Early-medieval Circular Fortresses In Zeeland

Vroeg-Middeleeuwse ringwalburgen in Zeeland (Early-Medieval circular fortresses in Zeeland) Robert M. van Heeringen, Peter A. Henderikx & Alexandra Mars (red) Authors: J. Buurman, R.M van Heeringen, P.A. Henderikx, I. Joosten, H. Kars, H.W. van Klaveren, R.C.G.M. Lauwerier, B. Oele, A. Pol, F. Verhaeghe. Goes/Amersfoort 1995. ISBN 90-72138-41-4

English summaries pages 229-239 1. History of the study - The natural landscape and the first human presence on the salt marshes - Archaeological evidence of occupation along the coast - Historical data on the period 810-839 - Historical data on the period 839-892 - The Zeeland fortresses: size and construction - Dating of the construction of the fortresses - The function of the fortresses - Historical data on the tenth century - Occupation within the fortresses - The finds - Historical data on the eleventh and twelfth centuries - Developments from the thirteenth century to the present day 2. Summaries of the material studies - Medieval Coins from the Beach at Domburg - Plant Remains from a Sewer of the Fortress at Middelburg - The Early Glazed Pottery found in Oost-Souburg - Slag and Ironworking in Oost-Souburg - he Stone Objects from Oost-Souburg - Objects of Bone, Antler and Horn from Oost-Souburg - Distribution and Function of Bone Three-pronged Objects - Stock breeding in Oost-Souburg

1. General summary This book is the culmination of the research project launched in 1990 by the State Service for Archaeological Investigations (ROB) on the early medieval occupation history of Zeeland in general, and more particular on the circular fortresses and their later occupation. An attempt has been made to present all the archaeological and historical information currently available, as well as the relationship between them. The book also contains the final elaboration of the results of the excavations at the fortress of Oost-Souburg, and a series of studies concerning the material culture that give us a better insight into how the early medieval society in Zeeland functioned. The main results are reviewed briefly below.

History of the study As shall be explained below, the province of Zeeland has at least five fortresses that were constructed in the last quarter of the ninth century. Their locations, where traces of some of them are still present below ground, are all typified by the toponym burg in the present place name: Oostburg in Zeeuws-Vlaanderen, Oost-Souburg, Middelburg and Domburg on the island of Walcheren and Burgh on the island of SchouwenDuiveland. A sixth fortress might be located near to the village of Kloetinge on the island of ZuidBeveland . In 1935 the historian J. Huizinga published a historical-topographical study which drew

attention to the fortresses on Walcheren and their relationship with the earliest history of the island. He convincingly linked the circular site on the border of the late medieval town centre of OostSouburg and of the double circle located within the street plan of Middelburg with the construction of fortresses in the Early Middle Ages, which is mentioned in the historical sources fort the Flanders and northern France region. Huizinga presented sound arguments for the existence of a burg at Domburg, although its precise location could not be determined. He called upon archaeologists to substantiate or dispute his conclusions. The archaeologist W.C. Braat of the Archaeological Museum in Leiden took up the challenge. With a series of trial excavations carried out between 1939 and 1952 he demonstrated conclusively that there had been fortresses and moats at Oost-Souburg, Middelburg and Burgh. The dating of the few finds roughly concurred with the historical dating to the ninth or tenth century. In 1969 it became necessary to conduct an archaeological investigation into the western half of the fortress at Oost-Souburg. The investigation was carried out by the ROB and was led by the then provincial archaeologist, J.A. Trimpe Burger. The excavations continued until 1971 and produced a wealth of information, particularly about the later settlement situated within the ramparts. In view of the historical significance of the fortresses location, the first, preliminary publication of the excavations in 1973 referred to the study published in 1965 by H. van Werveke, who had assumed that the Zeeland fortresses formed part of the line of defences that extended from northern France to Zeeland. This theory was based largely on a historical reference to fortifications erected prior to 891 (castella ibi recens facta) near St. Omaars in the estuary of the river Aa in northern France. From 1990 to 1994, the ROB undertook a new project, led by provincial archaeologist R.M. van Heeringen, focusing on the early medieval occupation history of Zeeland. The fortresses were one of the main subjects of study. The project involved a problem-oriented archaeological investigation designed to increase our knowledge of the fortress sites. To this end,

necessary rescue excavations were carried out at Burgh, Domburg, Middelburg and Oostburg. One of the most remarkable results of the field work was the discovery of the fortress at Domburg, and possibly a fortress at Kloetinge. Another aspect of the project involved collating all the available information on the occupation history. Independently of the ROB project, the historian P.A. Henderikx recently undertook a study of the early medieval occupation history of Walcheren. He has argued that the northern French, Flemish and Zeeland fortresses did not actually form a chain along the coast, but in fact can be divided into groups. They were concentrated in places where the seafaring Normans generally entered a country: at points where large rivers flowed into the sea. For instance, the fortress at Oostburg is situated on one of the two estuaries of the Zwin, and the other Zeeland fortresses are on the estuary of the river Scheldt, which flowed towards the sea along what is now the Eastern Scheldt. This new interpretation of the historical facts once more underlines the need for an indepen-dent archaeological dating of the construction of the Zeeland fortresses. The natural landscape and the first human presence on the salt marshes With the exception of a narrow strip comprising beach barriers and Older Dunes along the coast, the whole of Zeeland was flooded after the Roman period. Geologists refer to this as the Dunkirk II transgression phase. Early medieval occupation, in the sixth and seventh centuries, has been established only in the elongated dune landscape along the coast. As a result of erosion, in many places this original landscape can be reconstructed only on paper. In the course of the ninth century, and possibly as early as the end of the eighth century, a number of areas of marine sandy clay and clay had silted up to such a level that people were able to settle on them. Initially, visits to the 'new land' would have been only temporary. But since the huge salt marsh area was naturally highly suitable for the raising of sheep, in particular, and the mineral subsoil was found to be very fertile for arable use, the human presence there soon became a permanent feature. The prospect of salt extraction would also have attracted newcomers. Unfortunately, we have no precise archaeological dating for this pioneering phase on the salt marshes.

The oldest archaeological traces are layers of manure of unknown date, which have been observed under a number of small terps that can be roughly dated to the tenth century although their exact age is uncertain. The layers of manure might indicate places where sheep were kept. No settlements are known from this period. One may assume, by way of a working hypothesis, that the activity observed does not pre-date the ninth century. However, there are references to salt marsh meadows and salt extraction in eighth century texts. It seems obvious to assume that the huge salt marsh area was especially exploited from the occupa-tion centres along its edges, thus from the Wa1cheren and Schouwen dune area, and from the Flemish sandy area.. Archaeological evidence of occupation along the coast There was an important settlement in the Older Dunes at Domburg from the sixth to the early ninth century. The remains of this settlement and the associated cemetery(ies) were visible for 600 m along the high tide line on the beach in 1866. Remains of wooden houses and numerous graves with coffins were recorded by amateur archaeologist avant la lettre I.C. Frederiks from Oostkapelle. Unfortunately, no attention was paid to the dating of any of these remains or their interrelationship. Attention was subsequently focused mainly on washed up coins and other, mainly metal, objects. Few of the other artefacts found, including fragments of pottery and stone objects, were collected. Study of the coins, more than a thousand in number, has shown that the settlement existed from the sixth to the ninth century. The settlement might be the Walichrum mentioned in the sources; the name literally means 'wet sandy ridge'. The name Walcheren eventually came to be applied to the entire island. As an emporium, the settlement was involved in long-distance trade between the Frisian area and England, among others, and its status can be compared with that of Dorestad in the central Netherlands river area. A renewed study of the youngest coins, Carolingian denarii, carried out by the numismatist A. Pol of the Cabinet of Coins and Medals in Leiden, has suggested that Wa1cheren had become less important by the mid-ninth century. Clearly, this may well have been linked to the historically recorded Viking attack on Wa1cheren in 837.

As we stated above, the island had a seditio, or military settlement, although no evidence of physical defences were found under the settlement remains on the beach. There is at any rate no link with the fortress at Domburg. These defences lie around a kilometre to the south-west and are almost half a century younger. There must also have been a significant site on the island of Schouwen, since similar finds were found on the beach on the other side of the Scheldt estuary. Identification with Scaltheim mentioned in the early medieval sources however is nothing more than speculation. Historical data on the period 810-839 From the end of the seventh century the Zeeland coastal area formed part of the Frankish empire. Little is known for certain about how the emperor Charlemagne (768-814) and his successor Louis the Pious (814-840) ruled the province of Zeeland, situated as it was on the periphery of their empire. In principle, the emperor delegated some of his power to a count, who held sway over a certain area of one or more pagi (countships). We know the name of one countship in the Zeeland coastal area, the pagus Rodaninsis. This countship, which was situated in what is now Zeeuws-Vlaanderen, was named after the river Rodana, for which Aardenburg (Rodenburg) was also named. The countship occurs in historical sources between 707 and 840. We also know the name of a count, Eggihard, who was mentioned in 837 in relation to the island of Wa1cheren. The king possessed a number of estates which were managed by bailiffs. The management of the countship and the king's lands was probably in the hands of one person, the count. The sources also refer to homines franci, who played a role in the military defence of strategically located areas. The Vikings first attacked the Frankish empire in 810. Charlemagne had watches set up along all the estuaries, including the Scheldt estuary, and built ships to stop the Vikings landing. In 811 he even visited Ghent personally to inspect the progress of the work. In 820 thirteen Viking ships appeared off the Flemish coast. The defence system appears to have worked, since the Vikings were forced into retreat by men who had entrenched themselves in

a praesidium (a fortified guard post). One can deduce from a report in 834 that the custodia maritima, the coast guard, was still intact in the Flemish coastal area after this campaign. When the war for the succession of Louis the Pious broke out, the attacks by the Normans also became more frequent. From 834 to 839 the entire coastal area was pillaged and many Frankish trading settlements were destroyed, including Dorestad in the central Netherlands river area, Witla in the Maas estuary, and Antwerp and Walcheren (Walichrum) on the Scheldt. Unfortunately, we are unable to deduce with certainty from the written reports whether the settlement of Walcheren was made physically capable of defending itself prior to the attack in 837. There is mention of a seditio on Walcheren; in Medieval Latin, this word can refer either to the presence of a military settlement, a garrison, or to a reinforced garrison. The surviving text does however indicate that there was some kind of military settlement on the island of Walcheren. Eggihard, the count of the area, Hemming, a Danish leader fighting for the Franks and other high-ranking individuals are mentioned. Historical data on the period 839-892 After the death of Louis the Pious in 840, the Frankish empire was divided into three parts by the Treaty of Verdun in AD 843. The border between West Francia (the Kingdom of west franks) under Charles the Bald and the Middle Kingdom under Lothair I was the river Scheldt, although in Flanders (which included the present province of Zeeuws-Vlaanderen) the border turned slightly towards the south. Present-day Zeeland formed part of the Middle Kingdom, apart from the area around Aardenburg and Oostburg, which belonged to West Francia. After the division of the empire, regional rulers increased their power, as evidenced, for instance, by the creation of the county of Flanders under Baudouin I of Flanders (862-879). His successor, Baudouin II (879-918), was the first truly powerful example of these new regional rulers, whose county included the Oostburg area. Little is known about public authority in the Zeeland area governed by Lothair I. Since the time that Lotharius granted Walcheren as land in fief to the Norseman Harald (841), most likely under pressure from adverse political conditions, it can be assumed that the imposition of central

authority in the region was not wholly effective. After the death of Lothair II, Lorraine was divided by the Treaty of Meersen in 870. For a short time, the Zeeland coastal area then became part of West Francia under Charles the Bald. After his death in 877, Zeeland finally, and permanently, became part of East Francia (under the Treaty of Ribemont in 880). It was ruled by Charles III the Stout ( 882-887), Arnulf (887 -895), Zwentibold (895-900), Louis the Child (900-911) and Charles III the Simple (911-923). After 925, 'Zeeland' Lorraine became part of the Holy Roman empire. Again, a lack of contemporary written evidence means that we have no information regarding the actual power relationships in the Zeeland coastal area. During the periods 850-864, 879-885 and 891892 the Scheldt area was repeatedly attacked by Vikings. At the beginning of the first period, most of these were lightning attacks made by ship. Having entered the country via the Eastern Scheldt, the Vikings set St. Baaf's Abbey in Ghent on fire in 851. In 859 and 860 they plundered the area to the east of the river in Lorraine, and the area upstream towards France. They attacked Flanders from the sea in 850, 860, and 864. There are few sources about the coastal defences. Any reports that do exist refer to West Francia, the Flemish countship. There must have been fortifications in Bruges, and it is also known that Charles the Bald charged Count Adalelmus, count of Laon, with the defence of the Flemish coast in 853, but it is not known how he carried out the task. In July 879 the huge Norman army crossed to the continent after their defeat in England. They first sacked the Flemish coastal area and then sailed up the Scheldt to plunder Brabant. They set up camp for the winter six times between November 879 and October 885, in Ghent, Kortrijk, Asselt, Conde, Amiens and Louvain. By this time they had changed their battle tactics. Since most of the wealthy places near the river had already been pillaged, they used horses to venture further inland. Their forays mainly took them south and east, where they fought many battles with the local population. For instance, in November 880 the Normans defeated from their camp at Kortrijk the Flemings and Zeelanders. The chronicler of this incident stated that the Normans were particularly vengeful towards the Zeelanders,

perhaps because of earlier skirmishes, although this is not known for certain. In the period 890892 the Normans were mainly active on the Seine, Oise, Somme and the upper reaches of the Scheldt in northern France. The Scheldt was almost certainly used as an access route for the fleet, although no records have survived. Records of one of their journeys overland, in April 891, include a detailed description of the fortress of St. Omaars. Besides describing the ramparts, the author also mentions that on the way to Lorraine, in the West Francia coastal area they marched past the castella ibi recens facta. These 'recently constructed fortifications' cannot possibly be one of the Zeeland fortresses because, with the exception of Oostburg, they were all in Lorraine. If we consider all the voyages made by the Vikings between 879 and 892 of which records have survived, the fleet must have passed the Scheldt estuary upstream and downstream, in both directions, at least four times. Since during that period the Vikings spent the entire year in the Scheldt valley, it must have been a great effort for them to gather enough food and horses for their cavalry. It is very probable that the chronicles make no mention of these forays, since no treasure would have been plundered or important places attacked - the very things that interest chroniclers. And this is where the islands of Zeeland come in. It is quite likely that the Zeelanders did not appreciate the regular visits by Vikings in search of provisions, and therefore built the Zeeland fortresses. The Bayeux Tapestry (ut cibum raperentur) shows clearly how the Vikings went about collecting food. The Zeeland fortresses: size and construction Structures that have been interpreted as the remains of early medieval fortresses have been found at five locations in Zeeland. There is historical and geographical evidence to suggest that a sixth fortress may have existed. The defences all have the same construction, with circular ramparts of varying diameters. In cross section, the outer ramparts range, in ascending order of size, from 144 meters in Oost-Souburg, 200 m in Burgh, 220 m in Middelburg and Oostburg and 265 m in Domburg. The possible fortress site near Kloetinge could contain a fortress with a diameter of some 200 m.

The inside and outside of the ramparts are made of regular piles of clay sods. The centre usually comprises unstructured sandy clay or sand. There may have been a palisade on top or just inside the ramparts. Only in Burg was such a palisade well documented during an excavation. However, the radiocarbon dating does not exclude the possibility that the palisade was constructed later, during the occupation of the fortress site. The total width of the rampart base varies from 4 to 12 m. Various phases of the ramparts are often visible, although it is not clear whether these building phases should be interpreted as localised repairs or as a pars pro toto of an entirely new fortress. It was possible to calculate that there was an interval of some 50 to 70 years between the two phases at Domburg. The excavated fragments of the Domburg fortress contained at least one, and possibly even two, building sacrifices . Archaeological evidence and later maps show that all the fortresses had a system of two paths that crossed at right angles in the centre. These paths lead to the passages through the ramparts, each of which lead to a narrow bridge over the moat. These bridges have been confirmed archaeologically by the discovery of the west and south passages in Oost-Souburg and by the northwestern passage in Middelburg. The structure of the ramparts of the Zeeland fortresses concurs largely with a description of the construction of similar defences built between 878 and 882 around the Church of St. Omaars. These ramparts consisted of fuste, gleba et cespite, or 'posts, earth and sods'. The earth and clay sods used for the construction of the Zeeland fortresses came from outside the fortress sites themselves. The method of construction was extremely efficient. The ramparts were built and a wide, shallow moat containing water was created at the same time. The archaeologically determined width of the moat varies from 30 to 50m. Dating of the construction of the fortresses It is essential that the Zeeland fortresses will be dated as accurately as possible, since as we have seen, the dating based on the castella ibi recens facta text (Miracula Sancti Bertini) can no longer be regarded as applicable to the situation in Zeeland. Thanks to the favourable conditions in the Zeeland coastal area, a great deal of organic

material, such as wood and animal bone, has been preserved. A dendrochronological dating of an oak plank from a coffin in the rampart of the fortress of Domburg has been obtained and the radioactive carbon content of twenty samples of organic material from the ramparts and the later occupation of the different fortresses has been determined. The results of the measurements must be calibrated to establish the precise age of the samples. In around AD 900 there was a deviation in natural radiocarbon levels in the atmosphere, and in consequence there are several possible absolute datings for a number of results. Nevertheless it proved possible to draw further conclusions from comparison between the datings and with the archaeological context of the samples. This indicated that all the fortresses were almost certainly built in the same period, in the last quarter of the ninth century. In those cases where there is information about the dating of the later occupation of the fortress sites, it appears that, with one exception, the occupation began in the early tenth century. The exception is the fortress at Domburg, which was largely covered with dune sand in the second half of that century. The fortresses in Zeeland- Lorraine were therefore probably built in the reign of Charles III the Stout or of King Arnulf of East Francia. We do not know to what extent they were constructed on the orders of the local power in the Zeeland coastal area. The fortress at Oostburg, on the West Frankish side, could have been built under the 'authority of Count Baudouin 11 (879-918) of Flanders. However, it would be more obvious to assume that the fortresses were built on the orders of the powers that be in the area, given the manpower needed for their construction. The function of the fortresses No occupation remains on the old ground level within the fortresses have been found. It is therefore assumed that they were built as vluchtburgen (refuge places) to offer a safe haven for the population and - more especially livestock when danger loomed. We have assumed that the population at the time of the construction of the fortresses consisted of shepherds, at least in the salt marsh area around Oostburg, OostSouburg and Middelburg. It is uncertain whether the fortresses ever did actually sustain a Norman attack. Apart from the

attack on Walcheren in 837 - when the known fortresses had not yet been constructed - a few vague reports have survived from the tenth century which refer to Walchenars who were defeated, and their land despoiled. At this time, too, the area was apparently still important enough to the Vikings for them to continue visiting. However, the tenth-century sources also make no mention of defences. Fortresses of this kind were not unknown outside the province of Zeeland, in the Netherlands and neighbouring countries in the Early Middle Ages. A revival in the construction of wood and earth defences in fact occurred throughout northwestern Europe from the eighth century. The defences were advanced for their time, and could well withstand attacks from men on foot or horseback, armed with bows and arrows. The reasons for their construction differed from place to place. Of course, the primary need for protection in times of unrest and strife played a role, but the construction of such defences might also have been connected with power politics among local rulers. Real refuge places for civilians and fortresses that were constructed around a pre-existing settlement or the king's or count's base both occur. Other defences can best be interpreted as strongholds, while buildings were added to others whose strictly geometrical pattern could point to military use. The Danish fortresses are an example of this. Historical data on the tenth century In the course of the tenth century the fortress settlement of Middelburg developed into the central settlement on the island of Walcheren eventually becoming the regional centre for the whole of Zeeland to the south-west of the Scheldt (Zeeland Bewestenschelde). During this century Zeeland Bewestenschelde was one parish, under the church at Middelburg later known as Westmonster. This church was probably established in the mid-tenth century. On Schouwen and the neighbouring island of Duiveland, too, there must have been one or two parishes. The abbeys owned a lot of land, the majority of which was used for pasturing sheep. For instance, in 949 Count Arnulf I of Flanders presented some of his land to St. Peter's Abbey in Ghent, including pasture for 120 sheep at Morena. This

place is near to, or was actually, the settlement immediately bordering the fortress at Oostburg. Emperor Otto 11 placed St. Baaf's Abbey in possession of pasture for 900 sheep on Schouwen in 976. The area must have relied for a large proportion of its income on the production of wool or worsted, known better in other written sources as pallia Fresconia, or Frisian worsted. In 972 Otto 11 gave the provintia UUalacra, Walcheren, to his wife Theophanu as a gift (morgengave). Forty years later, in 1012, King Henry 11 rewarded Count Baudouin IV of Flanders with Walcheren. Zeeland to the east of the Scheldt (Zeeland Beoostenschelde) came under the influence of the counts of Holland. In 985 King Otto III gave Dirk 11 everything he had previously had in fief in the villa Sunnimeri. The time of the great Norman invasions was over by the tenth century, although they probably visited the Scheldt estuary several more times. Of the fortresses, in the tenth century only Oostburg is mentioned in passing in 949 in connection with the above-mentioned gift from Arnulf I. Occupation within the fortresses Occupation has been established in all the fortresses. In the cases where it was possible to see signs of occupation, it was preceded by a raising of the ground level using clean sandy clay. Apparently the sites were suitable for occupation only once they had been raised to a minimum height above the water table. The water table was of course associated with the average high tide, since the land was still undiked. Systematic dike building did not begin until the mid-twelfth century. The reports of the huge storm surge of 1014, which according to a Ghent analyst cost the lives of many inhabitants of Zeeland, show that even the artificial raisings were probably not always adequate. It is difficult to establish by archaeological means the amount of time that elapsed between the construction of the fortresses and the moment at which the raising and first occupation commenced. The dating of the pottery is too inaccurate for this purpose. However, one can argue that the occupation would not have begun any more than a generation later. With the exception of Domburg, the occupation ceased in the last quarter of the tenth century. The large-scale investigation at Oost-Souburg has revealed details of the internal structure of the western half of this fortress settlement. In the first

occupation phase the houses were built on individual sandy clay platforms. These podia were raised along a system of axes that crossed at right angles in the centre of the site. The systematic raisings automatically created ditches along the system of paths which were covered with wooden trackways. These 'ditches' carried rainwater through a wooden drain or sewer via the passage through the rampart to the moat. Remains of the drain were found in Oost-Souburg, and in Middelburg the wooden sewer was found in situ in the north-western passage. It is not possible to determine which houses were in use at the same time. However, it is highly likely that the excavation reflects the building over the entire occupation period, lasting around 75 to 100 years. There would therefore have been some four to six buildings in use at the same time in the western half of the fortress site. The buildings constructed on the podia are all rectangular, but display little uniformity. Those with walls more than 1 m thick made out of clay sods are undoubtedly among the oldest. No doorways or inner walls were found, and there are no divisions that might indicate the presence of stalls. Large, rectangular pits were found in many buildings, however. These might have been used for storing sheep dung, and the rooms they are in may have been used as sheep stalls. Round hearths made of clay sods were found in most of the houses. Some of these hearths contained smithing slag and have therefore been interpreted as smithy furnaces. It can be shown, on the basis of the age at which pigs were slaughtered and the presence of bone skates, that people lived in the settlement in the autumn and winter at least. Given the fact that spindle whorls were found - wool spinning being traditionally women's work - and that horse breeding and the growing of arable crops such as barley, wheat and linseed have been established, it is likely that the fortress settlement was occupied all year round. The finds No finds are known from the time of the construction of the fortresses, with the exception of wooden posts from the ramparts at Burgh and Domburg, a vague posthole on top of the rampart of the Middelburg fortress and two irregular burials. The other find material can be linked with

the occupation of the inner sites, which almost immediately followed the construction of the fortress. Most of the material naturally comes from the large-scale investigation at OostSouburg. Since the find material from the other fortresses is virtually identical, only the OostSouburg material is discussed here. In terms of numbers, the most important category of finds from Oost-Souburg is the sherds of earthenware pots. Since very few tenth-century find assemblages have been discovered in the Netherlands, the pottery from Oost-Souburg is an important point of reference for the study of the development of form, distribution and dating of ceramic types in the transition from the early to High Middle Ages. Around half of the total of sherds are from coarse grey pottery. This is also known as Kugeltopfware, after the type of sagging base prevalent among this pottery. But not all the jars have this type of base. In fact, geographically Zeeland marks the transition from the actual Kugeltopfware of the north to the grey ware of the south. Although the origins of the 'local' pottery found at OostSouburg are unknown, it is assumed that it was produced nearby in the area. In the case of the Zeeland findspots, this might also mean that it came from production centres further up the Scheldt. The second important category of pottery is Pingsdorf ware, which comes from Pingsdorf in the German Rhineland. More than 40% of the sherds are of this type. In relation to the findspots outside Zeeland this is a very high proportion, which suggests that this high quality product was easily available, probably by means of the port of Middelburg. However, a more well-founded explanation for the high percentage of Pingsdorf ware must wait until further, urgently required, research has been carried out. The other types of pottery found in Oost-Souburg occur only in low percentages. They comprise, in descending order of abundance, white pottery with a yellow-green glaze, reliefband, Paffrath, red-fired, Hunneschans and Badorf ware. Braat's suggestion that the white pottery with yellowgreen glaze from Oost-Souburg represents an early group that pre-dates the Andenne production, has long been the subject of doubt.

A further study by F. Verhaege, a senior researcher for the National Fund for Academic Research in Belgium, has clearly shown that this category of pottery opens up interesting prospects for the study of the origins of glazed pottery in north-west Europe. The white pottery with yellow-green glaze found at Oost-Souburg probably originated in Hoei in the Belgian Meuse valley. The groups of reliefband, Hunneschans and Badorf ware are the youngest representatives of these categories of imported ceramics, which are so characteristic of the Early Middle Ages. The Paffrath ware, which generally dates from later, is one of the earliest occurrences. Not only a lot of the pottery was imported. All the stone implements - querns and whetstones and a bowl or bucket - came from outside the Scheldt region. The analysis of the type of stone shows that the querns came from the Eifel in Germany, while the whetstones come from Scandinavia. The iron, too - whether objects, semi-finished products or raw iron - must have been imported, as is shown by the slag that was studied by Ms C. Joosten as part of a larger study on the early medieval iron industry on the Ve1uwe. In a number of cases, the distribution of iron slags and further study of the slag material were used as a basis to show that some of the hearths found were used for smithing iron. This traditional activity will mainly have involved manufacturing or repairing the implements used in the settlement. These included simple knives, many of which are included in the find material. The category of bronze objects makes a fairly poor impression and consists of a series of simple fibulae. These items were almost certainly made elsewhere. A round bone fibula might well have been made in the settlement, and might serve to illustrate the difficulty of obtaining metal equivalents. The study of the objects of bone, antler and horn have shown that at least some of them, such as the combs and spindle whorls, came from outside the Scheldt region. This is suggested by the wide variety of types and the fact that the raw material, red deer antler, cannot have come from Walcheren. The salt marsh landscape is not suitable for red deer, and no red deer remains were found among the slaughtered animal remains within the settlement. If one looks at all the objects found from the tenth-century settlement in Oost-Souburg, one is

forced to conclude that many of these objects must come from elsewhere. Clearly, the settlement was dependent on trade or barter in order to obtain these goods. The most obvious place for the exchange of goods was of course Middelburg, which probably already had port facilities in the tenth century. Exported products will have included sheep's wool (possibly spun), salt, and perhaps horses. The opportunities for surplus agricultural production would not have been great in the brackish landscape, where there was a great risk of flooding. Imported products at any rate included quality pottery and querns from the Belgian Meuse valley and the Rhineland. Both areas were then part of the Holy Roman empire, to which Zeeland also belonged in this period. Historical data on the eleventh and twelfth centuries In 1012 King Henry 11 rewarded Count Baudouin IV (988-1035) of Flanders with the villa Walcras, thus adding Zeeland Bewestenschelde (the area between the Eastern- and Westernscheldt) to Flanders. The punitive expedition carried out by Robrecht de Vries, the brother of the count of Flanders in 1067 indicates that the inhabitants of Middelburg and Walcheren were not always obedient. The Count was defeated and the inhabitants of Walcheren subsequently became embroiled in skirmishes among themselves. We know this because abbot Thiofried of Echternach wrote of it in 1103 in his biography of Willibrord. Thiofried also described how he tried to settle the dispute, travelling to Middelburg by ship for this purpose. This is also the earliest indication that Middelburg had a harbour or roads. Until the end of the first quarter of the twelfth century Zeeland Bewestenschelde belonged to Flanders, after which the Count of Flanders presented the Count of Holland with the whole of Zeeland Bewestenschelde. In the eleventh century, many village settlements arose, almost all of them with their own church. Since the presence of a parish church was a precondition for an independent trade, the local lords played a major role in founding churches. This is indicated among other things by village names, which were composed of an individual's name with the suffix -kerke (church), such as 'sHeer-Arendskerke on ZuidBeveland. The fact that in the eleventh century Zeeland Bewestenschelde was owned by the Count of

Flanders and the Count of Holland alternately indicates that, in practice, the area found itself in a kind of power vacuum. Internal defences were therefore organized by the local lords, each of whom built a motte-and-bailey castle, generally a simple moated timber-earth defence. At one time there were over 135 such small timber-earth castles in Zeeland. Nowadays no more than 30 survive, which are known by the somewhat confusing name of vliedberg (refuge mound). The population also had to fight a constant battle against the water. Floods - those in 1014 and 1134 were recorded as particularly serious - led to the building of dams and dikes. Much of this work was done by the large Flemish abbeys. The written sources are not terribly informative when it comes to the old fortress sites in the eleventh and twelfth centuries. The following information is, however, available about the individual fortress settlements. Oostburg The mother church of the northeastern part of the countship of Flanders, the Church of St. Eligius, would appear to have stood directly to the south of the fortress site in 1038. It was a private church belonging to the count, which makes it likely that the fortress, too, belonged to the count. The centre of the settlement was probably already outside the fortress by 1038. Middelburg In 1123 the count placed the northern part of the fortress site at Middelburg at the disposal of the Abbey of Our Lady and in the mid-thirteenth century, the area directly to the west was given to a monastery of the Order of Friars Minor. The manifestations of the count's social and judicial power were then concentrated within the fortress. In addition, the fortress also functioned as a spiritual centre. The actual trade settlement would appear to have had two centres at that time: one to the south-west of the fortress around the Westmonster Church established there in the tenth century, which had a market function. And one immediately to the north-east of the fortress, centred around the Noordmonster Church, which had been founded at the end of the eleventh century, and which was the focus of port activities. Oost-Souburg The old parish church of the village of Souburg was a daughter church of the

Westmonster church in Middelburg. The Souburg parish covered a large part of south Walcheren. Part of this area probably originally belonged to the count. The church stood at some distance from the fortress in what would later (after 1250) become known as West-Souburg, possibly at the central manor house, on the counts estate. The parish was first mentioned in 1162, but the church probably dates from the first half of the eleventh century. The church at Oost-Souburg dates from 1250, when the parish of (West-) Souburg was split into an eastern and western section. Domburg When, in around the year 1000, the parishes of Oost- and Westkapelle were created as divisions of the parish of St. Maarten, which that time covered the whole of Walcheren, the fortress, which had partially disappeared under the sand, was used as the border. The site of the fortress was assigned to the territory of Oostkapelle. In the course of the eleventh century, a settlement called Dumburgh infra castellum, 'Dune fortress within the fortress' arose on the site of the fortress. The name suggests that the fortress must still have been partially visible. Subsequently, a settlement known as Dumburgh extra castellum, 'Dune fortress outside the fortress', was established directly to the west of the former fortress site, in the parish of Westkapelle. In both places village churches were created as divisions of the parishes of Oostkapelle and Westkapelle. The names Domburg binnen (within) and Domburg buiten (outside) the fortress occur for the first time in a charter establishing the abbey of Middelburg in 1235. Burgh Little mention is made of Burgh in the written sources, except that in 1219 a local lord by the name of Wolfert van Burgt was appointed. The parishes of Burgh and Haamstede were first mentioned in the second quarter of the thirteenth century. Both parishes are divisions of the mother church of western Schouwen; Haamstede is probably the eldest of the two. The mother church probably stood to the west of Westenschouwen, and is mentioned in a written source concerning possessions of the Lorsch Abbey in 776.

Kloetinge It is known for a fact that one of the oldest dikes on Zuid-Beveland, dating from the mid-twelfth century, bisects the presumed fortress site. The relatively low location of the site possibly explains why no settlement arose at the fortress site in the tenth century. This is why the natural character of the place where the settlement was founded eventually determined its name. The name Clotinge means something like 'place characterized by an elevation'. The village was first mentioned in the sources in 1216. Developments from the thirteenth century to the present day After the period when the towns and villages were established, the state of the fortress sites changed little. The fortress at Domburg, hidden by dune sand, was forgotten. The circular form of the fortress of Middelburg can still be seen within the street plan of Middelburg. The fortress sites at Oost-Souburg, Burgh (and Kloetinge) have continued to be used for agriculture. Finally, the uninhabited fortress site at Oostburg was incorporated into the construction of town defences in the early seventeenth century. In 1860 the round fortress at Oost-Souburg was still regarded as the site of a castle built against the Flemish invasions. The true nature of the fortresses has only been revealed by the interest of twentieth century archaeologists and historians. The investigations prompted by this interest have shown that many remains of the fortresses are still present in the soil, remains which have much to tell, as demonstrated by this book. The sites of these, more than a thousand year old defences, will in future enjoy the status of 'archaeological monuments'. A good example of the benefits of this new status was the reconstruction in 1994 of the fortress rampart at Oost-Souburg in the municipality of Vlissingen. With its double role of archaeological monument and local park, this site has taken the lead among the fortresses in terms of their role in the preservation of archaeological monuments and the cultural and historical life of the province of Zeeland.

2 Summaries of the material studies Medieval Coins from the Beach at Domburg On the beach at Domburg many early medieval coins have been picked up in the last few centuries. The finds include quite a few Merovingian and other gold pieces (tremisses) from the sixth and seventh centuries, but the bulk of the coins found there are Merovingian, Anglo Saxon and Frisian silver coins (denarii and sceattas) struck over a period of three-quarters of a century around the year 700. This large discrepancy need not come as a surprise, for the rate of production of the coins in the second group was much higher than that of the gold coins. The picture, however, changed in the Carolingian period. While the silver coins (denarii) continued to be struck on a large scale until far into the second half of the ninth century, there is a considerable decline in the incidence of finds on the Domburg beach: coins struck before 840 abound, whereas coins struck after that date are much more scarce. This seems to confirm the supposition that 'Domburg' was severely hit by the Viking attack of 837, because a similar break does not occur in the general pattern of finds in the Netherlands around that date.

It appears that the plant species could not have grown together, but must originate from very different vegetations. Plants which grow on the highest parts of the salt marshes are particularly well represented in the sludge. They indicate the presence of extensive natural grasslands suitable for grazing. The pollen analysis also shows a very open vegetation. The animals might have been kept in the settlement within the fortress during the night, so that dung passed into the sewer from the stables. The plant remains may also derive from bedding, hay or sods originating from the high salt marshes. Plants from ruderal habitats may have grown in the fortress itself on dung heaps, roadsides, waste ground, neglected gardens, etc. Barley, wheat and linseed/flax were grown on the highest parts of the salt marshes. Whether or not oats were also cultivated is not clear. Rye may have been imported. The seeds of arable weeds, together with some chaff remains, may derive from the waste from crop processing activities. These weeds may also have grown in vegetable gardens in the fortress. The environment was very suitable for animal husbandry, but the archaeobotanical investigation of the sludge from the sewer shows that arable farming was also practised.

Plant Remains from a Sewer of the Fortress at Middelburg

The Early Glazed Pottery found in OostSouburg

Janneke Buurman

Frans Verhaeghe

During the excavation of the early medieval fortress at the Balans in Middelburg the remains of a wooden sewer containing undisturbed sludge were found. This sludge has been dated to the last phase of use of the sewer in the first half of the tenth century. The sludge was subjected to an archaeobotanical investigation to allow a reconstruction of the vegetation and environmental conditions in and near the fortress and to provide an insight into the plant component of the food. A sample from the sludge was sieved using meshes of 1.0, 0.5 and 0.25 mm. Seeds and fruits appeared to be very well preserved in an uncarbonized condition. Some carbonized plant remains were also present. Pollen analysis was also carried out on two small samples from the sludge. ...

This contribution discusses the early white-fired pottery with yellow lead glaze which was associated with the settlement within the late Carolingian circular fortress at Oost-Souburg and dates to the period 900-975. The small, but nevertheless indicative, amount of these wares represents 4.6% of the total sherd assemblage and may well include two separate fabric groups, both tempered with quartz-like inclusions: a finegrained fabric with a smooth surface, reminiscent of the typical Andenne-type wares and a slightly coarser fabric with a more pimply surface. The finds comprise a coherent group of specific types of tableware: medium-sized to fairly large, ovoidshaped spouted pitchers with small strap-handles on the shoulder and an reverted rim which often

Arent Pol

has a lid-seat (at least six and possibly twelve examples), conical lids (four examples) and small globular-shaped drinking vessels (at least seven and possibly twelve examples). Many of the vessels are decorated with block or diamond rouletting or with applied strips (some with thumb impressions); only a few are decorated with bosses or incised lines. Cooking pots seem to be totally absent from the assemblage, but the fragmentary state of the finds makes definite conclusions in this respect difficult. Technical and typological features indicate that the finds are closely comparable to others from Alkmaar, Antwerp, Bruges, Douai, London and elsewhere, a fact which, in addition, confirms their chronology. The origins of such features are, however, more difficult to identify. A brief survey of the main finds in France, England, Belgium and the Netherlands makes it possible to emphasize that white-fired glazed wares occurred fairly regularly from the late ninth century onwards, having been produced in many regions. Two possible sources are identified for the OostSouburg and related finds: north-western France (including the Paris area) and the Belgian Meuse valley. A reassessment of some of the earlier finds suggests the Meuse valley as the more likely provenance and kilns may quite possibly have been located in or near the town of Hoei. The evidence re-establishes the credibility of W.E. Braat's somewhat intuitive hypothesis for the existence of an early, pre-Andenne group of glazed wares, characterized by a more lavish use of glaze. The interpretation of the finds in terms of socioeconomic patterns remains difficult. These quality, or even luxury, items do not necessarily indicate the presence of an elite, though they do point to a certain degree of well-being. Together with the fairly large amount of other imports found at Oost-Souburg, mainly Rhenish redpainted wares, they also suggest that the site had close links with the new urban network of international exchange which emerged after the Viking invasions and after the demise of the major emporia such as Dorestad. Many questions related to the production and socio-economic significance of these wares remain unanswered pending further research (including science-based approaches). Similarly, more detailed comparative work on other early

finds of this type is needed, particularly in Flanders and in the Netherlands, including attempts at ranking the relevant sites. Together with related finds, the Oost-Souburg examples of early glazed wares do point to tenth-century processes of innovation and behavioural patterns which were far more complex than was previously generally believed. They also show the dangers of linking white-fired, glazed sherds intuitively to the well-known Andenne-type products and of using them indiscriminately as reliable chronological guide-fossils for the late eleventh and twelfth centuries.

Slag and Iron-working in Oost-Souburg Ineke Joosten

Excavations in the early medieval fortress at Oost-Souburg produced 16 kg of slag and 1 kg of furnace lining which was associated with the settlement from the tenth century. Macroscopical and chemical analyses of the slag material showed that it originated exclusively from smithing activities and not from iron production. A perforated clay disc, fired on the outside and vitrified on the inside, was also found. This had been used as a bellows protector. In order to investigate the provenance of the iron, a metallographic study of some of the iron objects from the site was undertaken. The slag inclusions in the iron objects were known to carry the chemical characteristics of the ore from which the iron was produced. Analysis of these inclusions with a micro-probe indicated high phosphorus and low manganese contents. Such a chemical composition is thought to be the 'fingerprint' of bog iron ore.

The Stone Objects from Oost-Souburg Henk Kars

The stone object assemblage excavated at OostSouburg (900-975) comprises fragments of fifteen querns, around 30 whetstones and one fragment of a bowl or bucket. The querns, which are typologically comparable to early medieval querns (as described earlier), are all made of tephrite deriving from the Mayen quarries in the Eifel region of Germany. Amongst the whetstones, one rotary whetstone of new red sandstone (Buntsandstein) was identified. This

sandstone type also derives from the Eifel region, most probably from a quarry south of Aachen. The other whetstones are bar-shaped and mostly heavily used; three of them have a hole bored through them. Most of these whetstones are of quartz phyllite. Whereas one may have come from south-eastern England, most are probably from quarries in Telemark, southern Norway. The relatively high frequency of good quality whetstones at Oost-Souburg is linked to forging activities. The most intriguing find is the fragment of soapstone that must have been part of a large bowl or a bucket. Although soapstone finds are rare in the Netherlands, they are more common in areas with natural deposits of soapstone, such as southern Scandinavia and the Alpine regions. On the basis of a comparison between the material from Oost-Souburg and that from the early medieval trading site of Haithabu near Schleswig in Germany, it is concluded that the Oost-Souburg specimen originally derived from either southern Norway or Sweden.

wares in the fortress. In addition, it is possible that combs were acquired via the market. There is no evidence to suggest that the handled combs from Oost-Souburg were imported from Friesland or England, as has been put forward for handled combs found elsewhere.

Objects of Bone, Antler and Horn from Oost-Souburg

The discovery of production waste means the three-pronged objects were certainly made within the settlement. The function of these objects is unclear. Since the inhabitants of OostSouburg lived, in part, from the production of wool, one could speculate that these comb-like objects had something to do with its processing. The bone skates from Oost-Souburg were all actually used as skates rather than sledge runners. They were made from various bones (the radius, metacarpus and metatarsus of horse and cattle) and can be categorized into different types. Some types were made to be used without the need for securing ties. Other types were clearly used with ties. Yet others were possibly only tied on at the heel so that they couldn't slip away. Another possibility for this latter type is that they were also used without ties and the naturally occurring or drilled hole under the heel merely allowed the skates to be bound together with a cord. These coarsely-made objects in which, for example, the holes were made with a knife or another sharp tool and not with a drill, were clearly not professionally made but rather home produced. The same is true of the coarsely-finished spikes. These are probably spikes from skating sticks which were used for pushing off while skating. A few sawn-off horn-cores indicate the use of horn.

Roel C.C.M. Lauwerier

The 129 objects found in the fortress of OostSouburg, mainly combs, needles, needleshaped objects, spindle whorls, three-pronged objects, skates, and spikes from skating sticks, give a picture of the rich variety of object types and decoration current in the tenth century. All the combs are composite types and, except for one, are all made of antler. They can be roughly divided into four types with very diverse patterns of decoration. The fact that red deer was not indigenous on the island of Walcheren, in combination with the presence of semi-finished comb tooth plates and waste from antler working, leads to the conclusion that antler was imported as raw material for the on-site production or repair of combs. This does not exclude the possibility that finished combs could also have been imported from elsewhere. The very variety in comb types found, however, makes it unlikely that a specialised craftsman worked within the fortress. The variability, the half-finished state of some objects and the waste material can be better explained by assuming that the combs are the products of itinerant craftsman who sold their

As with the combs, the antler spindle whorls, or at least the raw material for their production, must have been imported from elsewhere. The varying dimensions and weights of the spindle whorls points to the spinning of threads of different thicknesses. The absence of loom weights, in a period when the technique of knitting was not yet known in north-western Europe, suggests that the inhabitants of OostSouburg probably traded wool in its spun form. The thirteen needles found are all made from the long bones of mammals. As well as needles used for stitching together very coarse material, or those used in the making of beehives or for mending fishing nets as eyeless needle shaped objects have been found. These were possibly used as hairpins, styli, or utensils for either needlework or leatherworking.

Distribution and Function of Bone Threepronged Objects Herman W. van Klaveren

During the excavations at the site of the early medieval fortress of Oost-Souburg ten specimens of worked animal bone were found, which belong to a single, three-pronged, artefacts. Similar artefacts are known from parts of the Netherlands, Belgium, France and Germany. Their name and function are unknown. The number of specimens found is steadily increasing. The most common material used in their production is a cattle metapodium; other bones do however occur. The form is tapering and resembles a 'D' in cross-section. It carries from two to four, usually three, teeth on top; the rounded side is mostly decorated in a simple way. Wear is clearly visible, but only on the outside. An enquiry was made into the limited available data concerning their distribution, context, dating and possible function. Dating falls roughly between the late ninth and the twelfth century or possibly even somewhat later. These artefacts can occur in any kind of settlement of this period and area. Their name and function could not be deduced from archaeological records. Written or pictorial evidence or parallels from other periods in bone or any other material are not known. They might have served some purpose in the working of perishable materials such as leather, wool, flax and the like. Stock breeding in Oost-Souburg Roel C.G.M. Lauwerier

The information in tables 23 and 24 gives a survey of the bone material found in the tenth century circular fortress of Oost-Souburg. No sieved samples were taken which means that little information is available pertaining to fowling and fishing activities. The remains of red deer are all parts of antler and are mostly found in the form of

worked tools. The meat of red deer was not eaten by the inhabitants of the site and the animals do not appear to have been hunted since they were probably not indigenous to Walcheren. The worked antler objects, or the raw material, would therefore have been imported from elsewhere. Cats and dogs were also not eaten by the inhabitants. The bones indicate great variability in the size and appearance of the dogs. The biggest dog, perhaps a watchdog or sheepdog, had a height of 65-68 cm. The smallest dog, with a height of only 23-25 cm, had legs bent like a dachshund and was possibly used for hunting foxes. The horses, with a height of about 138 cm, were rather small. As well as being used for riding and as draught animals, the horses were possibly bred for the market. Horse meat was not eaten, but the bones were used for making bone skates. The cattle, with a mean height of 115 cm, were mainly kept for their meat, but possibly also for milk. A number of sheep were clearly slaughtered young for their meat. The greater part of the flock, however, was kept primarily for wool and only consumed after some years. The sheep were probably not milked. Goats were kept for this purpose on a small scale, as is indicated in the bone assemblage from the circular fortress of Middelburg, dating from the first half of the tenth century. For purposes of consumption, beef was far more important than mutton: expressed in weight, probably three or four times more important. The contribution of pork to the diet was markedly small. The overall picture given by the bone assemblage for the inhabitants of OostSouburg is, in the first place, not that of a group of shepherds, but much more that of stockbreeders. It is concluded that there must have been a surplus of animal products. Horses were possibly bred for the market and the inhabitants sold, amongst other things, wool, including spun wool, indicated by the presence of spindle whorls and the absence of loom weights on the site.