by Patrick Nunn
In most Welsh stories about Cantre’r Gwaelod, the king is Gwyddno (rather than Brittany’s Gradlon), and it is his intemperate steward Seithenhin (rather than a dastardly daughter) who unlocks the flood gates causing the city to be drowned.18 The city, often said to be one of many, dominated a massive area of low ground named Maes Gwyddno (the Plain of Gwyddno), perhaps more than 2,000km2 (770mi2), in an area between Cardigan and Bardsey Island and the shores of modern Cardigan Bay.19 This land was said to have been ‘extremely good and fruitful and flat’, and its loss must have had a memorable impact on its occupants; an account from about ad 1450 mentions ‘the lament of Gwyddno … over whose land God turned the sea’.20
While some scientists have railed against any literal interpretation of these Welsh stories, rightly so in some of their more extravagant permutations, others have been persuaded that the story of Cantre’r Gwaelod is more likely ‘a legendary account’ dating from at least the fifth or sixth century ‘of an actual event’.21 As was argued for Australian Aboriginal societies (Chapter 2), ‘the conservatism of this tradition in both Wales and Brittany ensured the survival of this tale to the present day’. This analysis does not compel us to believe that these stories were based on observations in the fifth or sixth century, but rather that older stories about ‘great inundations’ became attached in medieval times to the heroes of the fifth and sixth centuries, people credited with founding Welsh (and perhaps Cornish and Breton) traditions.22 Similar stories about the Welsh coast that scholars have interpreted as built around memories of coastal inundation include that about Tyno Helig ([Lord] Helig’s Valley), which may have covered some 46km2 (18mi2) of land now underwater in Conway Bay,23 and the town of Caer Arianrhod in Caernarvon Bay, the ruins of which are reported to sometimes emerge at low tide.24
In the case of Wales we do not have to depend solely on such echoes of ancient traditions, for there is an abundance of physical evidence of former shorelines in Cardigan Bay and elsewhere by which scientists have been able to track the progress of the postglacial sea-level rise that flooded the area. There are submerged forests within a kilometre or two of the modern coast of Cardigan Bay (see colour plate section), but further out there are sediments containing peats that mark a time when forests of oak and pine stretched across much of this shallow bay.25 Were the sea level around 30m (100ft) below its present level, much of Cardigan Bay – at least that between Cardigan and Bardsey – would be dry land, a condition that would have been met at least 9,000 years ago. Given that the present form of Cardigan Bay was established a thousand years or so later, it is difficult to imagine a sizeable land mass being emergent and connected to the mainland more recently.
During the Last Glacial Maximum, about 20,000 years ago, when the sea level in this part of the world lay some 125m (410ft) or so below its present level, Ireland was connected to Wales and England, which were in turn connected to what is today mainland Europe. Postglacial sea-level rise gradually drowned the Ireland–Wales land bridge, leading to its submergence about 9,600 years ago, or perhaps a few millennia earlier.26 The timing is important, for it has the potential to lend credence to or to dismiss the idea that the traditions of unassisted human crossing of the Irish Sea may be based on actual events. The most common story is that of Brân the Blessed (Bendigaidfran), a Welsh hero, who went with his warriors to Ireland to rescue his sister from a bad marriage. In one account, at a time when the Irish Sea was ‘not so wide’, Brân crossed at least part of it ‘by wading’, but thereafter ‘the deep water grew wider when the deep overflowed the kingdom’.27 The story is reminiscent of Aboriginal Australians once having crossed places like Backstairs Passage or Clarence Strait by ‘a combination of walking and wading’, something that has not been possible for several thousand years (see Chapter 3). So might a similar explanation apply to the Irish Sea at the time of Brân? It may, but there are currently too many uncertainties to consider this likely, although one can see how, to rationalise the crossing of what is now a deep sea, it is necessary for Brân to be a giant in most extant versions of this story.28
What, therefore, can we say about the possible antiquity of stories of coastal drowning from the Atlantic coasts of north-west Europe? Using the same methods as for the Australian stories, explained at the end of the last chapter, we start by estimating the minimum depths below present sea level at which each group of stories (clustered by location) would be true. These depths are shown in Figure 5.2 below, together with sea-level histories for Brittany, Cornwall and Wales that cover the later part of the period of postglacial sea-level rise.
Figure 5.2 Sea-level changes along the coasts of north-west Europe over the past 10,000 years or so are shown on the left. On the right is shown a selection of locations of compelling stories about sea-level rise from north-west Europe, represented by bars showing the minimum depths at which these stories would be true.
The depth minima on the right of Figure 5.2 are of course estimates – the stories are too vague for anything else – but considered realistic nonetheless. Thus, for example, a submerged town (like Ys) a kilometre or so off the modern coast of the Baie de Douarnenez would today be 15–30m (50–100ft) below sea level. And if Cantre’r Gwaelod existed somewhere between 5 and 20km (3–12 miles) off the modern coast of Cardigan Bay, then it would today lie beneath 10–20m (33–66ft) of water. These kinds of numbers are not intended to be definitive, beyond disputation, but rather realistic approximations of where submerged places may once have been – a necessary preliminary to determining the minimum ages of these stories.
By relating the minimum depths shown on the right of Figure 5.2 to the appropriate sea-level curve on its left, minimum ages for these groups of stories can be calculated. For instance, by drawing horizontal lines from the top and bottom of the depth range for stories from the Baie de Douarnenez to the Brittany sea-level curve, then vertical lines from the points of intersection to the age scale, we can determine that for these stories to be true – to be a distant recollection of something people actually observed here – they must have endured for a minimum of 8,750–10,650 years. Similarly, for the stories from Cardigan Bay to be true, assuming the sea level at the time of observation to have been 10–20m (33–66ft) lower than it is today, they must have been created at least 9,000–10,250 years ago. The full set of dates for these stories is shown in Table 5.1.
Table 5.1 Water depths and age ranges for the seven groups of sea-level rise stories from the Atlantic coasts of north-west Europe. Age ranges refer to the most recent time at which the observations of lower-than-present water depths could have been made.
Similar caveats that were identified for the analysis of the Australian stories at the end of the last chapter apply here, principally those involving the crudity of both the water-depth data and the uncertainties in the stories.
Before proceeding, mention needs to made of Lyonesse, which in several accounts is said to have once occupied the area between the Scilly Isles and Cornwall now covered by ocean. In the analysis below (Table 5.1), it is assumed – solely for reasons of parsimony – that the stories about Lyonesse recall a time when the shallow platform from which the Scilly Isles rise today was dry land; in other words, Lyonesse was Scilly. This is assumed because the ocean floor between the Scilly Islands and Cornwall is today comparatively deep. A narrow, sinuous dry-land connection may have been possible when the ocean surface was 65m (210ft) lower, but for this to have been a few kilometres wide, it would have to have been 70m (230ft) lower. Although these depths are not shown in Figure 5.2, it is likely that the minimum age for the narrow land bridge would have been around 14,600 years ago, that for the wider one a millennium or so earlier.29
There are other drowning stories in Europe that do not lend themselves to being dated in this way but are still worth mentioning.
The east coast of England defines the western limit of the North Sea, the county of East Anglia marking the western stepping-off point for Doggerland, the broad land bridge that connect
ed the British Isles to mainland Europe during the last ice age. The rise of the sea level in postglacial times that saw the dismemberment and finally the drowning of Doggerland has also had dramatic effects on the poorly consolidated (and easily eroded) sediments that comprise much of the East Anglian coast. When you visit the coastal village of Dunwich (population about 183) today, it is astonishing to consider that it was once among the 18 largest towns in England, sitting at the head of a sheltered natural harbour, the bustling hub of a vibrant trade in fish, salt and cloth.30 However, well before it reached its zenith, Dunwich had already started losing its battle with the sea. Long affected by shoreline erosion (marked by cliff retreat), some 600 houses are reported to have been washed offshore during the winter storms of ad 1287–1288. Since then this sad story has continued, involving a loss of status, abandonment by merchants and a continuing physical loss; the last piece of All Saints, the only church to remain standing after the ad 1740 storms, collapsed into the sea in ad 1922 (see colour plate section).
The sedimentary vulnerability of the East Anglia coast is amplified by land sinking, one expression here of the slow, enduring downwarping of the North Sea Basin. Not only is this basin sinking – like many similar basins – under the growing weight of the sediment accumulated on its watery floor over millions of years, but it is also being undermined by the flow of its underlying crust to areas of surface rise, for example in Scandinavia.31
Given such a situation, you might expect East Anglian traditions to be rife with stories of ‘sunken cities’ and the like, but they are not. Aside from those that insist – as in countless other places with similar histories – that the droll sound of tolling bells can be heard from beneath the sea’s surface when it becomes agitated, I know of no other known folk memories of what was a prolonged and memorable process for the region’s inhabitants. This may be because, unlike in most other places from which we have such ‘drowning stories’, East Anglia has been substantially repopulated – quite frequently, perhaps, during the past few millennia – as successive waves of migrants and invaders, from Vikings to Normans, overran the country. Traditions would have been lost, and connections between people and place forgotten, while the vagaries of living along an unstable coast would have been rediscovered anew on many occasions.
Along the shores of the Mediterranean Sea are numerous ‘sunken cities’ that are of only passing interest to the topic of this book. For these cities are well known, and many have been fulsomely documented, not least because of the mass of written records that recall their existence … and sometimes the turmoil that accompanying their disappearance. Many of the cultures associated with them were literate so there is little dispute, if the obvious fictionalisations are sidelined, that they once existed and are now underwater. There is no need to depend in most instances on oral traditions. Add to this the fact that the material culture associated with most submerged Mediterranean cities is generally conspicuous; you cannot readily miss the statuary, the columns and harbours, towers and temples, walls and wells, indeed all the constituents of Greco-Roman civilisation that are well known to travellers and museum-goers in much of modern Europe.
A good example comes from the Nile Delta, where the Greeks built two cities – Herakleion32 and Eastern Canopus – to facilitate trade with Egypt and other parts of North Africa around 2,000 years ago. Once located on the now-infilled Canopic distributary channel of the Nile, the remains of these cities currently lie 5–7m (16–23ft) below sea level. Some 60 per cent of the sinking these cities have experienced is attributed to a combination of rising sea levels and the slow yet inexorable compaction of the delta sediments on which they were built. The remaining 40 per cent or so happened more abruptly as a result of the episodic collapse of the water-soaked sediments on which the cities were stood.33 Owing to the abundant written records about and from these cities, they were never truly ‘lost’ to humanity, although until about 1999 no one was quite sure exactly where they had been. Deltas are dynamic places, routinely flooded, with waters moving the loose sediments from which they are built from place to place; new distributary channels are carved during floods while others may be come infilled. Physical collapse often occurs, particularly along a delta’s seaward fringes, where underwater slopes are generally steepest. And of course in the eastern Mediterranean, which is more affected by tectonic paroxysms than most places, the impact of nearby earthquakes and locally generated tsunamis has often been to destabilise places like the Nile Delta.34
The seismo-tectonic and volcanic activity that affects much of the Mediterranean is another reason why this region’s submerged cities do not get much of a mention in this book. The land moves up and down – sometimes slowly, occasionally abruptly – as a result of the converging African and Eurasian plates. The former pushes northwards, while the latter resists but of course periodically fails, so in places each is sliced up and thrust beneath its neighbour. Friction causes earthquakes and oftentimes these cause the abrupt – and catastrophic – sinking of coastal lands. Take the Greek city of Helike, once the principal city in the district of Achaea, which is recorded as having been comprehensively destroyed during an earthquake (and accompanying tsunamis) in 373 bc. In fact, so comprehensive was its destruction that for centuries no one was quite sure where Helike had actually been … or indeed whether it had ever been! The basic problem was that the earthquake, which occurred along a thrust fault associated with the movement of the African Plate beneath the Eurasian, had caused Helike to sink by 3m (10ft) within a very short period of time.35 Then the tsunami came and dumped a layer of sediments over the top and thereafter, because the geography of the area had been comprehensively reconfigured, the site of Helike found itself within an ocean passage, and subsequently became buried by river and ocean sediments. It took some extraordinarily persistent detective work to relocate Helike, which today, because of all this sedimentation, is now back on the land – where almost nobody had thought to look for a sunken city.36
A few hundred kilometres away from these thrust belts, perhaps the same distance below the surface, the melting of downthrust crust produces masses of liquid rock that often finds its way to the ground surface, producing lines of active volcanoes in places like the Aegean Volcanic Arc (along which Santorini famously erupted, perhaps inspiring Plato to invent Atlantis, in 360 BC) and the Calabrian Arc (responsible for a range of volcanic activity from Etna to Vesuvius and beyond). A good example from the latter region of how volcanic activity can cause coasts to sink concerns the Roman-era ports of Baia, Miseno and Pozzuoli, parts of all of which now languish several metres below sea level. These former ports are all located within a caldera of the Phlegrean Fields volcanic complex (Campi Flegrei), which is notoriously active today. They were repeatedly submerged and uplifted on at least three occasions since their Roman occupation (see colour plate section).37
European and North African coasts do indeed have many submerged cities associated with them. Yet both the submergence of most of these is ascribable to processes other than postglacial sea-level rise (so submergence is consequently more recent than most of the other examples described in this book), and written records exist for most, removing any need to rely on oral traditions (which are the main source for the ‘recollection’ of stories about the other examples in this book). We now therefore shift our attention east to the coasts of the Indian subcontinent, where drowning stories more akin to those of Australia may exist. But before we describe these stories, consider that parts of the coast of India have been affected more recently by some epic subsidence.
On 17 December 1846, the members of the Geological Society of London assembled for their quarterly meeting. The title of one of the items, forwarded from a correspondent in India and read to the meeting – Extract from a Letter concerning a Depression lately produced in consequence of an Earthquake in Cutch, by Mrs Derinzy – puts one in mind of a nervous memsahib suffering regular attacks of the vapours, who experienced a minor earth tremor and suffered a br
eakdown as a consequence. But it is not that kind of depression that Mrs Derinzy wrote about. The depression was that of the ground surface.
The Rann of Kachchh (Cutch or Kutch) is a low-lying, shallow basin, fault bounded, and close to the modern border between India and Pakistan. It is unclear how long this rann (salt marsh) has existed, for in medieval times the area where it is now was connected to the Arabian Sea by a prominent distributary of the Indus River named the Kori. Today the Kori is no more, a natural dam 80km (50 miles) in length having blocked it. This dam, named Allah Bund, was created during the earthquake that rocked the area on 16 June 1819. The uplift that created Allah Bund was complemented by subsidence, which dropped an area at least 7km (4 miles) in diameter (including the British fort at Sindri) more than a metre.38 The resulting landscape change was even more impressive, for the combination of the uplift and the subsidence created a basin with a surface area of more than 1,000km2 (385m2) – the Rann of Kachchh.
Earthquakes affected the same area in the 1840s; Mrs Derinzy’s Depression probably occurred in the course of the one in June 1845. During this event, as in most, the earthquake was followed by a tsunami that flooded the low-lying Rann. Subsidence also occurred. An extract from Mrs Derinzy’s letter talks of a man crossing the area on a camel, expecting it to be dry. Instead,
… the guide travelled 20 miles [32km] through water … up to the beast’s body. Of Lak [Lakhpat] nothing was above water but a Fakeer’s pole (the flagstaff always erected by the tomb of some holy man); and of Veyre and other villages only the remains of a few houses were to be seen. 39
