The North Sea in the northwest European context

The North Sea, in conjunction with Britain, Belgium, the Netherlands, northern Germany and Denmark, is often portrayed as the very northern margin of the earliest expansion of anatomically modern humans. As far as can be observed from the archaeological record, Upper Palaeolithic occupation (Aurignacian, Gravettian) of northwest Europe seems to have been thin prior to the LGM (Roebroeks, Reference Roebroeks, Roebroeks, Mussi, Svoboda and Fennema2000, Reference Roebroeks2014). Since the onset of the LGM, northwest Europe has been deserted, and it took about 10,000 years before Magdalenian hunter-gatherers recolonised the territory. The archaeological record suggests a more structural presence of Late Palaeolithic hunter-gatherer populations (Magdalenian/Creswellian/Hamburgian, Federmesser Gruppen, Ahrensburgian) ever since, possibly inducing the emergence of some regional differentiation, although some differences may be partly explained from taphonomical processes (see below).

At the time of Magdalenian initial recolonisation (∼13,000 BP), the North Sea basin, again, was in the zone of the occasional presence of hunter-gatherer groups. A small number of sites recorded in the limits of the loess belt in the southeastern part of the Netherlands represent the most northern occurrences of the late Magdalenian (Arts & Deeben, Reference Arts, Deeben, Burdukiewicz and Kobusiewicz1987; Rensink, Reference Rensink2010). The ‘classic’ Magdalenian seems to be absent outside the loess zone of continental northwest Europe, as it is in Britain. In this zone, however, the Hamburgian and Creswellian have been defined. The Hamburgian is generally characterised as a subgroup of the Magdalenian, and is identified in the northern Netherlands, northern Germany and extending to Poland. In addition to a number of English sites, including Hengistbury Head, the recent discovery of a Hamburgian site at Howburn, Scotland, has extended its geographical occurrence westward (Ballin et al., Reference Ballin, Saville, Tipping and Ward2010). Some Hamburgian implements have furthermore been reported from Skåne, Sweden (Andersson & Knarrström, Reference Andersson and Knarrström1999). As to the affinities of the Creswellian, which is mainly identified in Britain and to a limited extent the Netherlands, there has been considerable debate (Jacobi, Reference Jacobi, Barton, Roberts and Roe1991). Barton et al. (Reference Barton, Jacobi, Stapert and Street2003), however, considered the Creswellian to be Final Magdalenian.

What is of importance here, however, is the fact that the Magdalenian ‘range’ (i.e. Magdalenian/Creswellian/Hamburgian) extended well into the North, and that three interrelated techno-complexes theoretically ‘meet’ in the North Sea basin, roughly between 13,000 and 12,000/11,500 BP. Both Jacobi (Reference Jacobi, Barton, Roberts and Roe1991) and Keeley (Reference Keeley, Barton, Roberts and Roe1991) have drawn attention to the fact that the contemporary coastlines are now under water, and suggested that we have lost ‘the areas of densest and most continuous settlement’ (Jacobi, Reference Jacobi, Barton, Roberts and Roe1991, p.137). With sea levels about 70–80 m lower than today, the contemporary coasts have to be sought far in the North Sea and Channel/Manche (see Cohen et al., Reference Cohen, Gibbard and Weerts2014). Indeed, one may wonder to what extent the northwest European Magdalenian/Creswellian/Hamburgian represent an inland range of landscape use. Notwithstanding the technological similarities, regional variability can be recognised at the level of resource exploitation and settlement, e.g. in the Paris Basin, Ardennes and Rhineland (Audouze & Enloe, Reference Audouze, Enloe, Barton, Roberts and Roe1991; Street, Reference Street1998; Rensink, Reference Rensink2010). From this perspective, then, an important question has to be, what happened in the North Sea area? Did large herds of reindeer traverse the river valleys (cf. Grøn, Reference Grøn2005), and did this attract hunter-gatherers? Was it, as Jacobi (Reference Jacobi, Barton, Roberts and Roe1991) and Keeley (Reference Keeley, Barton, Roberts and Roe1991) believe, a relatively densely populated area that saw continuous settlement along the coast? Were the coastal zones exploited differently, giving way to the development of more complex hunter-gatherers (cf. Keeley Reference Keeley, Barton, Roberts and Roe1991, p. 188), and subsequent diversification in group identity, as can be seen among (near-) present-day coastal hunter-gatherers (Kelly Reference Kelly1995)?

In a way, much of what counts for the Magdalenian range also applies to the subsequent Federmesser settlement between 12,000/11,500 and 11,000 BP. Temporarily slowed down by the Dryas 2 cold spell, sea levels continued to rise at the time of the Federmesser occupation, which is mainly correlated to the Alleröd interstadium. Nonetheless, at ∼60 m below sea level, the contemporary coast was still far off the present-day coast line. Hence, the archaeological evidence from continental northwest Europe and Britain can be considered to represent an inland range of landscape use. The Federmesser Gruppen, as it is currently defined, covers an area spanning from Britain to the Ukraine, and from northern France/the Alps to Denmark and possibly southern Sweden.

Despite the vast number of sites, remarkably little is known with respect to the use of the landscape, mainly due to taphonomic processes (see below). As far as can be concluded from the few sites that produced faunal remains (e.g. Bedburg-Königshoven, Germany, and Doetinchem and Wierden, the Netherlands), the Federmesser hunter-gatherers exploited a broad range of animal resources, including a variety of mammals (elk, red deer, roe deer, horse, auroch, wild boar), fowl (goose) and fish (pike) (Street, Reference Street1998; Baales, Reference Baales2002; Lauwerier & Deeben, Reference Lauwerier and Deeben2011). In the sandy plain of the Low Countries, sites are frequent along the western and northern shores of lakes, as well as on river terraces (Deeben, Reference Deeben and Otte1988; de Bie & Caspar, Reference de Bie and Caspar2000). The British and German records show a comparable picture, albeit from a very limited sample of sites, with the addition of caves and short-lived uses of the forested inland (Elburg & van der Kroft, Reference Elburg and van der Kroft1997; Street et al., Reference Street, Baales, Cziesla, Hartz, Heinen, Jöris, Koch, Pasda, Terberger and Vollbrecht2001). A number of site types can be discerned (cf. Stapert, Reference Stapert1985): isolated projectile points, flint knapping sites, small special-purpose camps, multi-purpose dwelling sites and extended complexes that could relate to the contemporaneous use of several dwelling structures where people aggregated for some time (de Bie & Caspar, Reference de Bie and Caspar2000). The latter interpretation is, however, not undisputed, as there is the possibility of repetitive site use and palimpsest formation. In the case of Niederbieber, Germany, for instance, Baales (Reference Baales, Crombé and Vermeersch2004) has presented evidence for frequent and rather short-lived dwelling episodes, indicating high mobility among Federmesser hunter-gatherers as opposed to the Magdalenian.

In many aspects, the Federmesser settlement of northwest Europe is somewhat comparable to the Early Mesolithic: the broad-spectrum exploitation of animal resources, the variability in settlement locations and site types, and the repeated use of specific places, leading to the formation of extended palimpsests. This raises the question of how Federmesser hunter-gatherers used the North Sea area, where several large lakes must have been present (Gaffney et al., Reference Gaffney, Fitch and Smith2009). Could larger aggregation settlements, which we seem to ‘miss’ in the archaeological record, be present on the shores of these large lakes? ^{Footnote 1} Or have concentrations of people been present in the coastal zones, in a way comparable to what can be seen for the Late Mesolithic?

The Dryas 3 cold spell brought the favourable conditions of the Alleröd to an end, leading to an open landscape in which the dominant western winds had free play. Sea level was still ∼50 m lower than today, and the contemporaneous shore was far off the present-day North Sea coasts (Cohen et al., Reference Cohen, Gibbard and Weerts2014). As the polar conditions were of a very short nature, Ahrensburgian hunter-gatherers would soon exploit the landscape, in which large herds of reindeer were present. While the Ahrensburgian occurs in Belgium, the Netherlands and northern Germany (Rensink et al., Reference Rensink, Deeben and Stapert1996; Jöris & Thissen, Reference Jöris, Thissen, Fagnart and Thévenin1997), related and contemporaneous Bromme hunter-gatherers were present in the south Scandinavian territory (Andersson & Knarrström, Reference Andersson and Knarrström1999). In Britain, the ‘long-blade tradition’, synonymous with the Ahrensburgian, is considered to be of comparable date (Barton, Reference Barton, Aschton, Healy and Pettitt1998). As this was the case for the Magdalenian range, the North Sea basin, again, seems to be the meeting zone of these so-called tanged-point technocomplexes.

Several authors have attributed a significant role to the North Sea area, in particular with regard to Ahrensburgian hunting strategies and mobility (Arts & Deeben, Reference Arts and Deeben1981; van Noort & Wouters, Reference van Noort and Wouters1987; Arts, Reference Arts and Otte1988; Baales, Reference Baales, Kozlowski, Gruba and Zaliznyak1999). The river valleys and larger lakes in particular will have been attractors for game and humans. Baales (Reference Baales, Kozlowski, Gruba and Zaliznyak1999), for instance, proposed a model in which the Ardennes formed the spring territory for Ahrensburgian hunter-gatherers, while the sandy plains of the Low Countries, northern Germany and the North Sea area formed the major hunting grounds in winter. However, our knowledge on migratory patterns of reindeer during the Late Glacial is, as yet, limited. In this respect, the reindeer remains from the North Sea (Glimmerveen et al., Reference Glimmerveen, Mol and van der Plicht2006) are of major importance (see below for further discussion). If reindeer was an important resource to the Ahrensburgian (and related technocomplexes), it is necessary to understand the relationships between reindeer habitats and Ahrensburgian hunting strategies. Ahrensburgian technology persisted into the Preboreal (Deeben et al., Reference Deeben, Dijkstra and van Gisbergen2000), but would eventually be replaced by full Mesolithic micro-blade technologies. To what extent, then, does this relate to changes in the palaeoenvironmental setting?

In addition to questions about the scale of Late Glacial hunter-gatherer landscapes, migratory patterns and the exploitation of resources, the North Sea takes an important position with regard to questions about environmental diversity. Large parts of the Late Glacial landscape are now covered with North Sea water. What did the palaeogeographical layout of northwest Europe look like, and how did it change in the course of the Late Palaeolithic? To what extent and at what scale did spatial and temporal diversity in vegetation occur? How was the animal community composed in terms of species representation, population structure and geographical occurrence in space and time?

At the Pleistocene/Holocene transition, all parts of northwest Europe, including Scandinavia, would eventually be colonised by hunter-gatherers. Rapid sea-level rise invoked dramatic changes in the palaeogeography, while increasing temperatures, combined with altering precipitation patterns, led to rapidly changing plant and animal communities. Based on the Scandinavian evidence, it seems plausible that the early Holocene colonisation of the northern latitudes was strongly related to coastal exploitation (Bjerck, Reference Bjerck, McCartan, Schulting, Warren and Woodman2009). Indeed, coastal adaptations have been considered crucial to the spread of humans, and the emergence of complex societies (e.g. Bailey & Milner, Reference Bailey and Milner2002; Bailey, Reference Bailey and Flemming2004). Furthermore, the rapid submergence of Coles’ (Reference Coles1998) Doggerland led to the loss of territory, a process which is believed to have triggered increasing competition among Mesolithic hunter-gatherer groups (Waddington, Reference Waddington2007a; Gaffney et al., Reference Gaffney, Fitch and Smith2009; Momber, Reference Momber, Benjamin, Bonsall, Pickard and Fischer2011).

The concept of territory loss in the North Sea basin is one that sets the Later Mesolithic people apart from the Late Palaeolithic cultures of the relatively warm Alleröd period. At this time, the low sea levels following the LGM allowed people to occupy an extensive land mass that became habitable on the northwest European continental shelf. The subsequent depopulation of the region during the Younger Dryas triggered by the advance of arctic conditions would have seen the retreat of the flora and fauna. The process allowed the technologies to move where they could continue to be used, as discussed above.

The depletions caused by the climatic changes left a land that was denuded and ready for population growth when the conditions ameliorated. The Mesolithic hunter-gathers who dominated the more promising environment at the onset of the Holocene were the Maglemosian. Their presence has been recorded on both sides of the North Sea basin from Scotland to Poland (Clark, Reference Clark1936; Bang-Anderson, Reference Bang-Anderson, Kindgren, Knutsson, Loeffler and Ǻkerlund2003; David Reference David, McCartan, Schulting, Warren and Woodman2009). Like the Late Palaeolithic that preceded them, the first Maglemose people were pioneering, but they rapidly colonised the land in its full extent (Leakey, Reference Leakey1951; Rankine, Reference Rankine1952; Clark, Reference Clark1954; Reynier Reference Reynier and Young2000; Conneller, Reference Conneller, McCartan, Schulting, Warren and Woodman2009a).

The introduction of Later Mesolithic technologies within the next few thousand years occurred at a time when broad-leaf woodlands were advancing from the south and the natural landscape was changing considerably. Sea levels had also risen markedly and in doing so would have overwhelmed territories that were previously available for exploitation. This is also a time that saw an increase in the divergence of cultural technologies and the development of more regional identities within the European arena (Clark, Reference Clark1936; Jacobi, Reference Jacobi, Shennan and Schadla Hall1981; Gumiński & Michniewicz, Reference Gumiński, Michniewicz, Kindgren, Knutsson, Loeffler and Åkerlund2003; Bell, Reference Bell2007; Suddaby, Reference Suddaby, Waddington and Pedersen2007; Warren, Reference Warren, Waddington and Pedersen2007; Conneller, Reference Conneller, Conneller and Warren2009b; Lübke, Reference Lübke, McCartan, Schulting, Warren and Woodman2009; Schulting, Reference Schulting, McCartan, Schulting, Warren and Woodman2009; Wickham-Jones, Reference Wickham-Jones, McCartan, Schulting, Warren and Woodman2009; Momber, Reference Momber, Benjamin, Bonsall, Pickard and Fischer2011). Europe included the North Sea basin, which was rich in favourable fluvial environments. Being that the evidence recovered from the seabed (as outlined in this paper) suggests the area was occupied, the reduction in available land is very significant. This was compounded by the potential for increased attraction as it was potentially becoming more productive with the increased growth of protein-rich estuaries extending from the north and south. Such environmental changes have been recorded in the Baltic and the Solent, where the archaeology suggests technological and social change (Fischer, Reference Fischer, Pedersen, Fischer and Aaby1997, Reference Fischer and Flemming2004; Pedersen, Reference Pedersen, Pedersen, Fischer and Aaby1997; Grøn, Reference Grøn2003; Åstveit, Reference Åstveit, McCartan, Schulting, Warren and Woodman2009; Momber, Reference Momber, Benjamin, Bonsall, Pickard and Fischer2011).

The positioning of the North Sea as an important part of the Mesolithic world goes as far back as the early 20th century (Reid, Reference Reid1913; Clark, Reference Clark1936). The whole idea was based on assertions about the presence of Maglemose hunter-gatherers in Scandinavia and Britain, but, significantly, also on the presence of tree remains that were frequently ‘caught’ in fishing nets or were observed along the British coasts at low tide. Clearly, these indicated the existence of submerged forests, at the time referred to as ‘Noah’s Woods’ by many locals (Gaffney et al., Reference Gaffney, Fitch and Smith2009). Submerged forests were not only known from the British coast, but also from the Baltic (Pedersen et al., 1997) and the Netherlands (Wiggers, Reference Wiggers1955). Occasional finds of bone and antler implements of undisputable Mesolithic age, and trawled up from the sea floor, provided definite proof for human presence (Clark, Reference Clark1936; Louwe Kooijmans, Reference Louwe Kooijmans1971). However, the perspective taken was very much a terrestrial one, where the North Sea formed a natural land bridge, connecting Britain with the European continent. Coles’ (Reference Coles1998) paper took a radically different turn, positioning Doggerland as one of the core areas of Mesolithic northwest Europe (Fig. 1).

Fig. 1. Time series of palaeogeographic reconstructions of the North Sea basin as published by Coles (Reference Coles1998).

In view of the palaeogeographical evolution of the early Holocene northwest European landscape (Ward et al., Reference Ward, Larcombe and Lillie2006; Cohen et al., Reference Cohen, Gibbard and Weerts2014), it is apparent that much of the early Mesolithic world is now under water. This raises questions about the representativeness of the archaeological record as we know it from terrestrial northwest Europe. If coasts were highly attractive to hunter-gatherers, what did settlements along the coasts look like? Were there higher population densities, and was there continuous, year-round presence of people? How were coastal zones related to the inland? To what extent did the exploitation of coastal zones change with the shifting of coastlines due to sea-level rise? After all, sea-level rise does not only imply changes in the shoreline, but also in the characteristics of marine coastal dynamics (Beets & van der Spek, Reference Beets and van der Spek2000) as well as coastal ecology.

It is also within this framework of questions that the definition of a coastal site as opposed to an inland site should be considered. The work at Bouldnor Cliff in the UK (Fig. 2a) has revealed a site with the potential for year-round occupation and indications of sedentism (Momber et al., Reference Momber, Tomalin, Scaife, Satchell and Gillespie2011). At the time of the human recorded occupation at the site, which now lies 11–12 m below British Ordnance Datum, sea levels were lower still. The area was within a lacustrine basin just 10–12 km from the coastline and provided a wide diversity of resources. It could be argued that it benefited from being both a coastal and inland location as it was a site that was inland but yet within the coastal zone. However, by 8000 cal BP it became the coastline as the sea level rose. When this happened people adjusted and occupation continued. This was a process that all the people of southern England had to respond to if they were to continue exploiting the same established hunting and gathering territories.

Fig. 2. a. Geographic location of Bouldnor Cliff in the Solent, UK. b. Geographic location of archaeological sites and find zones off the Dutch coast. 1, Maasvlakte-Rotterdam harbour (Europoort); 2, Eurogeul; 3, Middeldiep; 4, Brown Bank De Stekels.

None of the answers to these questions will be straightforward as sea-level rise meant the environment was changing continuously and the effects would have varied across a geographically diverse area. We need to look at the sub-regional scale to ask, did the loss of territories have a negative or positive influence on evolving social interactions? Did the extension of coastlines induce greater technological adaptations in each case? If so can we see continuity of traditions after lands were drowned? Research into sites from Great Britain demonstrates that there was a divergence in cultural activity from mainland Europe during and particularly after severance (Woodman, Reference Woodman, Kindgren, Knutsson, Loeffler and Åkerlund2003; Bell, Reference Bell2007; Waddington, Reference Waddington, Waddington and Pedersen2007b; Conneller, Reference Conneller, McCartan, Schulting, Warren and Woodman2009a; Wickham-Jones, Reference Wickham-Jones, McCartan, Schulting, Warren and Woodman2009). It has been postulated that technological developments could have been stifled once Britain had become an island (Momber, Reference Momber, Benjamin, Bonsall, Pickard and Fischer2011); if so what impact did the transgression have on the smaller islands that were formed by the rise in sea level? Some of these may have lasted for a thousand years and more before finally being extinguished by the sea.

Apart from the effects of environmental change on the presence and distribution of crucial resources to hunter-gatherer communities, the gradual loss of land will undoubtedly have affected people in a mental way (e.g. Leary, Reference Leary2009), as did the occurrence of a tsunami, triggered by the Storrega land slide off the Norwegian coast at ∼8200/8300 cal BP (Weninger et al., Reference Weninger, Schulting, Bradtmöller, Clare, Collard, Edinborough, Hilpert, Jöris, Niekus, Rohling and Wagner2008) or the sea-level jump resulting from the Lake Agassiz drainage (Hijma & Cohen, Reference Hijma and Cohen2010). However, the impact of land loss and particular events such as a tsunami on people will have been very different, depending on the specific and variable conditions along the coasts. To draw inferences from the archaeological record with regard to such a topic, detailed insight in the timing and extent of these processes and events is needed with explicit attention to (sub-) regional variability. To what extent was the environment affected? Did this have an effect of technological developments between different locales? How long were highs such as Dogger Bank and Brown Bank accessible? Do polished flint axe blades from the North Sea indicate the presence of islands in the Neolithic, or are they related to perceptions of this seascape in reference to the land that once was? Indeed, this invokes further questions with regard to the sea-faring abilities of the Mesolithic and Neolithic people (Garrow & Sturt, Reference Garrow and Sturt2011; van de Noort, Reference van de Noort2011).

Archaeological evidence and taphonomy

So, what archaeological evidence do we have then to underpin the theoretical importance of the North Sea as outlined above? Admittedly, this is not very much as yet. In some cases we are dealing with direct evidence, that is, artefacts, bones with butchery marks and human remains. In other cases, indirect evidence permits some first inferences to be made with regard to some of the many questions passed in review. We will briefly discuss the evidence below.

Ever since the discovery of the ‘Colinda point’ on the Leman and Ower Banks in 1931, prehistoric artefacts have been brought up frequently. The majority of these finds originate from the Brown Bank and Eurogeul area, off the Dutch coast (Fig. 2b). An overview was published by Louwe Kooijmans (Reference Louwe Kooijmans1971), who discussed the finds in relation to sea-level rise, and, at the time, coarse-grained models of palaeogeographic developments. Owing to innovations in the fishing industry, and the use of beam-trawling in particular, there was a steady rise in the number of finds from the mid-1980s (Verhart, Reference Verhart and Fischer1995, Reference Verhart and Flemming2004; Glimmerveen et al., Reference Glimmerveen, Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Flemming2004). Close cooperation between Dutch fishermen and palaeontologists, who were keen to collect the many thousands of fossil bone remains that were caught in the fishing nets, was crucial. More direct research was initiated in Britain, where conditions for underwater surveys are somewhat better. One of the more exiting discoveries was that made at Bouldnor Cliff in the Solent (see above), where in situ remains of a Mesolithic settlement have been exposed (Momber, Reference Momber, Benjamin, Bonsall, Pickard and Fischer2011). However, the total number and quality of observations from the North Sea lags far behind those from the Baltic (Fischer, Reference Fischer1995, Reference Fischer and Flemming2004; Jöns et al., Reference Jöns, Lübke, Lüth, Terberger, Harff and Lüth2007) because of differences in accessibility and taphonomy.

The record as we presently know it shows some striking gaps. Several areas in the North Sea have produced Middle Palaeolithic artefacts (Glimmerveen et al., Reference Glimmerveen, Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Flemming2004; Verhart, Reference Verhart and Flemming2004; Tizzard et al., Reference Tizzard, Baggaley, Firth, Benjamin, Bonsall, Pickard and Fischer2011), and even a fragment of a Neanderthal skull (Hublin et al., Reference Hublin, Weston, Gunz, Richards, Roebroeks, Glimmerveen and Anthonis2009) in addition to uncountable numbers of Pleistocene mammal remains (Mol et al., Reference Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Glimmerveen2006). Next comes a reasonable number of finds, securely anchored in the Mesolithic (Preboreal-Boreal) by means of radiocarbon dates (Glimmerveen et al., Reference Glimmerveen, Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Flemming2004). However, finds that can securely be dated in the Late Palaeolithic are absent, except for the Colinda harpoon, which has been dated to 11,740 ± 150 BP (Housley, Reference Housley, Barton, Roberts and Roe1991). The Viking Bergen flint artefact found in a vibrocore off the Norwegian coast (Long et al., Reference Long, Wickham-Jones, Ruckley and Roe1986) could be of Upper Palaeolithic age, but as the specimen originates from washed sands, nothing can corroborate this; even a Middle Palaeolithic date cannot be excluded. In fact, this absence of evidence for Late Paleolithic activity is as surprising as the indication for Neolithic activity, which is evident from a number of Neolithic axe blades from Brown Bank (Maarleveld, Reference Maarleveld1984; Glimmerveen, Reference Glimmerveen2007) and Dogger Bank (van de Noort, Reference van de Noort2011).

As Roebroeks (Reference Roebroeks2014) points out, the North Sea basin may have been the limit of Upper Palaeolithic hunter-gatherer presence. Notwithstanding the taphonomical bias that affects the integrity of the archaeological record, the number of Aurignacian and Gravettian sites in the surroundings of the North Sea is extremely low indeed. Hence, it is not surprising that this phase is not represented in the North Sea record; it is a matter of statistical probability that such material would be found if present at all. Clearly, this is not the case for the preceding late Middle Palaeolithic, for which there is evidence in the North Sea. Notably, the Neanderthal frontal bone discovered off the Dutch coast (Middeldiep) seems to fit best the morphological features of late Neanderthals, to which several of the northwest European cave specimens (Spy, Neanderthal) belong. Indeed, the North Sea basin with its surrounding highs may have been one of the last Neanderthal refuges (Roebroeks, Reference Roebroeks2014). Neanderthals were present in this region during the early Weichselian/Devensian, and at least as far north as the northern Netherlands.

Why then do we hardly have any solid evidence for human presence in the North Sea area after the LGM, before the Mesolithic? After all, as the terrestrial evidence in both Britain and the continent shows, hunter-gatherers were definitely occupying the territory. In fact, as of ∼13,000 BP there is more or less continuous occupation that was only interrupted (locally?) during short-lasting Dryas cold spells. It is important to note that the palaeontological evidence equally shows a lack of post-LGM material that can securely be dated in the Late Glacial (Fig. 3). So far, only two bones (the aurochs Bos primigenius and a goat-like animal) are dated between 20,000 and 10,000–11,000 BP. Typically, a series of radiocarbon analyses on reindeer remains delivered only pre-LGM dates, while reindeer bone/antler from the North Sea shows no convincing indications of human activity (Glimmerveen et al., Reference Glimmerveen, Mol and van der Plicht2006). ^{Footnote 2} The contrasting evidence, then, is all the more surprising.

Fig. 3. Number of radiocarbon dates (in 1000 year lags) on bone remains from the North Sea (data from Glimmerveen et al., Reference Glimmerveen, Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Flemming2004, Reference Glimmerveen, Mol and van der Plicht2006; Mol et al., Reference Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Glimmerveen2006). x-axis, radiocarbon age; y-axis, frequency of radiocarbon dates.

Of course, one important question that has to be considered is to what extent erosion of Late Glacial sediments may have contributed to this gap in the archaeological record. The dynamics related to sea-level rise and the transformation of the area into a full marine environment will certainly have affected the preservation of earlier deposits. Reference to the geological records shows that the effects would have occurred locally rather than on a ‘landscape-wide’ scale (Cohen et al., Reference Cohen, Gibbard and Weerts2014). The impact of the transgression would depend on the topography at any given location. Where land is exposed, it will suffer attrition from the sea and erode, but where sites are protected from the force of the rising water, they can be protected by sediments.

Generally speaking, sedimentation will occur in sheltered, deltaic estuarine environments. Fluvial channels and coastal lakes would act as catchments for particulate matter. These sediments would drop from the water column when rivers slow as they meet the sea. In addition, as the sea levels increase pressure on the coast, water tables are forced up creating stagnant lagoons on land that was previously dry. Where fine sediments drop to form silt, terrestrial material can be encapsulated and preserved in an anaerobic environment. If undisturbed it can survive for many thousands of years. With time, as the sea continues to rise, sedimentation would continue to extend steadily inland, overwhelming coastal fringes, river banks and the shores of lakes. All such fluvial and marine site types are known to be attractive for human occupation.

A consequence of these processes will be the filling of palaeoriver channels, lakes and coastal plains with deep sedimentary deposits. Recent studies have identified a multitude of such palaeofeatures (Gaffney et al., Reference Gaffney, Thomson and Fitch2007). Where archaeological material is covered by these deposits they would be buried and preserved within the anaerobic silts beneath the seabed. Once the palaeolandscape becomes covered by open water they can become masked by mobile marine sediments or sand banks. However, the seabed is continually transformed and marine hydrodynamic regimes can modify erosive currents to expose areas of seabed that have remained undisturbed since their deposition. These exposures would invariably represent only a small part of any site at a given time. Notwithstanding taphonomic variations, this suggests there could be further Mesolithic material preserved beneath the seabed in associated archaeological horizons (Fig. 4). Unfortunately, once prehistoric material becomes exposed on the seafloor, organic artefacts will quickly degrade, and fine flakes and flints will be dispersed. The chance for their discovery would have to occur within a very narrow temporal and spatial window, therefore very few sites have been found exposed on the sea floor.

Fig. 4. Bouldnor Cliff: seam of in situ archaeological remains eroding out of a Mesolithic horizon 11 m below UK, Ordnance Datum. Organic artefacts remained well preserved for over 8000 years beneath alluvial deposits that filled the palaeochannel. Changes to the modern waterway are now causing exposure and loss of the submerged landscape.

When considering finds recovered from the surface of the seafloor the only way to discover any temporal patterning in the palaeontological record is by means of radiocarbon dating, as there is no possibility to date bone remains chronologically on the basis of degrees of fossilisation other than by approximation (Mol and Glimmerveen, pers. comm.). Another factor may be an incidental bias owing to the geographical occurrence of remains from various periods and the exposure of materials on the seabed, and the subsequent change of discovery in the context of fishing and aggregate extraction.

In view of the above, it goes without saying that for the time being we can only reasonably draw some inferences with regard to the occupation of the North Sea during the early part of the Holocene. The most instructive find occurrences currently known are De Stekels near Brown Bank and Maasvlakte-Europoort off the Dutch coast, and Bouldnor Cliff in the Solent, Britain (Fig. 2).

De Stekels represents a zone southwest of Brown Bank, where more than 100 artefacts and human remains have been trawled up over the past two decades (Glimmerveen et al., Reference Glimmerveen, Mol, Post, Reumer, van der Plicht, de Vos, van Geel, van Reenen, Pals and Flemming2004). Most artefacts are of bone and antler, and comprise tools such as perforated and socketed adzes, but some stone implements (mace heads) have been collected as well. The human remains comprise two lower jaws and several cranial fragments. Among the bone and antler is one zigzag-decorated adze, while in another case part of the wooden shaft has been preserved. This points to generally good preservation conditions and limited transport of objects since the exposure at the seafloor. The human remains may point to the presence of Mesolithic burials, and possibly a cemetery. If this indeed is the case, De Stekels hosts one of the earliest Mesolithic cemeteries known to date in northwest Europe.

The second occurrence is the well-known site of Maasvlakte-Europoort (Rotterdam, the Netherlands). Over 500 bone and antler implements, mainly barbed points, were collected in the 1970 and 1980s (Louwe Kooijmans, Reference Louwe Kooijmans1971; Verhart, Reference Verhart1988). The exceptionally high number of bone points from Maasvlakte-Europoort is unique in northwest Europe, but has parallels at Star Carr, Britain, and Hohen Viecheln, Germany. As yet, not much is known about the original stratigraphic position of the Maasvlakte artefacts, but recently conducted research in the extension zone of Rotterdam harbour demonstrates the presence of an intact Late Glacial to Early Holocene sequence (Vos et al., Reference Vos, van den Berg, Bunnik, Cremer, de Keine, Maljers and Mesdag2010). Important new insights are expected to come from a large aeolian river dune located in the Yangtze harbour basin. Targeted research based on seismic and core data (Fig. 5) demonstrated the presence of a Mesolithic site on the flank of this dune, and several thousands of flint and bone fragments have been recovered during an excavation conducted in 2011. Despite the fact that there is only limited control over the exact stratigraphical (vertical and horizontal) position of the finds, the site is of major importance as it demonstrates the presence of early Mesolithic settlement locations next to the rich harpoon assemblage. It will allow investigation of the behavioural context of the points from a new perspective.

Fig. 5. Maasvlakte-Rotterdam harbour. Left, high-resolution vibrocore showing an undisturbed archaeological layer (A) in the top of aeolian riverdune sand; right, seismic profile showing the river dune in the middle (copyright Deltares, Utrecht).

Finally, in the case of Bouldnor Cliff, in the Solent, underwater excavation works have documented parts of an eroding Mesolithic settlement showing signs of industrial activity (Momber et al., Reference Momber, Tomalin, Scaife, Satchell and Gillespie2011). Apart from worked flint, hearths, spalls of wood, carbonised wood and stockpiles of heated flints, over 20 pieces of interconnected worked wood have been found, many with evidence of working. Furthermore, the unusual find of fragments of string made out of vegetable fibres (Fig. 6) has been reported and a large piece of tangentially split timber has been unearthed (Taylor in Momber et al., 2011). The method of splitting timber in this way is a technology that does not appear in the British archaeological record for another 2000 years. We have yet to determine why this is the case. Was it that evidence of similar technologies have not survived on dry land or were certain skills lost following the drowning of occupation sites and separation from mainland Europe? The Solent was an area with raw materials where fresh water and estuarine resources were easily accessible, enabling year-round occupation. Did these skills evolve here in isolation? Bouldnor Cliff was occupied prior to the formation of the North Sea. Were there cultures existing at the same time with similar technical skills distributed around the estuaries and lakes that still remained in the North Sea Basin? If so, and if sites could be found, can they provide evidence for the genesis of some of the technological idiosyncrasies that now appear in the archaeological record on land?

Fig. 6. a. Bouldnor Cliff: a 10-cm long piece of Mesolithic twisted fibre string. b. Bouldnor Cliff: detail of the string.

A final question that the work at Bouldnor Cliff is helping to address is the process that enabled the sites to survive. The preserved landscape is covered in 7–8 m of estuarine sediment. It became exposed due to a change in the marine geomorphology that has stripped a channel through the deposit. It became exposed due to a change in the marine geomorphology that has stripped a channel through the deposit, which sits in section at the base of a 7-m thick alluvial deposit (Fig. 7). Which sits in section at the base of a 7-m thick alluvial deposit (Fig. 7). If the channel had not cut through the deposit, it would have remained buried deep below the seabed. The investigations by the Vista team of the University of Birmingham (Gaffney et al., Reference Gaffney, Thomson and Fitch2007) have located many similar sub-bottom geomorphological characteristics and sediment sinks in the North Sea. Many of these are still filled with sediment, with a high potential for the preservation of palaeolandsurfaces. The recovery of Mesolithic artefacts by trawlers demonstrates that similar terrestrial deposits are becoming exposed. The work at Bouldnor Cliff has permitted, for the first time in British archaeology, the documentation of a well-preserved, yet quickly degrading Mesolithic site under water and is highlighting the possibility of comparable sites within the southern North Sea.

Fig. 7. Bouldnor Cliff: schematic representation of the stratigraphy and exposure of the archaeological layer. The alluvial deposits that protected the palaeo-landscape are now being eroded to form a 7-m high section that is steadily receding.

Survival and research potential

The good preservation of Mesolithic remains at the occurrences referred to above infers a high research potential. However, it should be born in mind that we are just getting a first glimpse of what the North Sea may be hiding; we are largely unaware of how much is still out there. Above we have addressed some problems with regard to the surprising absence of evidence of Upper Palaeolithic finds in particular. Clearly, the question of survival of sites, or for that matter entire landscapes, has to be considered for the period we focus on in this paper. How fair is it to expect all phases of Late Palaeolithic and Mesolithic settlement, as documented in the countries bordering the North Sea, to be equally present below the seafloor? Put differently, what is the research potential of the North Sea compared to terrestrial northwest Europe?

A first step to get a better understanding of this problem was taken by Ward & Larcombe (Reference Ward and Larcombe2008). In their study they combined geomorphological features with an estimate of site survival, in order to come to a preservation rating in the southern North Sea area (Table I). However simple their approach may appear, it provides a clear-cut framework that can be elaborated in far more detail. An important matter that has to be dealt with is the spatial and temporal scale at which we would like to assess the research potential. Clearly, as Cohen et al. (Reference Cohen, Gibbard and Weerts2014) show, and is also recognised by Ward & Larcombe (Reference Ward and Larcombe2008), the specific nature and sequence of geogenetic developments can be very complex and variable owing to (sub) regional and local differences. Relative sea-level rise, for example, is far from uniform, even along the Dutch coastline (Kiden et al., Reference Kiden, Denys and Johnston2002). Differences in timing, combined with coastal geometry and sedimentation rates, can produce major differences in the palaeogeographical evolution, even over relatively short distances. The North Sea basin is a highly complex environment that cannot be summarised in terms of some simple models.

The complex nature of sedimentation-erosion dynamics becomes all the more apparent in the spectacular results obtained from analysis of 3D seismic data by Gaffney and Fitch (Gaffney et al., Reference Gaffney, Thomson and Fitch2007; van Heteren et al., Reference van Heteren, Meekes, Bakker, Gaffney, Fitch, Gearey and Paap2014). In contrast to what was long believed, the Dogger Bank area has now been shown to contain a variety of geomorphological structures, dating from the periods before submergence. However, within the very same area, expectations with regard to preservation potential are not uniform (Gaffney et al., Reference Gaffney, Thomson and Fitch2007, p. 115). In fact, higher degrees of preservation potential are strongly related to relatively small-scale landscape features, such as former fluvial channels and lacustrine features (e.g. in the case of Bouldnor Cliff). For that matter, it is important to note that technological developments in the field of seismic surveying and digital data analysis are facilitating the collection of increasingly detailed data. Examples are the possible detection of a drowned forest on the eastern end of Dogger Bank (Hansen, Reference Hansen1981) and the detection of a Mid Holocene tidal channel network in shallow water, off the Dutch coast (Rieu et al., Reference Rieu, van Heteren, van der Spek and de Boer2005). The survival of Late Glacial and earlier Holocene landforms is, of course, vital to determine the significance of the North Sea for the archaeology of post-LGM hunter-gatherers. Notwithstanding the fact that there is still much to gain in the construction of spatial and temporal frameworks, as well as the geogenetical validation and temporal anchoring of mapped landscape features such as at Dogger Bank, the recent developments are overwhelmingly positive.

However, the significance of the North Sea is not only a matter of preservation conditions. From our outline of the post-LGM settlement history of northwest Europe until the final submergence of North Sea islands, it should be apparent that the North Sea not only fills a crucial geographical gap, but was also the scene for major changes in the relationships between people and the land. From this perspective, even coarse-grained data, such as those of dredged-up and trawled-up finds, can be extremely informative. Through a combination of high-level taphonomic studies and intrinsic information from the finds themselves, patterns may be revealed that are meaningful for the understanding of settlement histories in a wider setting. Hence, the degree to which we can frame the place and time of materials, be it artefacts or whatever else that is considered to be of importance, defines a baseline for the underpinning of the North Sea’s significance. Through cooperation between the research community and industry, protocols can be agreed upon to stimulate the recording of find coordinates and sedimentary context. Radiocarbon dating can help in chronologically positioning individual objects, although we must remain cautious with respect to the interpretation of results (cf. Roebroeks, Reference Roebroeks2014).