STUDIA HISTORICA GANDENSIA

UIT DE SEMINARIES VOOR GESCHIEDENIS V AN DE RIJKSUNIVERSITEIT TE GENT

217

A. VERHULST

An Aspect of Continuity between Antiquity and Middle Ages: the origin of Flemish cities

Overdruk !fit: Journal of Medieval History, vol. 3,1977, pp. 175-206

GENT 1977 STUDIA HISTORICA GANDENSIA is een ;eeks overdrukken gepubliceerd door de Afdeling Geschiedenis van de Faculteit der Letteren en Wijsbegeerte van de Rijksuniversiteit te Gent. Blandijnberg, 2, Gent (Beigie)

STUDIA HISTORICA GANDENSIA est une serie de tirages a part publiee par la Section d'Histoire de la Faculte de Philosophie et Lettres de l'Universite de Gand.

2. Blandijnberg, Gand (Belgique)

STUDIA HISTORICA GANDENSIA is a series of offprints published by the Department of History of the Arts Faculty of the University of Ghent. 2, Blandijnberg, Ghent (Belgium)

STUDIA HISTORICA GANDENSIA

ist eine Reihe von Sonderdrucken, veroffentlicht von der Historischen Abteilung der Philosophischen Fakultat der Staatsuniversitat Gent. Blandijnberg 2, Gent (Belgien)

D. 1977/053817 as a creation oj the Carolingian period and afler. An aspect of the But recent archaeological excavations have shown a substantial Roman presence at six sites which later were towns oj medium or considerable im­ question of portance. Although the buildings were evidently • • abandoned in the .fiflh century, settlement and political and ecclesiastical organization developed contInuIty around them in the Merovingian age. The Roman background thus had a considerable effect on the between development of town life in medieval Flanders .·~ In a brief article entitled Les villes jlamandes antiquity and avant Ie XIIe siecle, Henri Pirenne in 1905 provided an exceptionally clear exposition of his views on the origin and earliest history middle ages: of the Flemish cities (Pirenne 1905). He did not modify his theories significantly there­ after (Pirenne 1910, 1925, 1927). Subse­ the origin of the quent Belgian historians who dealt with this problem, particularly the historians of the Flemish cities Ghent school- Van Werveke, Ganshof, Ver­ cauteren, Dhondt, and others - remained faithful to the essential lines of Pirenne's between the concepts until the 1950s (Van Werveke 1950). In 1958 the German historian Petri tested North Sea and the validity of Pirenne' s theories critically in the light of the numerous studies of in­ dividual towns of the Low Countries and the the ScheIdt neighbouring regions of northern France

I am indebted to Mr M. Ryckaert, aspirant of Adriaan Verhulst the Nationaal Fonds Wetenschappelijk Onderzoek, who helped me particularly with the sketch-maps; to Mr H. Thoen and Mr F. Verhaeghe, assistants in the Department of Archaeology of Ghent University (director: Professor Dr 5.]. De Laet) for information; to Professor Dr D. M. Nicholas (University of Neb­ raska) for the translation of the original Dutc h text of Most historians oj the Flemish cities have argued this paper into English; to Professor Dr R. Vaughan that there was no significant Roman habitation for the final revision of the text; and to Mr W. De Schamphelaere for his technical help in the execution north of the civitates oJBelgica secunda. Urban of the sketch-maps. To all of them I expl:ess my development in Flanders was thus seen essentially gratitude.

Journal of Medieval History 3 (197 7): 175-206. © North-Holland Publishing Company 175

S ';q 79, ~.. ------~ which had appeared during the previous two completely free of Roman inAuences or decades (Petri 1958a). Among other things, antecedents, although they were in an area his synthesis demonstrated that Pirenne's which had belonged to the Roman Empire view of the origin of the medieval cities of for quite some time. Their origin as trading the Low Countries had to be nuanced and centers in the Carolingian period, according corrected, particularly for the towns of the to Pirenne, was due entirely to long distance southern Low Countries, situated in mod­ trade. Their continuation or revival after the ern northern France, and for the towns of the Viking invasions of the second half of the Meuse region, and that Pirenne also had to ninth century was made possible only by the be supplemented with new information protection afforded by the new military fort­ about early medieval commercial settle­ resses which the counts of Flanders had ments in the delta area of the great rivers. erected in their immediate neighbourhood. But Petri basically confirmed Pirenne's I have noted that a number of recent theories for the great Flemish cities of the individual studies, some by historians but Scheldt basin and along the North Sea coast, most by archaeologists, have contradicted specifically Ghent and Bruges. Vercaut­ Pirenne's essentially simple views on several eren's bibliographical overview of the cities important points. Since most of them have of medieval Europe, published in 1967, appeared since Petri wrote in 1958, their reached the same conclusion (Vercauteren results have not yet been incorporated into a 1967). The extremely detailed and analyti­ synthesis. I wish to do this here, so that two cal bibliographical survey which Nicholas important aspects of the new views can be published in 1969 on the origin of the medi­ tested against the theory considered valid eval cities of northwestern Europe (Nicholas since Pirenne: the extent and intensity of 1969) confirmed Petri's findings on a more urban life between the North Sea and the general European scale. But numerous ScheIdt in the Roman period and its import­ detailed studies of the Flemish towns which ance for early medieval urban development; had appeared since Petri wrote led Nicho­ and the traces and forms of urban life in this las to the conclusion that the by now classic area during the early Middle Ages up to the theories of Pirenne on these cities could no time of the Viking invasions. longer be maintained without serious qualification. Pirenne proceeded on the assumption Urban life in the Roman Empire that in Flanders proper, between the North Sea and the Scheidt, every form of urban life Archaeological investigations over the past and organization had been annihilated dur­ fifteen to twenty years have considerably ing the third and fourth centuries. He broadened our understanding of urban life thought that in contrast to a few civitates in between the North Sea and the ScheIdt the southern part of this region, which during the Roman period. Not only began to revive as urban centres from the Pirenne's ideas, but also the more recent early Merovingian age, the great Flemish expositions of Petri and even of De Laet cities, particularly Bruges and Ghent, were (1960) on this point, must be thoroughly

176 r-­ \, r-- " ...... _\ t p. . ..'

'" TH ... NET NOS ~ '" "- ~ o II \ . Aa rd;'~~'~rg ...... , ...J['~ ,. II \.. \ \ "". /' --- r '- .. -.- .., .. w.l .. • ...... , .. " Bruges _ •... ,' •.J·~~ ...... C \_.. Antwerp . Oudenburg

B E L G u M N

~~\

'; o 5 10 15 20 25 km ( I I

" Figure 1. Cities in Flanders with Roman origins. revised and supplemented. Together with burg over Poperinge and Bruges, which did new studies of material which had lain not bisect the vieus noted above, for it ran untouched in museums and collections for farther south, passing through the territory many years, the most recent excavations of the medieval town from west to east. The have emphasized the persistence of settlement may have had the character of a comparatively numerous vici from the statio along this route. Its relation to the vicus second half of the first century A.D. to the farther north is still uncertain (Figure 2). second half of the third. If we confine Of cardinal importance is the fact that ourselves strictly to the North Sea-Scheidt while the vicus north of the town was area, we find Roman vici at Courtrai, Harel­ abandoned as a result of Germanic attacks beke, Wervik, Oudenburg, Wenduine, from the sea between 268 and 270, and Bruges, Aardenburg, Ghent and Antwerp. shortly thereafter destroyed and flooded by Some of these survived the Germanic in­ creeks during the Dunkirk-II transgression, vasions of the third quarter of the third the settlement along the southern road was century as castella (Raepsaet 1975). What not flooded because it was located on higher strikes us is the appearance in this list of sandy ground, so that although it too was places along the North Sea coast and the devastated in 268-70, subsequent settle­ Scheldt which merit our special attention in ment there remained possible. Finds from a discussion of the traces of urban life the late third and fourth centuries prove that during the early middle ages (Figure 1). this indeed happened. For this reason, Thoen thinks it not inconceivable that the A still unpublished thesis of H. Thoen, fourth-century settlement in the city of directed by Professor S. J. De Laet at the Bruges acquired a military character, .just as University of Ghent (Thoen 1973), shows happened with other places in Flanders that one of three mercantile vici discovered which survived the invasions of the late third to date along the North Sea coast was at century. Its center may have been a castellum, Bruges, in the dock area just north of the as at Oudenburg, Aardenburg, and perhaps modern city. The two others were at Ouden­ also Ghent and Antwerp. Thoen thinks that burg and Wenduine. The vieus of Bruges this fortification should be sought on the site originated rather late, around 200 A.D., of the later medieval castle at Bruges, which along an old creek created by the Dunkirk-I was crossed by the Roman road between transgression, where there had been settle­ Cassel and Aardenburg. Systematic excava­ ment since the Iron Age. This favourable tions have not yet been conducted along the location for commerce probably determined southern side of this route, in the open its mercantile character. There must also ground in the middle of the castle in front of have been significant Roman habitation in the modern city hall, but isolated Roman the old medieval centre of Bruges even from remains have been found there (De Vliegher the beginning of the imperial age, but its 1965). The site, on the edge of the exact location and character cannot yet be Pleistocene land mass along the then coast determined. It was probably along the and not far from a creek made by the Roman road leading from Cassel to Aarden- Dunkirk-II transgression, which in its turn • ~/~ • 2 • J • " N 5 6 I • ~

---­'- ..... ------

BRUGES

Figure 2. Bruges: 1 . Roman vicus; 2. Roman cemetery; 3. Roman finds; 4. Roman coin find; 5. Limitofthe Dun­ kirk-II transgression; 6. Medieval wall enclosing o 500m lhe city since 1297; A. Il=.-==--==II Medieval market place; B. Medieval cas/rum . was connected to the open sea, was quite at Bruges will correspond to the later square similar to that of the castellum of Ouden­ form of the medieval castle there. We have burg, which I shall discuss shortly. no certain knowledge of the latter structure But the square form and dimensions of before the mid-eleventh century (Dhondt the castellum at Oudenburg should not lead 1957a; Verhulst 1960), and it is very difficult us to assume that the plan of a Roman to reconstruct the topographical develop­ castellum which may eventually be discovered ment of the castle before that time. We do

179 know that the late Carolingian pre-Roman­ axis Zuidzandstraat-Steenstraat or Oudc­ esque church of St Donatian, which was burg-Hoogstraat-Langestraat, would have probably built in the first half of the tenth pennitted habitation at Bruges in the fourth century as a fortress church on the northern century only south of this road, on the side of the east-west Roman road which Pleistocene sand ridge, presumably on the crossed the area of the later fortification, was location of the later comital castle and/or built on extremely marshy terrain in a thick somewhat west of it. peat stratum on a foundation of transver­ Although the military character of the late sally positioned beech logs (Mertens 1955a, Roman settlement at Bruges remains some­ 1955b). The comital castle at that time thus what uncertain, no such doubt is possible very probably did not yet encompass the for Oudenburg, sixteen kilometres west of terrain north of the Roman road, but rather Bruges (Figure 3). The excavations of Pro­ the land south of it on the Pleistocene sand fessor J. Mertens between 1956 and 1970 ridge, where there was perhaps not enough revealed the existence here of a civilian vicus room even in the first half of the tenth and a castel/urn. The former was on the century to build the count's church on the western extremity and the latter on the open plain that would constitute a central eastern edge of the east-west Pleistocene and essential part of the castle in later sand ridge, which was the site of the Roman centuries (Verhulst 1960 :41-4). This road linking Oudenburg and Bruges and topographical situation virtually excludes which had a rather dense Roman settlement the possibility that the marshy fenland north (Mertens 1962; Creus 1975). The vicus, of the road, where a systematic excavation of which was extremely important, can be the pre-Romanesque church of St Donatian attested from the early imperial period and has revealed no Roman finds at all, could it was flourishing in the second century and have been part of a Roman castel/urn. The the first half of the third, but it was probably latter, if indeed it existed at all, can only be devastated shortly after the middle of the sought south of the Roman road which was third century. on the northernmost extremity of the A castellurn was constructed at about that Pleistocene sand ridge, perhaps in the time somewhat farther east, involving an southern half of the later comital castle, earthen wall with moat and a wooden perhaps farther west in the adjoining area of palisade. It was revamped and elevated the Oudeburg (Koch 1962 :59-64; Ryckaert shortly thereafter, at the end of the third or 1973: 165-7). beginning of the fourth century, to avoid the Until systematic excavation solves the rising water level of the Dunkirk-II trans­ problem of a castel/urn, we can only say with gression. As early as the second half of the certainty that the inundations of the fourth century, however, il was replaced by a Dunkirk-II transgression in the northern larger castel/urn (I63 x 146 metres) surroun­ and northwestern part of the area of the ded by a turreted stone wall, which was medieval city, together with the marshy fen connected with the establishment of the so­ character of the terrain on the northern edge called litus Saxonicurn on both the British and of the Roman road formed by the west-east continental shores of the North Sea during

180 00 1

+":+1+ 2 ...... /~ 3 OUDENBURG /

;:

N

o 500m

Figure 3. Oudenburg: 1. Limit of the Pleistocene sand ridge; 2. Roman cemetery; 3. Roman settlement; C. Roman castelLum . the fourth century. This was the Roman existed here, on grounds of the burg suffix in defense line of which the Notitia dignitatum the toponym and the later medieval plan of gives us information, and Mertens thinks the city. The excavations begun between that Oudenburg was the chief stronghold on 1961 and 1963 by Trimpe Burger have the Belgian North Sea coast. shown conclusively that Aardenburg was the The construction technique of the north­ most important settlement in the area of the ern wall, which faced the sea, also shows that great river mouths from about 100 A.D., the rising sea level may have been a and especially between 170 and 273 (Trimpe consideration in the replacement of the Burger 1971:51-2; 1973:141-4; Raepsaet original wooden eastellum at Oudenburg 1975:130) (Figure 4). It included numerous with one of stone. The originally favour­ large buildings of Tournai stone, and during able location of a eastellum on a Pleistocene its great age between 170 and 273 had an promontory in the flooded coastal area, essentially military character. The castellum, connected to the open sea by a wide channel whose existence has been demonstrated con­ and to the interior by a Roman road, was clusively by very recent and as yet unpub­ probably destroyed gradually by the effects lished work, seems to have been even larger of the Dunkirk-II transgression. The sudden than that of Oudenburg. But although it was abandonment of the eastellum in the first still inhabited during the fourth century, it years of the fifth century can be explained by seems to have been comparatively unimpor­ a combination of military factors connected tant then; for the settlement had been with the disorders in northern Gaul between destroyed in or about 273 by fire and shortly 406 and 410 and the Dunkirk-II flooding of thereafter flooded in large part by the the area east of the eastellum. This inundation Dunkirk-II transgression. In this, it forms a cut off the military base from its hinterland striking contrast to Oudenburg. and it consequently lost its importance. l We must also note briefly that both earlier This topographical isolation of Oudenburg and more recent investigations have attested is in contrast to the late Roman settlement at several Roman settlements in the delta area Bruges, where the Dunkirk-II inundations (Figure 5). They were commercial depots, probably did not reach the Roman road military fortifications, or cult centres (Trimpe along which the habitation was con­ Burger 1971, 1973). One of the most im­ centrated. It will be an important element in portant was the emporium which has since any evaluation of my argument concerning been submerged by the sea on the coast the importance of the two places during the at Domburg. In addition, a sanctuary of seventh and eighth centuries. Nehallenia has recently been discovered Aardenburg, seventeen kilometres east of on the coast of North Beveland northwest Bruges, is also on a ridge of sandy of Colij nsplaat. Farther north, on the islands Pleistocene topsoil, on the edge of the coast­ of Schouwen, Goeree-Overflakkee and line which had been flooded since about Voorne around the mouth of the Meuse, 300, and on the Roman road leading from archaeologists have identified a sanctuary, a Bruges (De Vries 1968). Earlier scholars vieus, and perhaps a castellum at Ouddorp, presumed that a Roman fortification had and a castellum which may have included a

182 N

.1 .. 2 o 100 200m AARDENBURG

Figure 4-. Aardenburg: A. West Gate or Kaaipoort; B. St Bavo Church; C. Town hall; D. Site of the former OL Vrouw Church; E. Modern gas works ; F. Original course of the River Ee. Excavations: I. Mixed Roman and medieval finds. 2. Medieval finds. commercial settlement at Oostvoorne. This Roman remains were revealed by the ex­ entire area must have been of considerable cavations of Professor A. Van de Walle commercial importance (it was on the route along the Scheidt at Antwerp in the vicinity from England to the Rhineland) at least of the so-called 'Steen' within the later until about 270 A.D., when military events medieval semi-circular fortification (Van de (the naval invasions) and probably also the Walle 1960, 1961). Interpretation of these rising sea level resulting from the Dunkirk­ finds was hindered, however, by the fact that II flooding ended its activity. The Roman they were not found in direct connection presence after that date, which had mainly a with Roman settlement traces in situ, and by military character, can only be attested the lack of a comprehensive report on the farther south, particularly at Bruges and excavations. Only the finds themselves, dat­ Oudenburg, but also at Aardenburg. ing from about 140 A.D. until the second The situation at Antwerp must also be half of the third century, have been considered, particularly in view of the sig­ interpreted somewhat more thoroughly nificance of the area of the mouth of the (Vandenborn 1965). Although they did not Scheidt and the Meuse in the Roman period. directly prove the existence of a settlement,

183 .. • 2 o 3 t 4

? o

N. BRABANT

.'~'" , " ...... I ..... ;...... " ..... ':':":'.'... ~ '\., ...... /' BE LGIUM ;:·,·····t ~.~ . r.J·: ·~'~ :~·~ .. ~······,···· ·······,···1 ... ,.• ,., ...... ; ...... ~ .... . Antwerp t Balgerhoeke \, ...... ~ .:j-... /. o 20km

1::::1======i l

Figure 5. Delta area, Roman finds: 1. Sanctuary; 2. View; 3. Castel/urn (only in the case of Aardenburg is the existence of a Roman castellum positively proved); 4. Cemetery. the fact that they include tiles, roofing the butchers' hall (Oost 1976) (Figure 6). material, and the like already suggested it. This suggests that the Roman settlement at Genuine direct traces of Roman settlement Antwerp was comparatively large, but we were found for the first time in 1975, just still need further information concerning its outside the wall of the fortification beside essential character. There have been no finds

184 ' igur 6. AllIwerp: I. '

185 to date from the fourth century. For this which the name Gentbrugge and the reason too it is impossible for the time being designation Ganda for the site of the later St to use archaeological evidence to hypo­ Bavo's Abbey both suggest was probably thesize a Roman fortification at Antwerp, named Ganda, existed from the mid-first although several pieces of written evidence century A.D. into the fourth. But the in­ from the eighth and ninth centuries which I vasions of the late third century cost the will discuss later might suggest this. eastern elements of the vicus most of its Such a hypothesis may not, however, be industrial and commercial character, with excluded farther upstream along the Scheidt the result that the core of the settlement, at Ghent, where it has been shown that a which was to be primarily military from that Roman castellum almost certainly existed. time on, was probably located thereafter in The systematic excavations of Professor S. J. the westernmost nucleus on the location of De Laet and his colleagues at the eastern end the medieval St Bavo's Abbey. We may safely of the modern city area of Ghent, about 1:\0\'0 assume that a late Roman castellum existed kilometres southeast of the medieval St here, in which the St Bavo's Abbey was built Bavo's Abbey on the border of the suburb of in the second quarter of the seventh century. St Amandsberg and the village of Destel­ I t was still being called castrum Gandavum in bergen, bet:\-veen the road to Dendermonde the first half of the ninth century (Verhulst and the course of the Scheidt, have revealed 1972a, 1972b). But archaeological evidence an industrial settlement, as well as one of the cannot yet prove directly that a castellum largest cemeteries discovered to date in existed here in the fourth century, although northern Gaul (De Laet 1969). Both were numerous random Roman finds have been undoubtedly part of a widespread vicus found on the site. The deduction rests on which, just as was the case with many other historical arguments, a toponymic con­ Gallo-Roman vici in these regions, must sideration, and several eighth- and ninth­ have had a very diffuse geographical struc­ century narrative sources which suggest that ture. The settlement and cemetery formed the remains of the castel/um were still visible the eastern rim of the vicus at that point, so in the Merovingian and Carolingian that the vicus most probably extended from periods. But the eastern sections of the vicus the site of the excavations westward along had been abandoned and had disappeared the Dendermondse Steenweg to the site of as a real settlement. They had a purely the medieval St Bavo's Abbey, at the con­ agrarian character in the early middle ages. fluence of the Lys and the Scheldt, where We conclude this summary with Courtrai. numerous Roman finds have already been Located in the far interior, at the junction of discovered. This vicus was reached by a the Roman roads Tongres-Boulogne and Roman road which archaeological evidence Tournai-Courtrai, this place is linked to the enables us to locate for a considerable territory of the civitates along the southern distance northward from its junction at edge of the area bet:\-veen the North Sea and Blicquy. It reached the south bank of the the Scheidt which we are presently consider­ Scheidt at Gentbrugge, just across from the ing. The name Cortoriacum and the location vieus on the north bank (Figure 7). The vicus, on a Roman road have meant that scholars

186 <- 00 1 III 2 ~3

ROOSBROEKEN -

DESTELBERGEN

DENDEO

HET STEENLANT CASTELBANT - ...... b q, -<: ~ (: <.0 ~-1 CI)

J a 500 1000 m .. I E

Figure 7. Ghent: 1. Marshes; 2. Presumed site of the Roman vicus; 3. Excavations by S. J. De Laet; A. Roman industrial area; B. Roman cemetery; C. Presumed site of the Roman settlement; D. St Bavo's Abbey, formerly castrum Gandavum and Roman castellum; E. Roman road Bavai-Ganda; F. Zandberg; G. St Bavo's Cathedral, formerly StJans Church; H. Gravensteen (the Count's Castle); J. St Peter's Abbey on the Blandinium hill. r

Kuurne

,

N

(".. ., .,/ '?; ..... ,.J ----- A

~ /////,0 B

a 1000 m 1====-_-== =-.....'

Figure 8. Courtrai: 1. Roman cemetery near the Moiemlraal; 2. Graves at the Zan db erg (Harelbeke); 3. Roman finds at Harelbeke; 4. Roman finds at Harelbeke--Stasegem; A. Presumed Roman road; B. Area with many Roman finds. have never seriously questioned the Roman much new archaeological evidence to light character of Courtrai. Several excavations which emphasizes the considerable impor­ during the past twenty years have brought tance of the place in the Roman period,

188 although the diggings were unfortunately Oudenburg, Aardenburg, Antwerp, Ghent unsystematic and disconnected (Raepsaet and Courtrai, practically all of which have 1975:100, 164-5). Although they have not only come to light during the intense yet been formulated into a synthesis, we archaeological investigations of the past fif­ should note them briefly, for these Roman teen or twenty years. They have substantially antecedents can explain the role which supplemented and modified our picture of Courtrai played as an administrative centre urban life in the far northwest of Roman in the Merovingian and Carolingian Gaul. Our previous knowledge was largely periods, .iust as we shall see was the case for confined to the civitates of Cambrai, Tour­ Bruges, Antwerp and Ghent. nai, Arras, Therouanne and Boulogne on The site of Courtrai and neighbouring the southern border of the region between Harelbeke contained two important vici or the North Sea and the Scheidt, in the zone twO segments of a single very extensive and between Bavai and the coast where Roman loosely structured vicus. At least part of it, if and German had their points of contact not the entire agglomeration, had important (Vercauteren 1934). Now we find that com­ industrial activity involving ironworking mercial industrial, and military settlements from the mid-first century A.D. until the which were appreciably different from great invasions of the late third. A sophis­ civitates and yet had an urban character ticated complex of trenches has also been existed north of this, with a striking con­ excavated from the mid-first century, as well centration along the coast, adjoining or in as remains of wooden and stone buildings. the area of the mouth of the Scheidt, and To date it is uncertain whether or not these along the Scheidt into the interior. In should be interpreted as remains of a contrast to the walled civitates, these were military encampment connected with the open vici with a very diffuse geographical preparation of Claudius' invasion of Britain. structure, consisting of several nuclei of Finally, a cemetery dating from the first and settlement at some distance from one second centuries has also been investigated another, some of which had a decidedly (Figure 8). No trace has yet been found at commercial or industrial character, while Courtrai of fourth-century remains, but the others were military or semi-agrarian. From reference in the Notitia dignitatum to a the mid-third century, one of the nuclei was Roman garrison there suggests that it was frequently a castellum, whose fortifications still inhabited in the fourth century and had made it increasingly important because of a military character (Dhondt 1948: 133). The the invasions of the second half of the third entire Roman settlement at Courtrai is at century. least as important as those at the other In contrast to the civitates, these vici owed places we have considered. The fragmentary their origin not to a function in the Roman yet unmistakable indications of life at administrative system, but to commerce on Courtrai in the Merovingian and Carolin­ the North Sea and the great rivers, par­ gian periods must be interpreted in the light ticularly between Britain and the Rhineland, of this. and to certain forms of local industry such These new data concerning Bruges, as iron extraction, pottery, and salt produc-

1~9 tion. In terms of these functions, practically also contrasted strongly the rise of such new all of them ceased to exist around 270. But coastal harbours along the English Channel whenever the settlement had included a and the North Sea coast as Quentovic and military element, generally a castellum, it Domburg with the decline of such Roman always survived the invasions of about 270 centres as Boulogne and Oudenburg. and continued to exist, in some cases until As Petri noted (I958b), this viewpoint is the beginning of the fifth century. This of perhaps too schematic and inadequately course does not exclude the possibility that nuanced, and leaves no role for elements of random civilian settlements tied in some way continuity with the Roman period. On the to the castella may have been repopulated contrary, for Dhondt as for Pirenne, the during the fourth century, but they cannot determining factor par excellence in establish­ have been of much practical importance in ing the first signs of urban life in the early any sense, and certainly not for commerce. middle ages was long-distance trade over­ seas and along waterways. But Dhondt in my opinion paid too little attention to the bare Traces and forms rif early medieval and quite striking fact of topographical urban life continuity with the Roman period, which I think is important for every place which he The problem of continuity from late Roman considered (Dhondt 1957b:60-Il, with the to early medieval urban life between the provisional exception of Quentovic (Dhondt North Sea and the Scheldt must be reviewed 1962:196-7). Dhondt furthermore did not in the light of the new material presented deal with Flanders proper between the above. In particular, we must consider the Scheldt and the North Sea except for the extent to which the combination of the new civitates in the extreme south,2 for he felt that archaeological data with historical material no traces of urban life could be detected long since available may alter the thesis here in the Merovingian age in terms of dealing with this problem which J. Dhondt trade or industry, nor indeed by any other formulated in 1957 (I957b). Dhondt noted criterion. Since he knew of no evidence of the blossoming during the Merovingian Roman urban life in this region save at period of commercial settlements at Dinant, Oudenburg and Courtrai, continuity as a Namur, Huy and Maastricht in the Meuse problem no more occurred to him than it valley. Save Maastricht, which had been had to Pirenne. But now, in the light of the more important, all of them had been new archaeological data concerning the merely small forts in the Roman period. Roman past of Bruges, Oudenburg, Dhondt drew a striking contrast between Aardenburg, Antwerp, Ghent and Courtrai, these places and the commercial and in­ the fragmentary information concerning dustrial decline of the old Roman civitates of these places in the Merovingian and early Cambrai, Tournai, Arras and Therouanne, Carolingian periods, to which Dhondt and the fate of which he attributed to the decline others assigned no particular significance, of the land routes, on which these civitates must be re-examined. Indeed Petri noted in were located, in favour of waterways. He 1958 the possible importance of such a

190

I • study, on the basis of the first Roman finds marshlands which were only brought into at Antwerp as they were then becoming use as pasture in the course of the eighth and known (Petri 1958a:24l). ninth centuries (Gysseling 1950; Noter­ We therefore return to our discussion of daeme 1958). When the name ofOudenburg the places along the Flemish coast: Ouden­ appears for the first time in a medieval text, burg, Bruges and Aardenburg. Because of in 866, the place was a cattle pasture and had their location on the edge of the Dunkirk-II no urban character at all. The name transgression from roughly 300 to 700 A.D. Aldenborg which was then used for it means (Verhulst 1959, 1962-3), any evaluation of 'old fortification' (Gysseling 1950 :48, 62). the importance of each of these sites in the Although the Roman castellum still existed early middle ages must take account of the and was visible - and would be until the end local consequences of these Roods for of the eleventh century (Gysseling 1950 :53- habitation and general commercial location. 8) - this name shows furthermore that the As we have seen, the castellum at Ouden­ earlier name of the Roman settlement had burg suffered directly at the end of the third been lost. The place had been depopulated century from the rise of the sea level. Even and was considered an abandoned ruin by the area east of the fortification was the newly arrived German-speaking inhabi­ Aooded, and the castellum was consequently tants. built higher (see above, p. 182). The areas The author of the Vita Eligii, writing in the directly south of the castellum, Groeninge and first quarter of the eighth century, listed the Riedinne, were three metres below sea level various urbes or rnunicipia appertaining to the and were still wet and marshy in the late diocese of Eligius, who had been chosen middle ages (Gysseling 1950:81) (Figure 3). bishop of Noyon-Tournai (Krusch The exposed position of the Pleistocene 1902 :695); but the pastoral character of sand ridge after the floods practically iso­ Oudenburg at that time makes it highly lated Oudenburg from its hinterland, a fact unlikely that he meant Oudenburg by his which would have had a deleterious effect on vaguely described and loosely identified its development as an eventual centre of the rnunicipium Flandrense. 3 This curiously in­ area. Thus it is hardly surprising that the direct description requires explanation. We cella which was erected at the beginning of get the clear impression, by the author's lhe eighth century in the vicinity of Ouden­ mention of all these places with the adjec­ burg and which was given in 745 to the tival use of a place name beside the indepen­ abbey of St Bertin at Saint-Omer was built dent noun urbs or rnunicipiurn, that the farth~r inland at Roksem, a few kilometres adjective has principal reference to the south of Oudenburg. The church of Roksem circumscription of which the author con­ was the mother parish for the more recent sidered the rnunicipium, which is not given its parish of Oudenburg itself, oj which only own proper name, the chief place.4 Three of the part farthest south, in the vicinity of the the Flemish places with which we are con­ castellurn, was above sea-level, forming a cerned - Tournai, Ghent and Courtrai - were small strip of sandy ground. The rest of the capitals of pagi which bore the town's name: parish of Oudenburg consisted of Aooded pagus Tornacensis, pagus Gandensis, pagus Cor-

191

BIDL. UNIV. l G J-~ J'1 '1' turiacensis (Ganshof 1949: 26 8). Bu t this is not agrarian agglomeration which was either true of the pagus Flandrensis, which was fortified or contained a fortification, or was named from an area and not from an of Roman origin - which at that time administrative centre. The use of the ex­ amounted to the same thing (Vcrcauteren pression municipium Flandrense in the same 1934 :350-1; Nienneyer 1960: 710-1). context can hardly be explained other than Only two places in the territory of the on the assumption that the author of the pagus Flandrensis, which extended from the Vita Eligii intended by analogy to indicate Yser on the west to just east of Bruges and the chief place of the then existing pagus consisted, outside the flooded coastal area, Flandrensis. This capital thus actually existed only of a small strip of sandy ground to the at the beginning of the eighth century, but it south (Dhondt and Gysseling 1948), merited certainly did not give its name to the pagus the term urbs or municipium at the beginning Flandrensis, as Tournai, Ghent and Courtrai of the eighth century: Oudenburg and did to their respective pagi. We may deduce Bruges. Most scholars have heretofore from this fact that the capital in question was assumed that Oudenburg was meant by the either somewhat less important than the phrase municipium Flandrense, since they others or was of more recent date as a major knew that it was of Roman origin (Gysseling centre. In this connection it is not without 1950 :68-71), while nothing of the sort had significance to note that the author of the been ascertained for Bruges. In view both of Vita Eligii neglected to name two other pagi, what we have noted above about early the pagus Mempiscus and the pagus Rodan­ medieval Oudenburg, and of my in­ ensis, which were constituent parts of Eligius' terpretation of the list of Flemish municipia diocese and which are mentioned in other in the Vita Eligii as capitals of pagi with a contemporary sources (Ganshof 1949:268). somewhat urban character and of Roman Neither of these pagi had an identifiable origin, we must identify Bruges as the muni­ settlement with a non-agrarian or only cipium Flandrense mentioned in this source. semi-urban character which could have Bruges was the capital of the pagus Flandren­ constituted a capital at that time and might sis from the second half of the ninth century have given its name to the pagus. 5 (Dhondt 1942:77-8), and I see no good The author of the Vita Eligii thus seems to reason why this could not have been the case have been limiting his remarks to the most a century and a half earlier, for it is now prominent centres of the bishopric of Tour­ absolutely certain that Bruges originated as nai which had a somewhat urban character a Roman settlement. and were simultaneously capitals of a pagus. If this conclusion is accurate, the problem The municipium Flandrense is hard to identify of the origin of medieval Bruges must be because of the author's obscure ter­ posed in a radically different light. Bruges minology, but it would fit this description has always been regarded up to now as a even if it were less important or of more completely new settlement owing its origin, recent origin than Tournai, Ghent and which could be dated no earlier than the Courtrai. But the author was undoubtedly beginning of the ninth century, to overseas using urbs or municipium to mean a non- trade made possible by its favourable

192 location. It was at the .iunction of an 1973). We have this information for Bruges overland route with a creek of the Dunkirk­ only at the beginning of the eleventh cen­ II transgression which provided a link to the tury, when the place was first called portus sea (Figure 2). A fortification provided the (Ganshof 1938 :282). settlement with the necessary protection a ur only piece of indirect evidence from from the third quarter of the ninth century an earlier period which suggests the role (Ganshof 1938; Dhondt 1957a; Verhulst which Bruges would eventually play in 1960; Koch 1962; Ryckaert 1972). It was international commerce is the derivation of uncertain whether the commercial settle­ the name of the city from the Old Norse ment originated during or before the first bryggja, which means landing bridge or quay. half of the ninth century, thereby predating Since the name is first recorded shortly after the erection of the fortress, or whether it the middle of the ninth century, it must have developed only after the castle had been been borrowed during the first half of that built in its shadow in the second half of century at the latest. It may have occurred by the century. But our new insights suggest on confusion with an original local name the contrary that Bruges was a centre with derived from the little river Reie which flows some urban characteristics which func­ from the hinterland of Bruges into the creek tioned as the capital of a pagus by the (Gysseling 1971). The meaning of the new beginning of the eighth century at the latest. name suggests that the place functioned as a It owed its position to its topographical harbour, so that it may have been borrowed continuity with an eventually fortified from Old Norse as a result of commercial Roman settlement which had been able to contacts with Scandinavia. We have no proof survive despite the floods of the Dunkirk-II of this, but Louis the Pious' grant to St transgression. Various considerations, Anskar of a small abbey at Torhout, twenty however, suggest that it can hardly have kilometres southwest ofBruges, as a training been very significant in the early eighth school for missionaries who were being sent century, among them the fact that it ·had still to Scandinavia, suggests something of this not given its name to the pagus Flandrensis of kind (Verhulst 1960:60). which it was the capital. The presence of a mint at Bruges, attested We are no more certain than we were around or shortly after the middle of the before concerning the time when this centre, ninth century (Verhulst 1960:59), says little favoured by its location at the edge of a of the commercial importance of the place. creek which crossed a road and led to the The area of circulation of the Bruges coins, sea, became a centre for commercial activity which could tell us something more, was not transcending mere regional importance. very extensive until about 900. Bruges coins The first reference to Bruges as a vieus dates of the Charles the Bald type have been only from about 900 (Koch 1949), but even recovered at Assebroek, Zelzate, Domburg, this does not necessarily imply a mercantile Glizyand Cuerdale (Morrison 1967:188). settlement, for vicus can be interpreted to The Bruges mint was thus essentially regio­ mean this only when other contemporary nal. Numerous Bruges coins found in the information confirms it (Petri 1958a; Kobler Baltic region suggest that it only assumed

193 more international overtones at the begin­ completely clear. 6 On the coast of Wal­ ning of the eleventh century (Ganshof cheren at Domburg, not far north of Bruges, 1938 :284). another commercial centre played a very It is extremely difficult to draw a con­ important role in continental trade with clusion from this sparse and indirect England until roughly the mid-ninth evidence. Provisionally, it seems unlikely century, when it disappeared (jankuhn that Bruges owed its origin and earliest 1958 :464-72). Coin hoards prove that it had prosperity to international trade, although ties with Bruges (De Man 1936:6; J ankuhn we cannot exclude the possibility that it had 1958 :471), and it seems probable that commercial significance as early as the first Bruges gradually assumed the role of this half of the ninth century. The infrequent place (Koch 1970:317-18, 321-2). On the direct evidence from written sources before other hand, a number of urban centres, 900 about the nature or importance of the particularly Ghent and Tournai, became settlement at Bruges all suggests that it was a important for trade in the interior along the political, military, administrative, and also Scheidt during the second quarter of the eventually ecclesiastical centre (Ganshof ninth century, as did one palatium, Valen­ 1938:281-2; Dhondt 1942:77-80; Noter­ ciennes (Petri 1958a:239, 248-9; Koch daeme 1954; Dhondt 1957a:4, 15; Koch 1970 :319). Their commercial activity during 1962 :34-6). The role of Bruges in politics the second and third quarters of the ninth and government continued to be the prin­ century was, I think, also a factor in the cipal element in the importance of the city evolution of Bruges into a trading centre. during the tenth century (Verhulst 1960:63). Just as was the case with Oudenburg and We still cannot prove even for that time that Bruges, the history of Aardenburg (Figure 4) Bruges owed its rise primarily to inter­ after the Roman period shows the impor­ national trade, and I think that this con­ tance for the eventual survival or disap­ clusion is even more valid for the ninth pearance of a place of the local conse­ century. We must beware of forcing all large quences of the floods of the Dunkirk-II Flemish cities into the same pattern and transgression from the end of the third deriving their origin from a single theory, as century. When the sea burst across the scholars since Pirenne have done. I do not Pleistocene sand ridge on which Roman think that Bruges originated as a Carolin­ Aardenburg was situated, the major part of gian trading centre beside which a castle was the settlement was flooded. Only the highest subsequently erected in the second half of point, the site of the castellum, was spared. the ninth century which would endow the The ruins of this fortress remained isolated place with its continuity and economic in a landscape which was uninhabitable prosperity as a city. Bruges was much older until the beginning of the eighth century. than this as a non-agrarian settlement, but From that time until well into the ninth on the other hand it only became an century, we have reference only to sheep important centre of international commerce meadows in this vicinity which were ex­ much later. The reasons for this slow and ploited from artificially constructed mounds secondary commercial development are not (De Vries 1968 :233-4, 244). The area was

194 known as pagus Rodanensis from that time on. in the ninth century, the fOlmer in a written The name was derived from the Rudanna, source, the latter on a coin, as we shall see. the little stream on which Roman Aarden­ Civitas was the translation of the Germanic burg was located and which had probably burg (Kobler 1973), and this qualification been widened into a creek by the floods. As I certainly fits Aardenburg. emphasized above, this pagus had no Ninth-century Aardenburg was thus in all population centre of an urban character, so probability a fortification rather than an that neither such a place nor the pagus itself agrarian settlement, but I hesitate to attri­ was included in the early eighth-century bute any urban function to it, and certainly summary of the Vita Eligii, discussed above none of a commercial nature. It only (De Vries 1968 :233-4). became a secondary town with trading The earliest surviving use of the name importance much later. Its existence as a Aardenburg may be in a barely legible non-agrarian centre from the ninth century inventory of the properties of St Bavo's is due entirely to its topographical continu­ Abbey at Ghent, dating from 810-11: apud ity with the Roman period. That it did not Rudburg . .. bercarias . .. (Verhulst 1971 :226). become a nucleus for urban development in This form is related to the old versions of the the ninth century, like Oudenburg but in toponym which are known from 966: Roden­ contrast to Bruges, can certainly be ex­ burg (De Vries 1968 :234). It is a Gennanic plained largely by the fact that it was unin­ word meaning fortification or burg on the habitable from the late Roman period until Rudanml. I t is striking that the inventory the beginning of the eighth century. Aarden­ does not call it a villa. Several archaeological burg and Oudenburg thus can constitute a finds confirm the fact that it was inhabited in demonstration a contrario of the accuracy of the ninth century (De Vries 1968 :233), but it my hypothesis concerning the origin of could have been only minimally important. medieval Bruges. It was probably limited to the ruins of the The history of early medieval Antwerp Roman caslellum, which may have been remains for various reasons a difficult and rebuilt during the ninth century as a defen­ complex problem which cannot be resolved sive outpost against the Viking danger (De satisfactorily within the limits of this study. Vries 1968 :235-6). Maurice Prou thought Although written evidence {i'om the seventh that a coin of Charles the Bald (840-75) with through the ninth centuries is somewhat the inscription Rotanis civitas was minted at more plentiful than for the three places Aardenburg rather than Rauen (Prou considered thus far, its authenticity and 1892:28). This identification is by no means interpretation pose such problems that certain (Dhondt 1948: 137; De Vries earlier scholars have not been able to reach 1968 :234), but the term civitas does merit definitive conclusions on the basis of it. 7 Nor attention, for it then meant a fortification, has any scholarly attempt been made to generally of Roman origin. Other places investigate critically the appreciable but which had been Roman and probably also fragmentary information about the earliest COntained ruins of a fortification, such as history of Antwerp and work it into a Antwerp and Courtrai, were called civitates coherent synthesis since Professor Van de

1'l:i ANTWERP

N

F

Figure 9. Medieval Antwerp: 1. Caslrum area; 2. First suburbium, enclosed by the ruien (eleventh century); 3. Ne\ (juarter ncar the O.L. Vrouw Church, enclosed about 1200 by the veslen; A. Former SI WalbU1ga Church; B. Stee. !castle); C. Nieuw Vleeshuis (Butchers' Hall); D. Town hall. E. O.L. Vrouw ChUl"ch; F. St Michael's Abbey.

I 'Hi Walle's excavations in the medieval castle daeme and Dekkers 1955: 146-7), a church along the scheldt revealed extremely in­ dedicated to SS Peter and Paul was in fact teresting additional archaeological data established at Antwerp after lhe mid-seventh from the Gallo-Roman, Carolingian and century. post-Carolingian periods (Van de Walle Solving the problem of the exact location 1960, 1961). This archaeological infor­ of the fortification within which the charter mation indeed is in some respects difficult to of 726 places the church would require a reconcile with the written sources. 8 lengthy digression which we shall omit here. Despite a hiatus in the archaeological Its existence however is confirmed by the evidence within and in the immediate reference Normanni Andwerpam civitatem neighbourhood of the tenth- and eleventh­ incendunt in the Annales Fuldenses for the year century castle between the end of the third 836 (Kurze 1891 :27). This source uses civitas century and the ninth (Van de Walle 1960 : 13; to mean a fortification of Roman origin 1961 :136),9 it is still possible to assume constituting the administrative and/or on the basis of the evidence of the written ecclesiastical centre of a wider area (Bonen­ sources that the nucleus of the early fant 1953 :422; Schlesinger 1954; Kobler medieval settlement at Antwerp was in this 1973). Since it seems improbable on the area rather than around the twelfth-century other hand that the civitas of 836 was a abbey of St Michael, nearly a kilometre recent fortification (Vercauteren 1936), we south of the later castle, a solution suggested may identify it with the castrum of 726, which by several authors including P. Bonenfant in its tum must be considered of older (1953:421-2) (see Figures 6 and 9). We origin. No doubt it originated in a late proceed on the basis of a charter of 726, Roman fortress, just as did pra."ctically all whose authenticity was demonstrated by Merovingian castra (Vercauteren 1936). Ganshof (1962:314) and Bonenfant (1953: Although archaeologists have not proven 420-1), in which a certain Rauchingus that a late Roman castellum existed at and his wife Bebelina gave to St Willi­ Antwerp, the topographical connection and brord the church of SS Peter and Paul, even the continuity of the Merovingian which according to the charter had been castrum with the now definitely attested built by St Amand infra castrum Antwerpis Gallo-Roman agglomeration near the area (Wampach 1930:78-80). Although the Vita of the later fortification seems very prob­ Amandi does not mention the foundation of able. A possible Merovingian settlement at this church (Krusch 1910), this omission can Antwerp was thus probably concentrated in be explained by analogy with other foun­ the neighbourhood of a Roman fortifica­ dations of this missionary which his bio­ tion in which a church was built in the graphy does not enumerate (Verhulst seventh century. Both church and fortifica­ .v 1953 :42-4). We may therefore assume that tion were probably burned by the North­ I! after the first unsuccessful missionary efforts men in 836. A new castle was built along the of Eligius (de Moreau 1947 :92; Noterdaeme Scheidt at the end of the ninth or beginning and Dekkers 1955: 141-9) and Amand (de of the tenth century, and a church was Moreau 1942 :26-7, 1947 :88-9; N oter- constructed within it shortly afterward

197

J3JBL. uNIV. GJllNT which was dedicated to St Walburga (Van de second and third quarters of the ninth Walle 1960, 1961; Van Werveke 1965 :46- century (Koch 1970). 7). 10 The expansion of Antwerp only re­ What else do we know of the nature and commenced about 900. A hint of this is the importance of this place between the seventh use of the term vicus in reference to it at that and tenth centuries? Coins were struck there time (Bonenfant 1953 :423), the construction in the seventh and eighth centuries, and this of a new fortification and of a new civilian happened more often than not in or beside quarter within the walls (Van de Walle, a Merovingian fortification (Bonenfant 1960, 1961), as well as the reappearance of 1953 :423). But minting does not necessarily coins under King Henry I the Fowler (919- indicate commercial activity, concerning 36) and the use of the term civitas (Bon­ which we have no other references, for the enfant 1953 :424). We still cannot say mention of a toll in the charter of 726 is whether this expansion was connected with probably an eleventh-century interpolation international trade, or that it resulted from (Ganshof 1962:314). No coins are known to the essentially administrative and military have been struck at Antwerp during the role still played by the castle of Antwerp, Carolingian period. But the area of the later which was on the border of the Empire in castle was still inhabited without marked the late tenth and early eleventh centuries hiatus in the ninth century, despite the anni­ (Dhondt 1952:8, 14; Van Werveke 1965:46; hilation of the earlier fortification by the Van Acker 1975 :35-8). Of course, such an N orthmen." The structure of the ninth­ administrative and military function is very century houses which have been excavated, probable for the seventh, eighth and early particularly their arrangement in a row ninth centuries, but its precise content is still along a single street and the correspond­ a mystery. Thus for example the bishopric to ingly advanced stage of division of the land which Antwerp appertained and the secular into lots, suggests that this settlement had an circumscription of which it was a part are urban character (Van de Walle 1960, 1961), still uncertain well into the eleventh century which may be attributed to a commercial (Dhondt 1952:8, 13-15). Further investi­ function of the town,12 but it can be ex­ gation must be made of the functions of plained equally well by its administrative Rauchingus, who was obviously a very dis­ and military role. The importance of this tinguished Frankish patens vir with power urban agglomeration in one or other of over the castrum at the beginning of the these aspects must have been affected by the eighth century (Ganshof 1958 :22), as well as invasion of the Northmen in 836. Antwerp of the role of the unknown monastery of shared this fate with the Carolingian em­ which the church of SS Peter and Paul was a poria at Domburg, Witla and Dorestad, the dependency before Rauchingus gave it to St prosperity of which also declined for the Willibrord (VanAcker 1975:15). same reason around 840, in contrast to the Although many aspects of the earliest places upstream along the Scheldt - Ghent, history of Antwerp thus remain unclear, this Tournai and Valenciennes - which were just preliminary investigation has in any case assuming a commercial character during the once again confirmed, as it did for Bruges,

198 the significance of Roman and Merovingian caused him to withdraw into the ruins of the antecedents for the establishment of the Roman castellum a few hundred metres settlement and later devdopment of an downstream, where a few years later, by 639 important medieval city. We see again, too, at the latest, he founded a church with a that the role of such a non-agrarian popu­ cloister. It was originally dedicated to St lation centre in military, monetary, eccle" Peter, but it was called St Bavo's Abbey from siastical and perhaps also administrative the beginning of the ninth century. Some affairs was primarily in the pre-Carolingian decades later, between about 650 and 675, period and preceded the eventual commer­ several of Amand's disciples founded a little cial importance of the place. cella dedicated to SS Peter and Paul on the A re-examination of the written evidence hill called Blandinium considerably farther for the history of early medieval Ghent south along the Scheidt. Its excentric which I undertook in 1972 in the light of the location was to prevent it from becoming excavations of Professor De Laet in and near important in the history of Ghent until the Roman Ganda and of the attractive hypo­ beginning of the tenth century, after St thesis which the toponymist Gysseling had Bavo's Abbey had been destroyed by the elaborated from toponymic and topo­ N orthmen. But the fate of the abbey of graphical elements (Verhulst 1972a, 1972b), Ganda, the later St Bavo's, was quite dif­ has thoroughly altered our knowledge of the ferent. It was at the confluence of Lys and history of Merovingian and Carolingian Scheldt and was being called castrum Gand­ Ghent. My new concepts can be summar­ avum by the beginning of the ninth century ized as follows. because of the fortified castellum within On a slight elevation along the left bank which it had been built. The abbey evident­ of the Scheidt at its confluence with the Lys, ly developed close ecclesiastical ties with the the ruins of a Roman castellum still existed at settlement upstream on the Zandberg in the the beginning of the seventh century. The course of the seventh and eighth centuries, castellum and its immediate environs were since it was functioning as the cult centre of uninhabited. The Roman vicus east of it had the Zandberg settlement in the ninth. At the disappeared and been replaced by several beginning of the eighth century the Vita small agrarian settlements of Frankish Eligii noted a municipium as the chief place of origin farther north. An overwhelmingly or a pagus Gandensis, as did the roughly contem­ even exclusively agrarian settlement must porary Vita Amandi. As we have seen, such have been located in the seventh century a language meant a settlement with urban few hundred metres upstream along the left character, fortified and as such considered at bank of the Scheidt before it joined the Lys, the time to be of Roman origin. The author at the crest of a hill on which the modern St very probably meant by this the complex Bavo's Cathedral is located; part of this formed by the Roman castellum and St Bavo's elevation is still called the Zandberg (Figure Abbey which had been built within it, which 7). St Amand tried in vain around 630 to was the part best known to him. But this convert the inhabitants of this settlement to does not necessarily mean that the ad­ Christianity. Their hostility toward him ministrative centre of the pagus was also

199 ..

located there. An examination from later tions. St Bavo's Abbey was burned by the sources of the juridical situation in the ninth Northmen in 851, but this occasioned only a century of the settlement on the Zandberg, brief caesura in the life of the abbey. several hundred metres upstream on the However, in 879 the foundation was chosen Scheldt, has shown that it probably still did and organized by the Vikings as a winter not belong to the jurisdiction of either of the encampment, and this time it meant the end two abbeys of Ghent. It is thus not incon­ of the monastic community, which was only ceivable that it may have been under the reinstalled several decades later, during the jurisdIction of the king or his representa­ second quarter of the tenth century. The tive, the count of the district of Ghent, in the nearby portus probably suffered as little per­ ninth century, and that the administrative manent damage from the first Viking raid in centre of the pagus was situated there as early 851 as the abbey, since the Martyrologium as the beginning of the eighth century. Usuardi mentions the portus around 865. Be that as it may, the vicus which the Vita Shortly after the second Viking occupation, Bavonis situates along the ScheIdt in the however, a completely new commercial second quarter of the ninth century, not far settlement originated along the Lys, far to from St Bavo's Abbey, and the portus of the west, in the immediate neighbourhood Ganda that is mentioned around 865 in the of a fortification constructed during the last Martyrologium Usuardi, both refer not to the quarter of the ninth century by the count or location of St Bavo's Abbey, but to the the lay abbot of St Bavo's to defend the settlement several hundred metres up­ region against the Northmen. This develop­ stream from the confluence of Lys and ment is characteristic in two respects: the Scheldt, on the Zandberg in the vicinity of the new commercial settlement originated when later St Bavo's Cathedral. Thus this settle­ St Bavo's Abbey no longer existed, and it ment must have developed a commercial developed at some distance from it, this time function along the Scheldt during the in the shadow of a new fortress and clearly second and third quarters of the ninth connected to it. century in addition to its eventual ad­ The earliest Carolingian mercantile settle­ ministrative function as capital of the dis­ ment on the Zandberg, however, was prob­ trict, which may be older but cannot be ably abandoned during the Viking invasion proven so. The proximity of the rich St of 879, and only revived in the first hall' or Bavo's Abbey was not unconnected with the tenth century. This time it was free of any this, as is shown by the fact that the fair connection with the earlier but not yet which was still being held around 1000 in rebuilt St Bavo' s Abbey but was probably in­ the portus at the festival of St Bavo had fluenced by the second nucleus of the portus probably originated in the ninth century in which had developed in the interim along the settlement on the Zandberg. the Lys. The two would grow toward each This connection with the abbey of St Bavo other during the tenth century. Thus, - the term 'dependence upon' would surely although the first commercial settlement at be too strong in the case of Ghent - is fur­ Ghent clearly originated in connection with ther confirmed by the following considera- the increase of trade along the Scheldt in the

200 ninth century, it cannot be considered apart an administrative centre in the Carolingian from the administrative or ecclesiastical period. The a e r C urll'ai reinforces my centre, in this case an important abbey, thesis that the adminisu-alive fa tor was pre­ whose location was ultimately determined eminent in h . continuari n an urban by topographical continuity with the Roman centre of Roman origin into the early period, just as was true of Bruges and middle ages. Antwerp. The considerable importance of Roman In conclusion, we see that Oudenburg and Courtrai seems to have continued into the Aardenburg, which were rendered unin­ early middle ages. Just like Ghent, Tournai, habitable by floods until the ninth century and (as I think I have demonstrated) Bruges, and which were conspicuous for centuries as Courtrai was called rnunicipiurn by the Vita abandoned Roman castella towering over an Eligii at the beginning of the eighth century. inhospitable region, were exceptional cases. We may therefore assume that it was an Bruges, Antwerp, Ghent and Courtrai, the agglomeration of urban type, in which for­ other places which we have considered here, tifications, walls, or buildings were still had also preserved their importance into the present, which had originated in the Roman late Roman period, for the most part as for­ period. Perhaps for this reason too the place tified centres. But sources from the seventh was the centre of the pagus Corturiacensis, or early eighth centuries show that these whose existence may be assumed on four were not merely more densely settled grounds of my interpretation of this text than the purely rural places in their en­ given above. A mint still functioned at virons, but also had an importance which Courtrai during the Carolingian period, transcended the immediate locality, mainly although this is less certain for the Mero­ as fortified administrative centres for a vingian age (Dhondt 1948: 136). The coins larger area. This fact as well as their struck there bear the reference civitas, a geographical location can only be explained qualification which says less of the impor­ by their Roman origin. Continuity with the tance of the place than of its fortified Roman period must be viewed primarily or character and its Roman origin. Finally, the exclusively as topographical, but in some choice of Courtrai as a winter encampment cases it probably involved the continuation by the Northmen in 880 (D'Haenens of the subdivision of a Roman civitas within 1967 :48) should be considered in connec­ the circumscription of the pagus (Ganshof tion with these facts; it may also help to 1949:268). We thus find that these places explain why we hear nothing else about had a distinctly urban character in the Courtrai for roughly a century. Merovingian and early Carolingian periods, Yet Courtrai, .iust like Aardenburg, would just as did the larger civitates farther south: n nClhcle achi \Ie a erwin imp nan c a Cambrai, Tournai, Arras and Therouanne. a ciry r es ' -ntia ll t ondal- rank mu(h There are no more indications than for the I( te .. : in Ih Iv drlh . I1lul'. Even m l' civitates of the eventual commercial impor­ striking i: th r I whkh il, and nearly al l tance of any of these centres on a more than til Iher pI ces w > have studi '(\, a um d a local scale before the second or third quar-

201 ter of the ninth century. Only then did trary to Pirenne's assumption, played an emporia develop, at least at Ghent and active economic role in themselves and in perhaps also at Bruges and Antwerp, beside their relations with their neighbourhoods. or within what we may consider pre-urban They can no longer be considered mere nuclei of Merovingian and Roman origin, passive consumers (Van Werveke and Ver­ just as happened at Cambrai, Tournai, and hulst 1960; Joris 1972). If the role of inter­ Arras at about the same time. These Carol­ national trade was less dominant than the ingian trading settlements in Flanders no active role of the pre-urban nuclei in the rise more developed ex nihilo, on virgin terrain and development of the mercantile settle­ and without historical antecedents or ties to ments during the period after the Viking centres of administration than those in other invasions, it was almost certainly less a factor areas. On the contrary, they were bound to in the age preceding them. previously existing pre-urban nuclei whose location had been determined primarily historically, definitely not for commercial reasons. The role of these pre-urban nuclei Notes in the rise of the mercantile settlements was for that reason perhaps more important than the revival of international trade, which The topographical isolation of the castel/um at Oudenburg as a result of the Roods of' the Dunkirk-II I think exerted an influence of only second­ transgression east of it is nOL mentioned in the studies ary importance. This role was moreover of Mertens (J962) and Creus (1975). H. Thoen has probably not the mere passive pro tection informed me that thel'e is conclusive archaeological afforded by the military character of the pre­ evidence to prove this. 2 It is striking that Petri (1958b) does the same urban nuclei. Although this military element thing. was present from their Roman past, it was 3 Vita Eligii, 2, c. 2 : Hoc ergo aurifi cem invitum detonsum constituerunt custodem urbium seu not primary. municipiorum his vocabulis, Vennandensi scilicet, Town development before the Viking quae est metropolis urbs, Tornacensi vero, quae invasions is thus similar in many respects to quondam regalis extitit civitas, Noviomagensi quoque the picture which Pirenne sketched for the et Flandrensi, Gandensi e tiam et Corturiacensi. 4 Reasoning further along these lines, Van Wer­ following period. Even for the post-Viking veke (1933: 15) considered these municipia simply as age, we may question whether the prior pagi and not as population centres. presence of a pre-urban nucleus, a com­ 5 I shall deal later with the nature of Aardenburg, located in the pagus Rodanensis. Both the pagus and pletely new castrum or the fortification of an Aardenburg itself were named for the river Rudanna. earlier centre, must not be interpreted as an On the pagus Mempiscu5 see Koch (1950). indication of its more active role in relation 6 I attempted in an earlier study to derive Lhese to the commercial settlement than Pirenne reasons from Bruges' problematical water link with the sea during the late ninth and tenth centuries was inclined to think, despite its now more (Verhulst 1960 :59), but this hypothesis was called into overtly military role. Recent studies of the serious question by Koch (1962 :6-8, 48-9). character and topography of individual pre­ 1 Bonenfant (1953 :420-4) provides a summary of the problems and the literature, as well as an overly urban nuclei from the late ninth and early brief sketch of the earliest history of Antwerp. I must tenth centuries show that they, again con- disagree with several of his views, and of course he

202 could take no account of the excavations, mentiolled Walhmge a Anvers. Analecta bollancliana SO :345- hereaJ'ter, orVan de Walle. 60. The most l'ecent attempt in this direction (Van Crells, I. 1975. De gallo-romeinse nedcrzetti ng onder Acker 1975:1)-62) is insu[hcienriy critical and leaves hel laat-romeins grah'eld van 0 udenbmg. many problems ullsolved. ' Archaeologia belgica 179. Brussel. ~ The gap in alThaeological finds corresponds D' Haenens, A. 1967. Les invasions [1ormandes en essentially to the MerOl'ingian period, and as such is Belgique au ge siecle. Leuven, lloticeable at llIany other places, such as Ghent, Dhondt, J. 1942. Het ontslaan van het vorstendOT11 10 I do not agl'ee with Van Wen'eke (J 965:47) lhal Vlaanderen. Belgisch tijdschrif't \"Oor filologie en the church oj' St Walburga was constniCled only at the geschiedenis 21 :53-93. beginning of the twelfth century, ['or he intel preted Dhondt, J. 1948. DeveIoppemcnt urbain et initiative Van de Walle's excavations too literally. Koch's col11tale en Flandre au lie siecle. Revue du Nord argulllellls (J 962 :44- 8) for Bruges and Coens' (J 962) 30: 133-56, inrormation on the cult of St Walburga at Antwerp Dholldt, J. 1952. Proloog van de Brabantse suggest to me that this ChUld1 could have been built geschicdenis. Bijdrage tot de studie van het no later than the tenth century. Brabantse heem 3: 1-40. Bergen op Zoom. II In view oj" the dilhculties in elating precisely the Dhondt, J. 1957a. De vroege topograrle van Brugge. pottery fragments uncovered at Antwerp, Van de Handclingen del' maatschappij voor Geschiedenis Walk's dating (19fiO, 196 J) of the Carolingian and en Oudheidkunde te Gent. Nieuwe reeks II :3-30. post-Carolingian traces oj" settlement within the castle Dhondt, J. 1957h. L'essor urbain elltre Meuse et Mer is not entirely exact, as indeed Vall Werveke noted du Nord a I'epoque mcrovingienne, Studi in onore (11)65 :46). Va~ de Walle dates the construction or the di Armando Sapori 57-78. Milano. earliest wall around the post-Viking castle artel' the Dhondt, J. 1962. Les problernes cle QuenlOvic. Stucli mid-ninth century, on grounds oj" potLery finds which in onore cli Amintore Fanrani 1: 183-2'18. Milano. could also belong to tbe tenth century. Bu t against this Dhondt, J. alld M. Gysseling. 1948. Vlaandcl·en. view we must consider that the material which Van de Oorspronkelijke ligging en etymologie, Album F. Walle did not note in his publications includes Bam I: 192-220. Antwerpen. material which dates from throughout the nilllh Ganshof~ F. L. 1938. lets over Brugge gedurende de century. [ owe this inJarmatio)) to F. Vcrhaeghe, who precollStitutioncle pcriode van haar geschiedenis. studied the Antwerp pottery finds (Vcrhaeghe Nederlandschc historiebladen 1:281-303. 1975: 163-4). Ganshof, F. L. [949. Het tijdperk van de Merovingen. 12 This is apparent from thc significance which In: Algemene geschiedenis clel' Nederlanden 1:252- must be attachcd in a number oj" placcs, such as H uy, 305. Utrecht. Dinant and ToUl nai, to the existence oj" ~l11all plots Ganshof, F. 1,. 1958. Het tolwezen in het Fl'ankisch with houses called Jedile (Koch 1970:319-:!O). Such rijk onder de Merovingen. Mededelingen van de Jedilia are also found in the countryside (Ganshof" Koninklijke Vlaamse Academic van Belgic, Klasse 1975:73,96), so that their occUlTence 'dot"s not auto­ del' Letteren 20.4. Brussel. matically permit the direct conclusion that the place Ganshof, F. L. 1962. A propos du tonlieu sous les had a commercial character. Mcrovingiens. Studi in onore di Amintore Fanfani 293-315. Milano. Gansbof, F. 1,. 1975. Le polyptyque de I'abbaye de Saint-Bertin (844-59). Edition critique et commen­ taire. Memoires de l'Academie des Inscriptions et Belles-Lettres 45. Paris. Literature Gysseling, M. [950. Toponymie \'an Ouclenburg. Brussel. Gysseling, M. 1971. Een nieuwe etymologie \'an Acker, J. Van 1975. Antwerpell, van rorneins veer tot Brugge. Naamkunde 3: 1-4. wereldhaven. Antwerpen. Jankuhn, H. [958. Die fj'uhmittelalterlichen Sce­ Bonenfant, P. 1953. L'originc des villes braban~onncs handelspIatze im NOI"d- und OstseeraUITl. Vortrage et la 'route' de Bruges a Cologne, Revue beIge de und Forschungen 4 :45 J-98. Konstanz. philoJogie et d'histoire 31 :399-44 7. J ods, A. 1972. A propos de 'b III gus' a H u)' et a Coens, M. 19fi2. Le sejoLll" Icgendaire cle Sainte Namur. In: Festschrift E. Ennen 192-9. Bonn.

203 Kobler, G. 1973. Civitas ulld virus, burg, stat, dor!" Note"daeme, J. 1954. De graven van Vlaandcren en und wik. Vor- und Friihtormen del' europaischen hun domeinen rond Brugge. Handelillgcll del' Stadt im Mittelalter. Abhandlungen del' Akademie Maatschappij voor Geschiedenis en Oudheidkllncle der Wissenscharlen in Gbllingen. Philologisch­ lc Gent. Nieuwe reeks 8:31-8. Histol'ische Klasse 3.83 :61-7 6. Go ttingen. Noterdaeme, J. 1958. Hel onstaan van de parochie Koch, A. C. F. 1949. De ouclenlom van de stad Oudenburg. Sacris erudiri 10: 151-6 J. Brugge. Handelingen van het genoolschap Societe Noterdaeme, J. and E. Dekkers. 1955. Sint Eligius in d'Elllulalion te Brugge 86: 145-50. de Pagus Flandrensis. Sacris crudiri 7: 140-61. Koch, A. C. F., 1950. Le territoil'c des Mcnapiens. Oost, T. 1976. De opgravingen 'Stadsparking' te Tijdschri!"l voor rechtsgeschicdellis 18: 19-35. Antwerpen. Tijdschrirt der slacl Antwerpen 22.2 :69- Koch, A. C. F. 1962. Brugge's topografische onl­ 76. wikkeling tol in de 12e eeuw. Handelingen van hel Petri, F. 1958a. Die Anfange des mittelalterlichen Gcnootschap Societe d'Emulation te Brugge 99 :5- Stadlewcsens in den Niederlanden und dem angren­ 67. zenden Frankreich. Vortrage und Forschungen Koch, A. C. F. 1970. Phasen ill der Entstehung von 4:227-95. Konstanz. Kau fmannsniederlassungen zwischen Maas und Petri, F. 1958b. Merovingerzeitliche VorausseLzungen Nordsee in der Karolingerzeit. In: Festschrift {"Or fur die Entwicklung des Stadtewesens zwischen Franz Perri 312-24. Bonn. Maas und Nordsee. Bernerkungen und Erganz­ Krusch, B. 1902. MGH SS rerum Merovingicarum 4. ungen zu einer Sllldie von J. Dhondt. Bonner Jahr­ Hannover. bOcher 158 :233-45. Krusch, B. 1910. MGH SS rerum Merovingicarum Pi renne, H. 1905. Les villes flamandes avant Ie 12e 5 :395-449. Hannover, Leipzig. siecle. Annales de l'Est et du Nord 1 :9-32. Nancy. Kurzc, F. 1891. Annales Fuldenses. MGH SS in mum Rep"inted in Pirenne 1939: 123-41. scholarum. Hannover. Pirenne, H. 19 10. Les anciennes democra ties des Laet, S. J. De. 1960. Schets van het o11lslaan en de Pays-Bas. Paris. Reprinted in Pit-enne 1939: 143- ontwikkeling van stedelijke agglomeraties in 30 J. Noord-Gallie in de Romeinse tijd. Mededelingen Pirenne, H . 1925. Medieval cities. Their origins and van de Koninklijke Vlaamse Acadcmic van Belgie, the revival of trade. Princeton. French translation: Klasse del' Letteren 22.6. Brussel. Pirenne 1927. Laet, S. J. De. 1969. Les fouilles de Deslelbergen et les Pirenne, H. 1927. Les villes du moyen age. Brussel. origines gallo-romaines de la ville de Gand. Reprinted in Pirenne 1939 :303-431. Archeologia 30:57-69. Pirenne, H. 1939. Les villes et les institutions urbaines. Man, M. De. 1936. De munlen, tot nu toe op en in het 2 vols. Paris. strand bij Domburg gevonden. Archief Zeeuwsch Prou, M. 1892. Les monnaies carolingiennes. Paris. Genoolschap del' Wetenschappen 1936: 1- 15. Mid­ Raepsaet, M.-Th. and G. 1975. Gallia Belgica et del burg. Germania Inferior. Vingt-cinq annees de re­ Mertens, J. 1955a. De opgravingen in de Sint­ cherches historiques et archeologiques. In: H. Tem­ Donaaskerk te Brugge. Streven 9 :57-60. porini and W. Haase (eds.), Aufstieg und Nieder­ Menens, J. 1955b. De opgravingen in de Sint gang del' romischen Welt. Geschichte und Kultur Donaaskerk te Brugge. Wetenschappelijke tijdingen Roms im Spiegel der neueren Forschung 2.4:3-299. 15:246-8. Berlin. Mertens, J. 1962. Oudenburg et Ie Litus Saxonicum en Ryckaert, M. 1972. De Brugse havens in de middel­ Belgique. Heliniulll 2:51-62. leeuwen. Handelingen van het genootschap Societe Moreau, E. de 1942. Saint-Amand, Ie principal d'Emulation te Brugge 109: 1-23. evangelisateur de la Belgique. Brussel. Ryckaert, M. 1973. De Oudeburg te Brugge. Album A. Moreau, E. de. 1947. Histoire de l'eglise en Belgique Schouteet 155-68. Brllgge. I. Brussel. Schlesinger, W. 1954. Burg und Stadt. In: Festschrift Morrison, K. F. 1967. Carolingian coinage. New York. fur Th. Mayer 1 :97-150. Konslanz. Nicholas, D. M. 1969. Medieval urban origins in Thoen, H. 1973. De Belgische kustvlakte in de northern continental Europe: state of resealTh and Romeinse tijd. Gent (unpublished ms.). some tentative conclusions. Studies in medieval and Trimpe Burger, J. A. 1971. Deae Nehalleniae. Gids Renaissance history 6 :55-114. bij de tentoonstelling. Zeeuws Genootschap der Niermeyer, J. F. 1960. Mediae Latinitatis lexicon Wetenschappen: 45-54. Middelburg. minus, 8. Leiden. Trimpe Burger, J. A. 1973. The islands of Zeeland

204

• • \ .. - WI"

. ," and South Holland in Roman times. Berichten Echternach im Friihmittelalter 1.2. Luxembourg. Rijksdienst Oudheidkundig Bodemonderzoek Werveke, H. Van. 1933. Kritische studien bctrelfende 23: 135-48. Amersfoort. de oudste geschiedenis van de stad Gent. Univer­ Vandenborn, O. 1965. Gallo-romeinse vondsten te siteit te Gent. Werken uitgegeven door de Faculteit Antwerpen. Helinium 5 :252-83. der Wijsbegeerte en Letteren 69. Antwerpen. Vercauteren, F. 1934. Etudes sur les civitates de la Werveke, H. Van. 1950. De steden, ontstaan en eerste Belgique seconde. Academie Royale de Belgique. groei. In: Algemene geschiedenis der Nederlanden Memoires in-8°, 33. Brusse!. 2:180-202, 52~4. Utrecht. Vercauteren, F. 1936. Comment s'est-on defendu au Werveke, H. Van. 1965. 'Burgus': versterking of ge siecle dans l'Empire franc contre les invasions nederzet.ting? Verhandelingen Koninklijke Vlaarnse normandes? Annales du 30e Congres de la Academle van Belgle, Klasse der letteren 27.59. Federation Archeologique et Historique de Belgi­ Brusse!. que 117-32. Brusse!. Werveke, H. Van and A. Verhulst. 1960. Castrurn en Vercauteren, F. 1967. Conceptions et methodes de Oudburg te Gent. Bijdrage tot de oudste I'histoire urbaine medievale. Cahiers Bruxellois geschiedenis van de Vlaarnse steden. Handelingen 12: 1-24. der Maatschappij voor Geschiedenis en Verhaeghe, F. 1975. De middeleeuwse keramiek. Oudheidkunde te Gent. N ieuwe reeks 14 :3-62. Gent, duizend jaar kunst en cultuur 3:160-78. Gent. Verhulst, A. 1953. Over de stichting en vroegste geschiedenis van de Sint-Pieters- en de Sint­ Baafsabdijen te Gent. Handelingen van de Maatschappij voor Geschiedenis en Oudheidkunde te Gent. Nieuwe reeks 7: 1-51. Verhulst, A. 1959. Historische geografie van de Vlaamse kustvlakte tot omstrceks 1200. Bijdragen voor de geschiedenis der Nederlanden 14: 1"":3 7. Verhulst, A. 1960. Les origines et l'histoire ancienne de la ville de Bruges (9e-12e siecle). Le moyen age 66:37-63. Verhulst, A. 1962-1963. L'evolution geographique de la plaine maritime Ramande au moyen age. Revue de l'Universite de Bruxelles 89-106. Verhulst, A. 1971. Das Besitzverzeichnis der Genter Sankt-Bavo-Abtei von ca. 800 (Clm 6333). Fnlh­ mittelalterliche S tudien 5: 193-234. M iinster. Verhulst, A. 1972a. Die Friihgeschichte der Stadt Gent. In: Festschrift E. Ennen 108-37. Bonn. Verhulst, A. 1972b. De vroegste geschiedenis en het ontstaan van de stad Gent. Handelingen der Maat­ schappij voor Geschiedenis en Oudheidkunde te Gent. Nieuwe reeks 26:5-39. Vliegher, L. De. 1965. Op de burg te Brugge. Archeologie 2 :54-5. Vries, D. De. 1968. The early history of Aardenburg to 1200. Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 18 :227-60. Amersfoort. Walle, A. L. J. Van de. 1960. De archeologische opgravingen in het oud stadscentrum te Anlwerpen. Tijdschrift der stad Antwerpen 6 :48-61. Walle, A. L. J. Van de. 1961. Excavations in the ancient centre of Antwerp. Medieval archaeology 5: 123-36. Wampach, C. 1930. Geschichte der Grundherrschaft

205

131BL. UNIV. GENT

...------"...... ------