itself at some point called Llyn Legion/Leion because Venta Silurum held out as a Roman stronghold for longer than Isca did. There must have been a period when the long wiggly lake on the Troggy was the only waterway in the area where there was still a Roman trading presence.

There was another possible connection with Llyn Llyw that I worked out only after John's death. In the Llyfr Taliesin we are told about the rescue of Gweir from a harsh prison called Caer Sidi, a four-towered fortress accessible by sea and near "flowing water". Caer Sidi is called the Fortress of Drunkenness, of Hardness, of Glass, of Hindrance, the Fortress of God's Heights, the Steep-Sided. A host equivalent to three ship's companies attacked it to rescue Gweir, and only seven returned.

Caer Sidi may or may not be the same as the castle of the demi-goddess Arianrhod, which is said to be the gateway between the mundane world and Annwn, the realm of spirits and the dead. Caer Arianrhod is mentioned as a prison in the Hanes Taliesin ("I have been three periods in the prison of Arianrhod"): it has become associated with a semi-submerged reef off the coast of north Wales but according to DruidDawn this identification only dates from the 16^th C, so its original position, assuming it to have ever had one, is open. Caer Sidi has been translated variously as the Castle on the Mound (linked to the word sidhe), or as the Spiral, Rotating or Zodiac Castle (sidydd means "Zodiac" in modern Welsh): these later translations may have been influenced by the presumed association with Caer Arianrhod, since Arianrhod probably means "silver wheel" and "Caer Arianrhod" is also the name of a constellation.

Whether or not Caer Sidi equates with Caer Arianrhod, there are definite similarities between the tale of the imprisonment and rescue of Gweir, incarcerated in Caer Sidi "from before the reiving of Annwn", and that of the Mabon in Culhwych and Olwen, imprisoned for so long that only the oldest creature in Britain now knows where he was taken. Mabon is rescued from somewhere which is accessible by water from Llyn Llyw: the location is named as Caer Loyw, Shining Castle, which is normally assumed to be Gloucester since that is Gloucester's name in modern Welsh.

Gloucester is indeed accessible from Llyn Llyw, if you swam right up the Severn, and in the Roman period it too was a grand city with high walls. However, it doesn't make much internal sense for Gloucester to be the prison of Mabon, for Mabon is described as wailing and crying so loudly in his misery that he can clearly be heard through the walls of his prison, at a point accessible by a salmon, and he is also capable of speaking audibly and coherently through the walls. I suppose the author might have been going for a Diagon-Alley-ish discord between the mundane and the magical, but still it's hard to see how the storyteller could expect his or her audience to believe that such a noisy and articulate prisoner could have been held at the waterline of a populous city thronging with boats, apparently for centuries, and yet nobody except a fish who lived in a pool several miles away knew he was there. It does suggest that in the original version, Mabon's prison was somewhere a bit less public.

If we assume then that the stories of Mabon and Gweir are the same story with a bit of drift, and that the castle in the story began as Glass Castle (Caer Sidi) and then was misremembered into Shining Castle (Caer Loyw/Gloucester), then there's another candidate.

It's hard to be really sure of anything about the original shape of the hill-fort at Sudbrook because about two thirds of it has been eroded and washed away by the Severn, but what's left suggests a rough square with bulging, convex sides, so it is very likely to have had four corner towers - or at any rate, to have looked to a Dark Ages storyteller as if it had once had four corner towers. It sits on a mound of sorts - the promontory of Sudbrook Point.

It also has additional ditches and banks, creating a terraced appearance, on the north-west flank where it slopes down to where the original mouth of the Troggy/Ned(d)ern was. The Rev. Samuel Sayer's drawing of 1821

Nowadays the place seems to have no special name - it's just Sudbrook Fort, or Sudbrook Camp. But since Sudbrook is or used to be the mouth of what in the Dark Ages was a very large brook, which entered the Severn immediately adjacent to the north-west flank of the fort, the name must be in two halves - Sud Brook - and there's a very good chance that the fort there, with its four corners and its spiral appearance when seen from the water, was once called something like Caer Sudd. And an ancient castle in the liminal space on the seashore, probably long abandoned and already beginning to be devoured by the sea by the time Culhwch ac Olwen was first told, seems like a much better bet for a magical prison than Gloucester. It may well be that in the original story it was assumed to be broken down and partially devoured by the sea because it had been the prison of Gweir/The Mabon, and Arthur's ships had battered it down.

Aerial view of a possible reconstruction of Sudbrook Point when it was less eroded by the Severn than at present. The map shows the original course of the Troggy, the position of the Whirlyhole which was the original outlet for the Great Spring and an approximate outline of Sudbrook Camp at a time when it was complete and intact.

█ Course of Caerwent Brook/the Troggy.
█ Sudbrook Whirlyhole.
█ Likely boundary of Llyn Lliwan.
█ Sudbrook Camp.

Another poem in the Llyfr Taliesin says of Caer Sidi that "...around its borders are the streams of the ocean // And the fruitful fountain is above it // Sweeter than white wine is the liquor therein", which again fits very well with a castle on a promontory sticking out into the seaward end of the Severn, with a vigorously upwelling (productive, fruitful) whirlpool upstream about 750yds to the north, fed by a spring which is known to produce "exceptionally pure" water which is now used by the local brewery and paper mill. Although the surviving copy of the Llyfr Taliesin dates from the early 14^th C much of the content is believed to be 10^th C and some 6^th C. The name "Sudbrook" is said to be Anglo-Norman but the naming of Caer Sidi in a (probably) pre-Norman Welsh text suggests that a word not all that dissimilar from "Sud" existed in Middle Welsh, and the "Sud" in Sudbrook might be a Welsh word which has been Anglicised - especially given the similarities between Sudbrook Camp and the description of Caer Sidi, and its proximity to and aquatic accessibility from the putative Llyn Llyw.

If, in fact, Sudbrook Camp was the inspiration for Caer Sidi, and given that one of the names of Caer Sidi in the Llyfr Taliesin is "the Fortress of Glass", that would provide at least a tenuous link between the concept lliw, shiny, clear, glossy, and the mouth of the Troggy. Why might it be called Glass Castle in the first place? Some burial mounds, such as Newgrange and the cairns at Carnbane West, both in Ireland, were faced with quartz when new and it has been suggested, possibly by Robert Graves, that Caer Sidi might also have had a glittery skin of quartz. An original facing layer containing at least a fair proportion of quartz is certainly possible at Sudbrook Camp, as the locally quarried Sudbrook Stone is "a bright, yellow sandstone with white quartz pebbles", and Sudbrook is only about nine miles across the Severn from Bristol, which is known for a kind of very hard, bright quartz crystal called "Bristol diamonds".

It is also true - although it could be coincidence - that all the photographs I've seen of the Troggy in flood, taken by at least four different photographers, show the water as reflecting the sky strongly and presenting a very silvery appearance. If this is an actual feature of the landscape it would account for the name "Shining Lake".

It was not only John himself, it seems, but parts of the landscape which he loved which were fictionalised and turned into magical story-book versions of themselves, and not just in the present but for more than a thousand years. Among the many reasons why I wish John was still alive, I wish he was still alive so that I could say "Look, look, see what I found!" - and I know he would be so pleased.