The Mesolithic-Neolithic transition in the sandy lowlands of Belgium: new evidence.

Philippe Crombe; Yves Perdaen; Joris Sergant; Jean-Pierre Van Roeyen; Mark Van Strydonck.

Until recently little was known about the Late Mesolithic-Neolithic transition in the sandy lowlands of northern Belgium. The only evidence was a few highly disturbed dry-land sites (Meeuwen, Dilsen, Weelde, etc.) which yielded a Late Mesolithic lithic industry, small amounts of Michelsberg/Hazendonk pottery and Neolithic tools (polished axes, arrowheads, large blades in mined flint, etc.). Based on these largely surface data, several neolithization models have been elaborated for the sandy area on the agrarian frontier in Belgium. In nearly all models, a long survival of the Late Mesolithic tradition is claimed. According to some of them, Late Mesolithic hunter-gatherers in the sandy area persisted almost without any influence from the adjacent Neolithic/agrarian groups of the Middle Belgian loss region until or after the arrival of the Michelsberg culture (Verhart 2000: 111-15, 231) or even until the start of the Bronze Age (Vermeersch 1990: 100-101). Some (Creemers & Vermeersch 1989; Vermeersch 1996) have even proposed a `transhumance' model in which these native hunter-gatherers were employed by Michelsberg farmer-herders for herding cattle in the sandy area. They also interpret the Michelsberg enclosures found at numerous locations in Middle Belgium as possible indications of some kind of tension between the two population groups.

Recently, however, new and more reliable evidence, mainly from wetland sites, sheds a totally new light on this topic. Salvage research in the valley of the river Schelde (Crombe 1998), in particular in the vicinity of the Antwerp harbour (FIGURE 1), revealed two important wetland sites--Melsele `Hof ten Damme' (van Berg et al. 1992) and Doel `Deurganckdok' (Crombe et al. 2000). The two sites are similar in their environmental setting: they are situated on elongated but narrow sand ridges of Lateglacial origin within the former flood plain of the Schelde river and are consequently sealed by thick layers of peat and alluvial sediments. The site of Melsele was investigated in 1984-1986 and 1990 over a relatively small area of c. 100 sq. m. A preliminary analysis of the findings combined with the results of extensive radiocarbon dating on different materials and contexts (Van Strydonck et al. 1995), however, indicates the existence of an enormous palimpsest at this locality. It is believed that the top of the dune has at least 3 different occupation phases--a Late Mesolithic, a Final Mesolithic with pottery and a Middle Neolithic--the remains of which have been irrevocably mixed by bioturbation processes. Although re-use and bioturbation are also attested at Doel, this site seems to offer better prospects.

The site of Doel `Deurganckdok'

The site of Doel was discovered and subsequently excavated during the construction of a dock in the Antwerp harbour situated on the left bank of the Schelde. The salvage operation was conducted by Ghent University in close collaboration with the regional archaeological service Archeologisch Dienst Waasland between May and September 2000. A total surface of about 4000 sq. m was investigated, but in extremely bad conditions, so that only part of the data could be collected properly.

The excavations revealed the presence of two prehistoric areas, one dated to the Final Mesolithic, the second to the Neolithic. Besides numerous hearth-pits, in both areas archaeological material, consisting of lithics, ceramics and burnt ecofacts (bones, hazelnut shells, burnt seeds, etc.) was collected from a c. 10-20-cm thick bioturbated layer at the transition of the peat and underlying cover sand. This layer has been interpreted as an old A-horizon which was sealed by peat from 5050 [+ or -] 55 BP (NZA-12075) onwards.

Final Mesolithic zone

Next to numerous flint artefacts belonging to an older Federmesser occupation, the lithic industry found at this location contains a series of artefacts with clear Late Mesolithic affinities. Among these (FIGURE 2) are a number of small parallel and regular (so called Montbani) blades, microliths--especially trapezes--and small scrapers made from Wommersom quartzite. Despite the close parallels with `pure' Late Mesolithic assemblages known from numerous locations in northern Belgium, these Doel artefacts, in particular the armatures, display two original characteristics. Compared to the Late Mesolithic trapezes, they are much smaller and of another morphology. Rectangular and rhombic trapezes are most common in the sandy lowlands of Belgium, generally provided with a flat inverse retouch at the base, on blades with an average width of c. 1.5 cm. The specimens of Doel belong to a symmetric or weakly asymmetric type of trapeze and do not display an additional basal retouch. Another difference is the total absence of microburins at Doel, while these waste products are generally well represented in Late Mesolithic assemblages.

Approximately 800 potsherds have been found stratigraphically, and in some places also spatially, associated with these lithics. From a techno-typological point of view, these ceramics closely resemble Swifterbant pottery, especially the Early Swifterbant pottery, which is regularly found in the wetland area of the western Netherlands (Louwe Kooijmans 2001a; 2001b; Raemaekers 1999) and which is partly related to the Ertebolle pottery from southern Scandinavia (Nielsen 1986) and the Dummer pottery from northwestern Germany (Deichmuller 1969). The pottery from Doel is moderately fired and tempered mainly with grog and organic material. Most reconstructed profiles (FIGURE 3) belong to S-shaped vessels with slightly outward-curving rim and round or pointed bottoms, although bowl-shaped vessels also occur. Decoration is limited and mainly consists of oblique impressions of finger-nails or spatulas on top of the rim as well as of small knobs.

Preliminary analysis of the numerous burnt faunal remains points to a small spectrum subsistence, principally red deer, boar and fish (Van Neer et al. 2001). No domesticates have been found so far. (1) The vast majority of fish bones (97%) belongs to the family of Cyprinidae, with species such as roach, rudd, and white bream. In addition there are charred remains of edible (hazelnut, wild apple, blackthorn) and less- or non-edible (red dogwood, hawthorn, ivy) plants. (2)

At present three radiocarbon dates are available for this sector (FIGURE 4). Two dates obtained on samples of organic (food?) residue preserved on pottery fragments--KIA-12260 (5980 [+ or -] 35 BP) and KIA-14339 (5835 [+ or -] 35 BP)--confirm the Early Swifterbant dating. Unless the dates are altered due to an unknown source of pollution these two dates cannot represent the same real age ([X.sup.2]-test: df=1 T=8.6 (5% 3.8)).

The interpretation of a third date, executed on a sample of carbonized hazelnut shell (NZA-12076: 5220 [+ or -] 55 BP), is much more difficult. Either it indicates a re-use of the location or an accidental burning of a hazelnut. But it is worth noting that indications of a younger occupation are rather scarce among the recovered findings.

Neolithic zone (3)

In this zone too, lithics and ceramics have been found in the same stratigraphical and spatial context. Compared with the previous zone, these finds are totally different. Most significant is the total absence of Late/Final Mesolithic type lithics, which are now constituted principally of typically Neolithic artefacts, such as a fragment of a polished axe, leaf-shaped and transverse arrowheads and a robust blade fragment with obliquely retouched edges (FIGURE 2). Some of these tools are made of a black-coloured high-quality flint which is not attested in the previous zone and which was probably imported from one of the known flint mines in southern Belgium (Hubert 1980).

Similarly there has been a clear shift in the process of pottery manufacturing. Despite the small number of potsherds and their extremely bad state of preservation, it is possible to detect major technological and typological differences. Pottery in this later sector is predominantly tempered with crushed and burnt flint, a tempering agent which is typical of the Middle Neolithic Michelsberg culture of the loss area in Middle Belgium and northern France. The morphology of the few vessels that could be reconstructed so far, e.g. a tulip-beaker (FIGURE 3), indicates the same cultural complex. This is corroborated by a radiocarbon date on the organic residue of a potsherd, which places the occupation at around 5110 [+ or -] 35 BP (KIA-14334).


The Doel data, although still under investigation, clearly contradict current theories concerning the Mesolithic-Neolithic transition in the lowlands of northern Belgium. They prove that indigenous Late Mesolithic hunter-gatherers, at least in the Schelde valley, were already changing around 5900/5800 BP, and thus well before the arrival of the Michelsberg culture, c. 5400/5300 BP. Changes in the lithic tradition are still minor but nevertheless significant, as in other Final Mesolithic traditions along the Atlantic coast, e.g. within the Ertebolle (Vang Petersen 1984: 10-13), the Swifterbant (Deckers 1979) and the `Teviecien' (Marchand 2000: 387-90). The tendency towards smaller and symmetric trapezes not made with the microburin technique seems to be widespread. The knowledge of pottery manufacturing was probably one of the first Neolithic innovations these native hunter-gatherers adopted from their agricultural neighbours, living in the loamy area of Middle Belgium and adjacent regions. The Early Swifterbant pottery found at Doel indirectly points to contacts with Grossgartach/Rossen groups, in which S-shaped pottery with rim decoration and knobs occur relatively frequently (Spatz 1996; Lichardus 1976). (4)

Another item that could have been exchanged between the two population groups in these early times are perforated antler-beam mattocks. Many mattocks are known from the main river valleys of northern Belgium, in particular from the Schelde basin. Nearly all are unassociated finds collected during large-scale dredging activities in the second half of the 19th and the first half of the 20th century. However, a recent AMS dating project (Crombe et al. 1999) in which a series of 15 perforated antler-beam mattocks have been directly dated, has proven their broad contemporaneity with comparable specimens from Swifterbant and Ertebolle contexts in northwest Europe. The dates span a period between c. 6100 and 4700 BP (or c. 5000 and 3450 cal BC). Some scholars (Troels-Schmidt 1967: 524; Fischer 1982: 11; Gebauer & Price 1990: 227) have suggested that these items could have been borrowed, imitated or brought from Neolithic communities, such as the Bandkeramik or the later Rossen. However, there is reason to believe that the `exchange' or `influence' went the other way round. Radiocarbon dates of two unperforated specimens (UtC-8239: 8940 [+ or -] 50 BP; UtC-8374: 8250 [+ or -] 45 BP) from the Schelde valley indicate that antler-beam mattocks were already in use during the Boreal period. It thus seems that the manufacture of this tool was a Mesolithic rather than a Neolithic tradition. Furthermore, direct dating of waste products from the production of antler-beam mattocks from the Schelde area (UtC-8229: 5770 [+ or -] 40 BP; UtC-8232: 5715 [+ or -] 35 BP) also proves that these implements were made locally rather than imported.

The Doel data would suggest that Mesolithic subsistence around 5900/5800 BP was probably not yet influenced by the Neolithic, as, at this point at least, no signs of domesticates or agriculture have been recorded. In the absence of organic remains on the Neolithic site of Doel, it is currently unclear at what moment animal husbandry and cultivation were really introduced into the sandy lowlands of Belgium. An interesting site in this matter is the Michelsberg site of Oudenaarde `Donk' (Parent et al. 1986-1987), situated approximately 100 km upstream from Doel, only 10 km beyond the sandy border in an area covered by sandy-loamy soils. This site, dated between 5240 BP and 4670 BP, has yielded numerous faunal remains, the analysis of which points to a semi-agrarian way of life. About 60-65% of the bones belong to domesticated animals, in particular to pig and cattle. Sheep/goat remains only account for 5 %. The remaining bones belong to different kinds of wild game, such as red deer, roe deer, horse, boar, wild cat, beaver and otter. In addition, there is strong evidence that fishing (pike, catfish, perch, etc.), fowling and gathering (hazelnuts, etc.) were still practised. Evidence of cereal cultivation, on the other hand, has not been found.

The above data clearly demonstrate that at least at the time of the arrival of the Michelsberg culture, subsistence in the southern part of the Schelde valley at the border between the sandy and (sand-)loamy region consisted of a mixture of
Mesolithic (hunting, fishing, gathering) and Neolithic (livestock) strategies, often referred to as an `extended broad spectrum subsistence system' (Louwe Kooijmans 1993: 102-3). Furthermore it can be deduced from the evidence of both Michelsberg sites (Doel and Oudenaarde) that around c. 5250-5100 BP the material culture in the Belgian lowlands was already completely `neolithisized', suggesting that the above mentioned dry-land sites which yielded Late Mesolithic and Neolithic artefacts and ceramics are most likely mixtures of once-separate occupation debris. (5) It is remarkable to note that a similar shift in culture occurred almost simultaneously in other parts of the Atlantic and Baltic coastal area, e.g. in England (Woodman 2000: 246-53), The Netherlands (Raemaekers 1999: 108-12), southern Scandinavia and northern Germany (Larsson 1986; Price 2000a: 272-3). This pattern has suggested to some that colonization was responsible, although others have emphasized the role of indigenous groups in the neolithization of the northwest European plain. There is, however, growing archaeological and anthropological evidence from various coastal regions in support of the acculturation model. Detailed studies in Scandinavia (Andersen 1989; Zvelebil 2000a; 2000b: 396-401; Price 2000a: 301-5), the western Netherlands (Louwe Kooijmans 1993: 101-3; Raemaekers 1999: 156-61) and the United Kingdom (Whittle 1990; Thomas 1997; Thorburnpe 1996: 99-118) have demonstrated a trend of continuity rather than a major break in the material culture (lithic and ceramic technology), economy, settlement pattern, burial practices and even social structure between the Mesolithic and Neolithic. Whether this also holds for the Belgian lowlands cannot be determined on the basis of presently available data. Yet the site of Oudenaarde `Donk' points to economic continuity. It is likely that wild resources remained attractive throughout the whole of the Neolithic period, as suggested by the discovery of a Final Neolithic Bell Beaker site with a mixed subsistence on top of the Michelsberg site at Oudenaarde (Parent et al. 1986-1987: 27-31).

Acknowledgements. Field-research at Doel was financially supported by Ghent University (Bijzonder Onderzoeksfonds, GOA-project n[degrees] 12050298), the Belgian Ministry of Public Works, the Archeologische Dienst Waasland and the European Commission (North Western Metropolitan Area Programme, Interreg II c). We would like to thank our colleagues A. Ervynck, A. Lentacker, J. Bastiaens (Institute for Archaeological Heritage, Asse-Zellik), W. Van Neer (Royal Museum of Central Africa, Tervuren) and B. Klinck (Ghent University) for supplying unpublished archaeo-zoological and botanical results. Thanks also to J. O'Driscoll (Ghent University) for correcting and improving the English text.

(1) However, it should be emphasized that, clue to the extreme fragmentation and burning, the analysis, in particular that of mammals bones, is severely hampered. So far it has been possible to assign to species only 27 out of more than 1000 analysed mammal bones. In contrast, about 55% of the collected fish remains are determinable.

(2) A [sup.14]C-dating programme will be initiated shortly in order to test the contemporaneity of these plant remains and the Swifterbant occupation.

(3) According to Zvelebil (2000a: 60) the term `Neolithic' should be restricted to indicate communities in which agro-pastoral farming has been introduced and developed. Despite the absence of subsistence data, we nevertheless chose to attribute tbis sector to the Neolithic on the basis of strong parallels in material culture with Middle Neolithic (semi) agrarian sites in the western Netherlands (Raemaekers 1999).

(4) According to W.-J. Hogestijn & H. Peeters (1996: 112), the Early Swifterbant pottery might have been influenced by the so-called Hoguette pottery, which is found in association with LBK material on many agrarian settlements at the edge of the northwest European loss area. Although the chronological framework of the Hoguette pottery remains uncertain, it is generally accepted that it is a product of Late Mesolithic population groups who lived simultaneously with and maybe partly prior to the LBK. In our opinion the currently available data are largely insufficient and do not allow verification of this assumption.


ANDERSEN, S.H., 1989. Norsminde. A `Kokkenmodding' with Late Mesolithic and Early Neolithic Occupation, Journal of Danish Archaeology 8: 13-40.

CAHEN, D. & M. OTTE (ed.). 1990. Actes du Colloque << Rubane et Cardial >> Liege: Universit de Liege. ERAUL 39.

CREEMERS, G. & P.M. VERMEERSCH. 1989. Meeuwen-Donderslagheide: A Middle Neolithic Site on the Limburg Kempen Plateau (Belgium), Helinium 29(2): 206-26.

CROMBE P. 1998. The Mesolithic in Northwestern Belgium. Recent excavations and surveys. Oxford: Archaeopress. BAR International series S716.

CROMBE P., J.-P. VAN ROEYEN, J. SERGANT, Y. PERDAEN, K. CORDEMANS & M. VAN STRYDONCK. 2000. Doel `Deurganckdok' (Flanders, Belgium): settlement traces from the Final Palaeolithic and the Early to Middle Mesolithic, Notae Praehistoricae 20: 111-19.

CROMBE P., M. VAN STRYDONCK & V. HENDRIX. 1999. AMS-dating of antler mattocks from the Schelde River in Northern Belgium, Notae Praehistoricae 19: 111-19.

DECKERS, P.H. 1979. The flint material from Swifterbant, earlier Neolithic of the Northern Netherlands. 1. Sites S2, S-4 and S-51. Final reports on Swifterbant II, Palaeohistoria 21: 143-80.

DEICHMULLER, J. 1969. Die Neolithische Moorsiedlung Hude 1 am Dummer, Kreis Grafschaft Diepholz, Neue Ausgrabungen und Forschungen in Niedersachsen 1: 28-36.

FISCHER, A. 1982. Trade in Danubian shaft-hole axes and the introduction of Neolithic economy in Denmark, Journal of Danish Archaeology, 1: 7-12.

GEBAUER, A.B. & T.D. PRICE. 1990. The end of the Mesolithic in eastern Denmark: A preliminary report on the Saltbaek Vig Project, in P.M. Vermeersch & P. Van Peer (ed.), Contributions to the Mesolithic in Europe: 259-80. Leuven: Leuven University Press.

HOGESTIJN, W.-J. & H. PEETERS. 1996. De opgraving van de mesolithische en vroegneolithische bewoningsresten van de vindplaats << Hoge Vaart >> bij Almere (prov. Fl.): een blik op een duistere periode van de Nederlandse prehistorie, Archeologie 7: 80-113.

HUBERT, F. 1980. Zum Silexbergbau van Spiennes, in 5000 Jahre Feuersteinbergbau. Die Suche nach dem Stahl der Steinzeit: 124-39. Bochum: Deutschen Bergbau-Museum.

LARSSON, M. 1986. Neolithization in Scania--A Funnel Beaker Perspective, Journal of Danish Archaeology 5: 244-7.

LICHARDUS, J. 1976. Rossen--Gatersleben--Baalberge. Ein beitrag zur chronologie des Mitteldeutschen Neolithikums and zur entstehung der Trichterbecher-Kalturen. Bonn: Rudolf Habelt Verlag. Saarbrucker Beitrage zur Altertumskunde 17.

LOUWE KOOIJMANS, L.P. 1993. Wetland exploitation and upland relations of prehistoric communities in the Netherlands, in J. Gardiner (ed.), Flatlands and wetlands: current themes in East Anglian archaeology: 71-116. Norwich: East Anglian Archaeology. Report 50.

2001a. Hardinxveld-Giessendam Polderweg. Een mesolithisch jachtkamp in het rivierengebied (5500-5000 v. Chr.). Amersfoort: Elve Ex Press. Rapportage Archeologische Monumentenzorg 83.

2001b. Hardinxveld-Giessendam De Bruin. Een kampplaats uit het Laat-Mesolithicum en het begin van de Swifterbant-cultuur (5500-4450 v. Chr.). Amersfoort: Elve Ex Press. Rapportage Archeologische Monumentenzorg 88.

MARCHAND, G. 2000. La neolithisation de l'ouest de la France: aires culturelles et transferts techniques dans l'industrie lithique. Bulletin de la Societe Prehistorique Francaise 97: 377-403.

NIELSEN, E.K. 1986. Ertebolle and Funnel Beaker pots as tools. On traces of production techniques and use, Acta Archaeologica 57: 107-20.

PARENT, J.-P., P. VAN DER PLAETSEN & J. VANMOERKERKE. 1986-1987. Prehistorische jagers en veetelers aan de donk te Oudenaarde, VOBOV-info: 24-25.

PRICE, T.D. 2000a. The introduction of farming in northern Europe, in Price (ed.): 260-300.

2000b. Europe's first farmers. Cambridge: Cambridge University Press.

RAEMAEKERS, D.C.M. 1999. The articulation of a `New Neolithic'. The meaning of the Swifterbant Culture for the process of neolithisation in the western part of the North European Plain (4900-3400 BC). Leiden: Leiden University. Archaeological Studies 3.

SPATZ, H. 1996. Beitrage zum Kulturenkomplex Hinkelstein--Grossgartach--Rossen. Der keramische Fundstoff des Mittelneolithikums aus dem mittleren Neckarland and seine zeitliche Gliederung. Stuttgart: Konrad Theiss Verlag. Materialhefte zur Archaologie 37.

THOMAS, J. 1997. The materiality of the Mesolithic-Neolithic transition in Britain, in A. van Gijn & M. Zvelebil (ed.), Ideology and social structure of stone age communities in Europe: 57-64. Leiden: University of Leiden. Analecta Praehistorica Leidensia 29.

THORPE, I.J., 1996. The origins of agriculture in Europe. London: Routledge.

TROELS-SMITH, J. 1967. The Ertebolle Culture and its background, Palaeohistoria 12: 505-28.

VAN BERG, P.-L., L.H. KEELEY, J.-P. VAN ROEYEN & R. VAN HOVE. 1992. Le gisement mesolithique de Melsele (Flandre-Orientale, Belgique) et le subneolithique en Europe occidentale, Revue Archeologique de l'Ouest supplement 5: 93-9.

VANG PETERSEN, P. 1984. Chronological and regional variation in the Late Mesolithic of Eastern Denmark, Journal of Danish Archaeology 3: 7-18.

VAN NEER, W., A. ERVYNCK, A. LENTACKER, P. CROMBE, J. SERGANT, Y. PERDAEN. M. VAN STRYDONCK & J.-P. VAN ROEYEN. 2001. Dierenresten uit een vroege Swifterbant-nederzetting te Doel-Deurganckdok (Vlaanderen, Belgie): jachtwild, maar vooral veel vis. Notae Praehistoricae 21: 85-96.

VAN STRYDONCK, M., J.-P. VAN ROEYEN, G. MINNAERT & C. VERBRUGGEN. 1995. Problems in dating Stone-Age settlements on sandy soils: the Hof ten Damme site near Melsele, Belgium, Radiocarbon 37(2): 291-7.

VERHART, L.B.M. 2000. ?mes fade away. The neolithization of the southern Netherlands in an anthropological and geographical perspective. Leiden: Leiden University: Archaeological Studies 6.

VERMEERSCH, P.M. 1990. La transition du Mesolithique au Neolithique en basse et moyenne Belgique, in Cahen & Otte (ed.): 95-103.

1996. Mesolithic in the Benelux, south of the Rhine, in S.K. Kozlowski & C. Tozzi (ed.), Acts of the UISPP colloquium `Formation of the European Mesolithic Complexes', Forli 8-14 September 1996: 33-9. Forli: ABACO Edizioni.

WHITTLE, A., 1990. Prolegomena to the study of the Mesolithic-Neolithic transition in Britain and Ireland, in Cahen & Otte (ed.): 209-27.

WOODMAN, P. 2000. Getting back to basics: transitions to farming in Ireland and Britain, in Price (ed.): 219-59.

ZVELEBIL, M. 2000a. Transition to agriculture in eastern Europe, in Price (ed.): 57-92.

2000b. Les derniers chasseurs-collecteurs d'Europe temperee, in A. Richard, Ch. Cupillard, H. Richard & A. Thevenin (ed.), Les derniers chasseurs-cueilleurs d'Europe occidentale. Actes du colloque international de Besancon, octobre 1998:379-406. Besanon: Presses Univeristaires Franc-Comtoises.

Antiquity, Sept 2002 v76 i293 p699(8)