Ummeland voyaging, maritime ‘monks’, and the emergence of the town of Marstrand

the Limfjord was a strait dividing Jutland in a northerly and a south-erly part. As such, it constituted the safest and most frequented sea route between the North Sea and Baltic Sea regions. During the first part of the 12th century, its western inlet silted up, creating today’s Limfjord, thus forcing vessels to take a more northerly route around the dangerous Skaw – a route that had previously been avoided. The present study suggests that this so-called ummeland voyaging had considerable economic and geo-political consequences for the Viken region. In particular, it argues that ummeland voyaging was the main driver to the establishment of the town of Marstrand in the southern part of this region. The hypothesis presented thus contradicts much of the previous research on Marstrand’s urbanisation, which instead states that the town derived from a fishing hamlet.

Ummeland voyaging, maritime 'monks', and the emergence of the town of Marstrand

Staffan von Arbin
For many centuries, the Limfjord was a strait dividing Jutland in a northerly and a southerly part. As such, it constituted the safest and most frequented sea route between the North Sea and Baltic Sea regions. During the first part of the 12th century, its western inlet silted up, creating today's Limfjord, thus forcing vessels to take a more northerly route around the dangerous Skaw -a route that had previously been avoided. The present study suggests that this so-called ummeland voyaging had considerable economic and geo-political consequences for the Viken region. In particular, it argues that ummeland voyaging was the main driver to the establishment of the town of Marstrand in the southern part of this region. The hypothesis presented thus contradicts much of the previous research on Marstrand's urbanisation, which instead states that the town derived from a fishing hamlet.
"Perhaps the answers to some of the questions regarding the emergence of our towns should not be sought within the boundaries of the town itself, but on board a ship approaching the coast." (Crumlin-Pedersen1978, p. 74, author's translation) The present study is spurred by the discovery and investigation of the wreck of a well-preserved 13th-century cog-like vessel off Skeppstad, situated at the mouth of Jorefjorden, north of Gothenburg, in the now-Swedish province of Bohuslän ( fig.  1). At that time, of course, Bohuslän was part of Norway. The shipwreck has been dated to the early 1260s by means of dendrochronology. Moreover, dendrochronological analysis has fixed timber provenance to the Ardennes, that is, the wooded upland areas of south-eastern Belgium and adjacent parts of Luxembourg and France. This implies that timber used for the building of the vessel was floa-ted down the river Meuse, and that it was either built somewhere along the river or further downstream in the Rhine-Meuse-Scheldt delta, which today constitutes the southern part of the Netherlands (von Arbin & Linderson 2011; von Arbin 2021). Investigations thus show that the Skeppstad wreck is likely to be an exponent of the so-called Ummelandfahrt, that is, the hazardous circumnavigation of the Skaw (Skagen), which allegedly commenced during this period in history due to the partial silting-up of the Limfjord strait. As such, it raises several questions in relation to this traffic, which, according to the historian Arnved Nedkvitne, resulted in that "one of Northern Europe's most important sea routes came to follow the Norwegian coast" (Nedkvitne 1983, p. 33, author's translation). One such question concerns what influence, geo-politically, economically and otherwise, ummeland voyaging had on Norway in general, and the Viken area in particular. This influence must have been quite substantial and may have affected various areas of contemporary Norwegian society -not least urbanisation, which is the subject dealt with in this paper.
However, as has been emphasised by Lennart Dalén (1941, pp. 40-74), and, later, Hans Höglund (1975, this 'extrapolation' is mostly based on indirect evidence of rather questionable source value. For instance, as one of the main arguments for an alleged 13th century herring period, the establishment of the town of Marstrand, in the southern part of Viken ( fig. 1), has been emphasised (e.g., Ljungman 1896, pp. 126-127;Hörman 1918, pp. 111-112;Jahnke 2009, p. 168). Even though a herring period very well may have occurred in the 13th century, as some sources suggest, this is, in my opinion, clearly a case of circular reasoning. As I intend to demonstrate in the following, it is also likely based on false assumptions regarding Marstrand's age and origin.
In the present study, which is based mainly on written sources and archaeological evidence, I instead stress ummeland voyaging as the main factor behind the emergence of Marstrand. However, the hypothesis presented suggests that the N development of the town and its different institutions was perhaps not as straightforward as has often been assumed. It should be stressed, though, that the importance of ummeland voyaging in this context has been addressed to a certain extent also by previous scholars (Lönnroth 1963, p. 111;Carlsson 1984, pp. 53-56;Andersson 1988, pp. 179-181;Jahnke 2000, p. 285). This relationship, however, has never been studied in detail. The present study builds on the discussion of Kristina Carlsson from 1984 (pp. 53-56), while following the line of research proposed by Hans Andersson, in 1988 (pp. 179-181). Newly attained knowledge, particularly in respect of Marstrand's Franciscan friary, may provide a better understanding of the matter in hand: the birth and development of Marstrand as a town.

The medieval 'town' concept
What constituted a medieval town? This question has long occupied the thoughts of medievalists and historical archaeologists. The answer, of course, ultimately depends on how one chooses to define 'town'. In older research, towns were often separated from population centres of other kinds by means of various formal criteria, such as privileges or judicial status. This, however, has proved to be insufficient for a number of reasons, not least due to the uneven preservation of written source material containing such information. More importantly, this approach gives us, at best, the date for the 'finished product', which is the fully developed town. Consequently, the relevance of such criteria has been in question in recent years, as they are of limited help in the recognition and understanding of the many facets and inherent dynamics of the urbanisation process (Helle 2006, pp. 11-12). Over the years, scholars have therefore felt compelled to propose their own definitions, introducing other criteria as well. These definitions have won varying acceptance within the academic community, some more than others. Most, if not all of them, however, seem to proceed from the general notion that a town needs to maintain a) a certain population density, and b) certain central functions in relation to its hinterland. Divergence between the different views thus appears to be mainly a question of where emphasis should be put, as for instance, on the town's judicial position and administrative role, or, on economy and trade (Andrén 1985, pp. 12-13).
Inspired by post-war German historical and cultural geographical urbanisation research, Hans Andersson in the early 1970s distinguished three main categories of criteria -functional, topographical and judicial-administrative -that, in his view, characterised the medieval town. The first category describes the town's central functions, as well as its interconnectivity with other towns and places. The second translates to settlement structure and density, whereas the third and final category pertains to its judicial and administrative status (Andersson 1973, p. 24). Andersson's definition has become very influential in Scandinavian, and particularly Swedish, medieval urbanisation research, which is largely due to its use within the seminal Medeltidsstaden project in the 1970s and 1980s.
Another highly influential Scandinavian scholar within the field of urbanisation research is Knut Helle. According to his original definition, published already in 1964, a town is characterised by a certain degree of settlement density, compared to the surrounding countryside. Moreover, it should have some degree of eco-nomic specialisation in relation to its hinterland. Lastly, according to Helle, it must stand out as a judicial and administrative unit of its own, to which certain economic rights could be tied (Helle 1964, p. 9). In a more recent and updated version of this definition, Helle also adds that a social structure, where different groups of specialists live closely together, and thus create complex and dynamic milieus, should be in place as well. At the same time, he de-emphasises his previous view regarding the necessity of judicial and administrative unity. Instead, he concludes that towns often, but not always, had judicial and administrative structures of their own, including town council, administration, and courts (Helle 2006, pp. 16-17).
Even with definitions like these, however, demarcations could be difficult. To what degree, for instance, must the urban structure have developed, in order for a settlement to be considered a town? How dense must the settlement be, and how large the population? Unfortunately, there are no definite answers to these questions. Instead, the answers depend on the societal and cultural preconditions of the particular region and period studied (Helle 2006, p. 17). When studying urbanisation in the medieval period, it is important to understand that there was likely a gliding scale between, on the one hand, purely rural villages, and, on the other hand, settlements with a higher degree of specialisation in handicraft and trade. This is perhaps even truer in a Norwegian context, since Norway had a very small population and was situated in the outermost periphery of European urbanisation (Andersson 1973, p. 24;Helle 2006, p. 10).
Related to this discussion is more recent years' focus on 'other' types of urban space -that is, places that do not automatically fit within the town concept, but still may have played an important role in the overall process of urbanisation. This "urban undergrowth", as it is sometimes termed, comprises, for instance, minor population centres, such as villages and fishing hamlets, periodic marketplaces, wharfs, and harbour sites of other kinds. These were places with specialised functions that separated them from their rural environments. Even in those cases where they did not evolve into towns themselves, they could be regarded as partially or temporarily urban, given their respective function (Brendalsmo et al. 2009, pp. 12-15). Apparently, this is a perspective that has been largely overlooked in previous research. I nonetheless believe it to be of key importance in order to understand places like Marstrand.

The Limfjord connection
Throughout history, the Jutland peninsula has functioned as a physical barrier between the North Sea and Baltic Sea regions, effectively restraining maritime transport ( fig. 1). From Blåvandshuk, Denmark's most westerly point, situated north of the Wadden Sea, to the Skaw in the north, the coast of Jutland is very flat and exposed. Prevailing westerly winds, strong currents, shallow waters, and shifting sandbars, in combination with an absence of natural harbours and sheltered anchorages, made navigation in these waters risky (Holm 1999, pp. 15-16;Hammel Kiesow 2002, p. 55). The name Jammerbugt ('Misery Bay') bears testimony of the dangers posed to seafarers navigating these waters (Ohler 1989, p. 35).
Rather than trying to circumnavigate the Skaw, and thereby being at risk of losing both ship, cargo, and crew, ancient merchants thus sought other options. The transport of trading goods across the neck of Jutland -partially over land, and partially on inland waterways -was long preferred and gave rise to towns like Haithabu and Ribe in the Late Germanic Iron Age, and Schleswig a few hundred years later. Substantial transhipment flows, particularly of highvalue goods, also took place further south along the Lübeck-Hamburg axis (Jensen 1995, p. 250;Kirby & Hinkkanen 2000, p. 72;Hybel & Poulsen 2007, pp. 357-359).
The importance of the Limfjord connection during the Late Iron Age and Viking Age is a longdebated issue. Carl Johan Becker argued already in the 1950s that the Limfjord was the predominant transport route between Western Europe and the countries around the Baltic (Becker 1953, pp. 153-154). This view has, however, been challenged by Olaf Olsen. Based on the relative absence of import finds, he suggests that the seaborne transport that took place within the fjord in this period was in fact of mainly local interest (Olsen 1975, pp. 239-240). Erik Johansen, who has conducted the most recent study of the archaeological evidence in the Limfjord area, does not exclude the possibility for long-distance transport. His conclusion, however, is that the Limfjord in this respect mainly served as a transit route (Johansen 1992, p. 22).
The question of when the western inlet of the Limfjord closed has also been the subject of much debate. Danish chronicler Saxo Grammaticus gives a hint in his work on the history of the Danes, Gesta Danorum, written around the year 1200. Here he describes how the flotilla of the Danish king, Canute the Holy (r. 1080-1086), gathered in 1085 in the Limfjord in advance of a planned attack against England. According to Saxo, the ships sailed to the "edge" of the Limfjord, "from which it is only a very short sail to the Ocean". He continues by saying that "at one time this passage was navigable, but now an intervening sandbar closes the exit" (Friis- Jensen & Fisher 2015, book XI, para. 13.5).
A similar account is given by Saxo's fellow-countryman Svend Aggesen, who in his Brevis Historia Regum Dacie, written in the 1180s, states that the flotilla gathered at Humlum, "which at the time was a harbour by the sea" (Olrik 1900(Olrik -1901. Humlum is situated at Nissum Bredning, in the western part of the Limfjord. If these two statements are to be trusted, it can be concluded that the closing probably occurred sometime between the latter part of the 11th century and the late 12th century. However, a further indication is given by the move between the years 1134 and 1139 of the bishop's seat from Vestervig, close to the presumed inlet at Agger Tange, to Børglum, further to the north. This initiative has been taken as proof of the area's reduced significance, and may thus indicate a blocked, or at least highly restricted, thoroughfare (Jensen 1992, p. 56;Lund 1998, p. 173). From this, it seems likely that the inlet closed during the first decades of the 12th century, which appears to be the opinion borne by most scholars (e.g., Knudsen & Kock 1992, p. 153;Westerdahl 1995, p. 222;Wohlfahrt 1998, p. 204). Since silting up is likely to have been a gradual process, it is perhaps not possible to settle a particular year for the event.
In other words, whether the Agger Tange passage was to be regarded navigable or not may, for a certain period of time, have been dependent on the type and size of the vessel being used.

The circumnavigation of the Skaw
The earliest mentions of ummeland voyaging stem from the year 1251. In two charters, dated September 24 and 25 respectively, King Abel of Denmark (r. 1250-1252) grants merchants, presumably from the town of Kampen, certain privileges in relation to the Scanian fairs. The merchants he calls vmlandsfarae, which could be translated to 'those who travel around [the Skaw]'. The first charter concerns customs tariffs and the merchants' right to salvage their own shipwrecked goods. The second one gives the merchants extended rights with regard to the loading and unloading of their vessels at times otherwise prohibited (Höhlbaum 1876, Nos. 411 & 413;Hermansen 1938, Nos. 50 & 52). Based on these charters, it has been assumed that the mid-13th century was the actual startingpoint of ummeland voyaging (e.g., Hørby 1966, pp. 246-247;Ventegodt 1982, pp. 60-62;Nedkvitne 1983, pp. 51, 58-59;Westerdahl 1995, p. 222). There is reason, however, to challenge this notion. The fact that the name used by the king to denote the merchants, vmlandsfarae, is not further explained in the charters actually implies that the phenomenon was not entirely new. Thus, King Abel's charters could likely be taken as proof of an already existing practice (cf. Ventegodt 1982, pp. 60-61;Hammel-Kiesow 2002, pp. 55, 89).
The notion that ummeland voyaging is likely to have occurred several decades before the 1250s is also supported by dendrochronological investigations of early Danish 'cog' finds -primarily the so-called Kollerup and Skagen cogs, dated to c. 1150 and c. 1193 respectively. Dendrochronological analysis shows that both vessels were most likely built in the south-eastern part of Jutland. Provenance thus suggests that they foundered en route, while attempting to sail around the Skaw (Crumlin-Pedersen 2000; Hocker & Daly 2006). Based on this evidence, I envisage a more or less seamless transition to ummeland voyaging after the closing of the Limfjord strait. As stated, such a scenario would however largely contradict the long-prevailing view on the matter.
At the same time, is important to stress that the transhipment of goods over the Jutland neck, and via Lübeck-Hamburg in particular, continued during the entire medieval period, particularly for high-value goods. From this, a continued reluctance to travel ummeland, which was probably still considered too risky a proposition in many cases, may be inferred. The Skaw route seems to have been used mainly for bulk commodities, such as herring, grain, tar, pitch and other products from the countries surrounding the Baltic. Such commodities were hugely important for the economic growth of the towns of the North Sea Region (Kirby & Hinkkanen, 2000, p. 72;Hammel-Kieslow 2002).
It appears that the term vmlandsfarae originally applied solely to merchants of the towns of present-day Netherlands. This is also confirmed by the fact that early customs charters, relating to ummeland voyaging, all seem to concern merchants resident in this region (Ventegodt 1982, pp. 60-61). The connection between early cogs and ummeland voyaging is illustrated not only by archaeological ship finds, but also by written sources. King Abel's 1251 charter, for instance, specifically mentions cogs (coggone) in the context of ummeland voyaging (Höhlbaum 1876, No. 411;Hermansen 1938, No. 50). This may indicate that the development of this ship type was in fact sparked by ummeland voyaging and the increase in bulk trade at that time (Westerdahl 1995, p. 222;Crumlin-Pedersen 2000, pp. 238-239;Zwick 2016, pp. 33-35).

The Skaw-Marstrand route
One of the most important written sources to medieval maritime transport geography and hands-on navigation is a 15th-century rutter (sailing instruction), the German Seebuch (Koppmann 1876). It covers most of the European coast, including parts of the Baltic. It is known from two handwritten manuscripts that are bound together into one volume and preserved in the Commerzialbibliothek in Ham-burg. A third, and lesser-preserved, version of the text used to be kept at the Gymnasialbibliothek in Halberstadt, but was presumably destroyed during World War II (Rösler 1998, pp. 107-108). All three manuscripts are written in Middle Low German. Based on language and context, the writing can be dated to around 1470. However, as has been shown by Albrecht Sauer (1997, pp. 219-221), it is likely to go back in part to 14th-century Dutch sources.
Unfortunately, the Seebuch leaves out most of the Norwegian coastline. Regarding ummeland navigation, it still gives us some important clues, though ( fig. 2). According to the Seebuch, a ship sailing northwards from any of the ports in the Rhine-Meuse-Scheldt delta, bound for the Baltic, would have had three main options after having circumnavigated the Skaw (Koppmann 1876, ch. XII). First, it could follow the Jutland coast southward to the Danish straits. Second, it could take a more direct course to the Sound, keeping east of both Laesø and Anholt. Third, it could go along with the Jutland Coastal Current -preferably with additional aid of the prevailing westerly winds -across the Kattegat, heading for the southern Bohuslän archipelago. In this case it would end up somewhere in the area of the island Marstrand. From there, however, the route is not further described. The first of these alternative routes was apparently used in order to reach destinations in the Kiel and Mecklenburg bays, whereas the two latter were used by vessels heading for ports in the Baltic Proper.
Doubtlessly, the Skaw-Marstrand route was also the preferred one for ships heading for Kungahälla, Oslo, Tønsberg, Skien, or any of the other contemporary trading ports in Viken. As indicated by the Seebuch, western ships bound for ports further north along the Norwegian coast probably chose the more direct course to Cape Lindesnes or Skudenes (Koppmann 1876, ch. XII, paras. 1 & 2). Cape Lindesnes' pivotal role for the traffic between the Baltic and North Sea regions in medieval times is well documented (e.g., Stylegar 2004). The Skaw-Marstrand route is, for some reason, much lesser known. It is described in both preserved versions of the Seebuch, although with slightly different wordings (table 1).
For a ship bound for the Baltic, this route would decrease the risk of the ship being stuck on the Skaw's ever-moving sand banks. However, it was in no ways a route without hazards. The many islets and skerries of the Bohuslän archipelago make coastal navigation in these waters a difficult undertaking, and especi-  Rösler (1998, p. 113), a kenning is equivalent to 4 or 5 nautical miles (1 nautical  ally so in an age where proper nautical charts were non-existent. The so-called Pater Noster skerries north of Marstrand, that stretch approximately 7.5 kilometres to the southwest from the Tjörn 'mainland', pose a particular danger to shipping in this area ( fig. 3). The place name, Pater Noster, can be found in, for instance, Laurentz Benedicht's Søkartet offuer Øster oc Vester Søen from 1568 (Benedicht 1915) and Lucas Janszoon Waghenaer's Vanden Spieghel der Zeevaert from 1585 (Waghenaer 1964), and is likely to be of medieval origin. According to toponymists, the name most probably alludes to the hazardousness of the skerries to passing seafarers, who used to read the Lord's Prayer (Pater Noster) in order to safeguard themselves and their ships (Ekenvall 1992, p. 211).
When approaching the coast from the open sea, it was thus of utmost importance to be able to establish ones' position as accurately as possible. For many centuries, this was mainly achieved with the aid of a sounding lead and different landmarks, natural as well as artificial (Kirby & Hinkkanen 2000, pp. 64-66).
As for Marstrand, there is written evidence of one or possibly more navigation marks on the island from the second part of the 16th century at the very least (Benedicht 1915, p. 120;Nilssøn 1981, p. 195). Possible navigation marks can also be identified on one of the charts in Waghenaer's Vanden Spieghel der Zeevaert from 1585 (Waghenaer 1964). Whether they may have had medieval predecessors is not known, but it is not in any way unlikely. However, for a sailor coming from the Skaw, Blåkullen -the peak of the island Brattön, situated just to the northeast of Marstrand -would be the first coastal feature that became visible (Holmberg 1979, p. 313). The use of Brattön for navigational purposes in historical times is widely recognised (fig. 3). It was included already in the first printed North European rutter, Jan Severszoon's De kaert va[n]der Zee, published in Holland in 1532 (Knudsen 1914). With a present height of c. 130 metres above the present sea level, its peak constitutes the highest point in the Bohuslän archipelago (Östholm 1999, p. 4).
Based on a simple formula (see Zwick 2016;pp. 19-20), the theoretical maximum distance a lookout positioned in a ship's top castle, 20 metres above sea level, would have been able to spot Brattön from in the mid-13th century can be calculated to 56.4 kilometres, or 30.5 nautical miles. As a comparison, the distance between Brattön and the Skaw Spit is approximately 70 kilometres, or 38 nautical miles. The time spent at sea without having any eye contact with either of the two coasts would in other words have been very limited, in good visibility and favourable sailing conditions possibly less than one hour.

Marstrand before Marstrand
Today, the name Marstrand denotes both the island of Marstrand, located in the outermost part of the southern Bohuslän archipelago, and the small coastal town situated on the very same island. The town, which originally evolved on the south-eastern part of the island, nowadays also comprises the settlement on the adjacent island Koön to the east. Thus, the strait between Marstrand and Koön, which also constitutes the original natural harbour basin, divides the settlement in two (fig. 4). The narrow strait is over one kilometre long and offers lee from most wind directions, including the prevailing westerly winds. This, in conjunction with great water depths, easy access from two directions, and largely ice-free conditions during winter due to more or less constant water currents, have given Marstrand a reputation as one of the finest harbours on the Swedish west coast (e.g., Holmberg 1979, p. 208).
Lönnroth certainly mentions the rise of the Hansa, as well as the increased significance of maritime trade in the 13th century, as part of the explanation, and he stresses ummeland voyaging as an important contributing factor. At the same time he states that the town "could hardly have developed otherwise than as a fishing hamlet, which in turn has attracted passing trading vessels; its total absence of an economic hinterland effectively rules out other explanations" (Lönnroth 1963, p. 112, author's translation). I do not share Lönnroth's view, simply because it lacks empirical backing and, presumably, is based on the wrong premises.
Marstrand is first mentioned in the Hákonar saga Hákonarsonar. It tells us that the envoys of Haakon Haakonson's (r. 1217-1263) coregent Skule Jarl sailed in the autumn of 1226 from Halland bound for Oslo, but were forced to stay the winter in Marstrand because of difficult ice conditions. During their stay, they were attacked by representatives of the revolting Ribbunger party. Even though many merchants were lying there with their ships, none of them came to their rescue, the saga states (Storm & Bugge 1914, pp. 148-149). The saga does not mention any sett- lement on the island. Thus, when Marstrand first appears in history, it is in the capacity of harbour, not town. This is something that has often been neglected.
According to the saga, the attack was directed towards the ships where Skule Jarl's men were staying. If there had been anything like a town-like settlement on Marstrand at the time, one would expect the retinue to have been quartered there instead. Consequently, the mention of wintering merchants and their ships cannot be taken as proof for the existence of a town or even of a trading post (see e.g., Berg 1912, p. 18;Krantz 1919, p. 9;Holmberg 1979, p. 191;Olán 1982, p. 18). Instead, as stated by Lennart Dalén (1941, p. 48), one must simply assume that the merchants were staying in the harbour for the same reasons as the envoys. In other words, the saga gives no proof of a settlement, permanent or seasonal, at the time. Nor do the archaeological observations that have been made within the town over the years.
Already in 1744, Andreas Olavi Rhyzelius brought forward the opinion that Marstrand was founded by the Norwegian king Harald Gille (r. 1130-1136) in the year 1132. He also contended that, in 1138, a church was built on the island, "as shown by an old stone inscription" (Rhyzelius 1744, p. 131-132, author's translation). However, since there are no primary sources to support these claims, they could most probably be dismissed as le-gends. A more common view is that Marstrand's establishment should be attributed to the above-mentioned Haakon Haakonson. The reason for this is that the Hákonar saga Hákonarsonar declares that King Haakon 'built' Marstrand (e.g., Krantz 1919, p. 6;Aasma 1974, p. 90;Hasselmo 1980, p. 27;Olán 1982, p. 16;Svensson 1983, p. 7;Helle 2006, p. 68).
This information occurs in a recitation of the many things the king managed to achieve during his long reign. As for the Viken area, it states that he, among other things, built a castle on Ragnhildsholmen, close to the town of Kungahälla. In Kungahälla he erected houses within the royal estate, or 'kungsgård'. Furthermore, he built Gulleyna (interpreted as Gullön in the river Nordre älv, or, possibly, Gullholmen northwest of Orust), and he deforested the Öckerö islands, where he also built houses and erected a wooden church. Finally, he "built Marstrand and other desolate islands in Viken" (Storm & Bugge 1914, p. 294, author's translation). It has been suggested that the king's building activities on Marstrand and other islands should be understood as a kind of confirmation of already existing settlements (Krantz 1934, p. 4;Lönnroth 1963, p. 109). Once again, I beg to differ.

What was it that King Haakon built?
There is probably no need to dispute the saga's claim that King Haakon  tament of King Magnus VI of Norway (r. 1257-1280), which was written down in 1277, contains a list of the monasteries in Norway (D.N. IV, No. 3). Since Marstrand is not included, it has been assumed that it was founded sometime between the years 1277 and 1291, that is, long after Haakon's reign. Thus, a more likely founder is Duke Haakon Magnusson of Norway, who later became King Haakon V (r. 1299-1319). He was known as a keen supporter of the Franciscans, and also founded a Franciscan friary in Oslo during this period (Ekroll 1993, pp. 136, 149;Digernes 2010, pp. 17, 27).
Another option is that he founded a church on Marstrand, as has been suggested by Claes Krantz (1919, p. 25). There seems to be two main views regarding the current parish church, Mariakyrkan (i.e., the Church of St Mary, fig. 5). First, there are those who argue that it constitutes the church belonging to the friary (e.g., Lönnroth 1963, p. 145;Aasma 1974, p. 94;Olán 1982, pp. 177-178). On the other side are those who suggest that there were two different churches, one parish church and one friary church, and that the latter was torn down together with the rest of the friary during the Reformation. Thus, its remains should be sought elsewhere on the island (e.g., Krantz 1919, p. 25;Wideen 1975, p. 261). The origins of Marstrand parish are unclear. In the tax register of bis-hop Eystein from the late 1300s, in which the estates of the Oslo diocese are listed (Huitfeldt 1958), Marstrand is not mentioned at all. The first written mentions of a parish church occur in a series of indulgence letters, issued by the archbishop Henrik Kalteisen during a stay in Marstrand in the winter of 1453/1454. He there explicitly states that the church was dedicated to The Holy Virgin Mary (Bugge 1899, pp. 208-211). These sources indicate that Marstrand parish is a fairly late creation, possibly no older than the 15th century. The idea of a church built on the initiative of King Haakon can thus probably be dismissed.
So, if there were no church in Marstrand other than the friary church, how did the inhabitants fulfil their Christian duties? Did they in fact, as contended by Eskil Olán (1983, p. 177), use the church of Lycke parish, further inland? Alternatively, was the church of the friary open also for ordinary people, as has been suggested by Claes Krantz (1934), and later Karin Aasma (1974, p. 94)? None of these explanations, however, has passed without criticism (e.g., Krantz 1919, pp. 25-26;Wideen 1975, p. 261;Hasselmo 1980, p. 30).
The discovery and partial excavation in 1981, and particularly in 2017-2018, of large building foundations, interpreted as belonging to the friary, to the north and south of the present church ( fig.  6), have finally put an end to the discussion (Carlsson 1984;Johansson & Bergstrand 2017;Johansson 2020a). These excavations show that the Mariakyrkan most definitely was built as part of the friary. A recent radiocarbon analysis of skeletal remains found beneath one of the church's counterforts, which is part of the vault construction, gave as a result AD 1246-1302(95.5%), or, AD 1265-1290Bergstrand 2018, p. 11, appx. 2). Since the vault is likely to be of a later date than the original church building (Aasma 1974, p. 126), this result does not contradict a suggested construction between the years 1277 and 1291.
This means that the church most probably served not only the friars, but also the other residents on the island. Of course, this dual function may not have been there from the beginning. Neither does this exclude the possibility for a phase that involved, for instance, the church of Lycke. However, the main point taken here is that there was probably no need for a church during the 13th century, due to the insignificant number of permanent inhabitants on Marstrand and surrounding islands.
Thus, what remains is the castle (castello). Possibly, it was more of a fortified royal estate. Its only mention is in conjunction with an attack by merchants from Kampen in 1368, during which both castle, friary and church were burnt (SDHK,No. 9730). It may very well be that Haakon Haakonson directed its building, which has also been suggested by Krantz (1919Krantz ( , p. 25, 1923. The incident in Marstrand harbour in the winter of 1226, as well as other, similar events, may have instigated its construction. However, ummeland voyaging, which during the course of the 13th century made Marstrand an internationally renowned harbour, was probably the main motive. In the beginning of that century, the Norwegian kingdom was still very fragile. King Haakon most certainly recognised the strategic need to exercise control of its important harbours and sea routes, in order to suppress unregulated trade and to ensure collection of taxes and tolls.

The Franciscans
The establishment of a Franciscan friary in the late 13th century has often been taken as a sign that Marstrand had reached some degree of urban development (e.g., Krantz 1919, p. 10;Hasselmo 1980, p. 29;Rasmussen 2002, pp. 88-89;Svedberg & Jonsson 2006, p. 14;Johansson & Bergstrand 2017, p. 8;Bergstrand 2018, p. 7;Johansson 2020a, p. 7). However, as will be shown in the following, this might be a misconception. Instead, I suggest that it was established prior to the emergence of a town-like settlement on the island. My hypothesis is that seafaring, due to the increasing number of calling vessels caused by ummeland voyaging, may have provided the necessary supply base. It must be admitted, though, that this idea is not entirely new. Both Kristina Carlsson (1984, p. 8) and Hans Andersson (1988, pp. 179-181) have hinted in a similar direction, although Andersson does not specifically discuss ummeland voyaging in this context. Hence, the objective here is to try to further elaborate arguments.
It is true that the Franciscans were mendicants, and therefore highly dependent of alms. It is also true that mendicant orders often settled in towns for reasons of sustenance and support. In fact, their connection to towns has been considered so strong, that their presence has commonly been used as a criterion for urbanisation (e.g., Andersson 1973, pp. 38-39;Andrén 1985, p. 59). This, however, may very well be a case of circular reasoning, as pointed out by Christian Lovén (2001, p. 255). There are, in fact, several examples of Franciscan establishments that were built and functioned in more rural environments. Within the realm of medieval Sweden, Jarl Gallén (1959, p. 569) lists Krokek in Kolmården, Uppgränna in Holaveden, Vätö in Roslagen, and Kökar in the Åland archipelago. Of these, Krokek and Kökar had proper friaries. As for medieval Denmark, there is Torkö in the Blekinge archipelago, which initially had a chapel that later became a friary. As noted by Gallén, these establishments were all located along frequented land or sea routes. Regardless of whether they were situated inland or by the coast, a common denominator seems to have been their location in communications networks. It thus seems as if the important thing was not the town per se, but the proximity to potential donors (Sjöstrand 1993, p. 71;Digernes 2010, pp. 18-19).
Obviously, donors did not have to be town dwellers. They could equally well be travelling merchants, pilgrims, sailors, or for that matter, fishermen. Several scholars have noted that the activities of the Franciscans, just like the Dominicans, seem to have been orientated largely towards the sea and maritime occupations. Apparently, coastal towns and ports were of particular importance (Gallén 1959(Gallén , p. 569, 1993Sjöstrand 1993, p. 77;Digernes 2010, p. 38). Here, besides begging and preaching, Franciscans were able to conduct different clerical services, such as performing pastoral care and giving sacraments. Although under a general interdict, they often seem to have performed proper church services as well. In towns, this turned out to be a recurrent cause of conflict between the friars and the secular clergy. Much of the conflict was due to the constantly changing papal statutes, which led to ambiguity (Gallén 1959, p. 569;Digernes 2010, pp. 53-55). Franciscans were also entitled to perform funeral services (Digernes 2010, p. 63;Larsen 2015, pp. 296-297), and in the late middle ages, they were involved in the issuing of letters of indulgence, which became an important source of income (Digernes 2010, pp. 77-79).
There is written mention of wintering sailors in Marstrand, who during the winter of 1531/1532 visited the church to commune and confess (Aasma 1974, p. 127). As this occurred shortly before the Reformation and the dissolution of the Franciscan order in Norway, the friary was in a very poor state and operated by just one friar. Still, the episode clearly illustrates the demand for clerical services that existed among travelling merchants and sailors at the time. However, the services offered by the Franciscans on the island of Marstrand and other, similar locations may have extended beyond the religious sphere.
Alms were an important means of support, but friars were also expected to work for subsistence. As pointed out by Knut Willem Ruyter (1982, p. 147), the denomination 'mendicant order' is therefore somewhat misleading. Unfortunately, not much is known about the work of the Franciscans in the Nordic countries, but among their presumed occupations were gardening, fishing and brick making (Gallén 1959, p. 569). Fortunately, in the case of Marstrand, a tinsmith who also happened to be the friary's gardian, or master, is mentioned in 1423 (D.N. III, No. 666).
Other pursuits that may have been undertaken at Marstrand includes the loading and unloading of ship cargo, the replenishment of drinking water and provisions, as well as the undertaking of ship and sail repairs. Some friars may even have worked as sailors themselves. It is even possible that some friars were employed as pilots. The need for persons with a local knowledge in coastal navigation must have been crucial and is, in my opinion, something that often has been overlooked by maritime historians and archaeologists in the past. As pointed out by Per Olof Sjöstrand, Franciscans were probably among the ones that had the best first-hand knowledge of frequented paths and sea routes in medieval Europe (Sjöstrand 1993, p. 72;cf. Gallén 1993, p. 82 regarding the so-called Danish Itinerary).

Foundation of the town of Marstrand
So, if the establishment of a castle and a friary cannot be taken as proof of an existing town-like settlement, as has been argued in the previous sections, at what point in time could Marstrand actually be considered a town? Archaeologically, there is not very much to cling onto. The only standing medieval building on the island is the present church, which, as I have shown, was built as a Franciscan friary church in the late 13th century. The oldest artefacts discovered during the Drottninggatan excavations in 1981 consisted of a few shards of red earthenware and Siegburg-type pottery, which can be typologically dated to the first part of the 14th century (Carlsson 1984, p. 37). However, since they were found in 16th-century secondary deposits some ten metres to the north of the church, and thus in the immediate vicinity of the friary, they cannot be taken as evidence of a secular settlement at the time. More recently, excavations in Kvarteret Korvetten have revealed foundation layers of probable late medieval date. These, however, are not likely to be older than c. 1400 (Johansson 2020b). Neither do Marstrand's town privileges provide us with any useful leads on this matter. Christopher of Bavaria (r. 1442-1448) first issued privileges in 1442 (Herlitz 1927, No. 300). At this time, however, Marstrand had most certainly been a town for a while already. King Christopher's privileges also granted the burghers of Marstrand trade rights for the entire Bohuslän, whereas foreign merchants were prohibited to trade outside the town borders. Thirteen years later, in 1455, King Christian I (r. 1449-1481) issued new privileges. In the letter of privileges, several offices and institutions, such as a lawman, a mayor, aldermen, a town council and a bailiff, are mentioned (Herlitz 1927, No. 322). Predecessors to these two letters are not known, but this of course does not exclude the possibilities for earlier privileges.
The oldest securely dated surviving town seal of Marstrand is marked with the year 1611 (Scheffer 1967, p. 49). It consists of the year and a six-pointed star, which are surrounded by three fish (presumably herring) arranged in a triangle, and the Latin text S * OPIDI * MASTRANDENSIS *. An older seal, apparently with a similar design and text, is described by Johan Oedman in 1746 (Oedman 1983, pp. 100-101). According to Oedman, this seal was marked Anno MCCCXXX (1330) in the margin. It is not clear from Oedman's description, however, if the latter is part of the original motif, or if it is a later addition to the seal matrix. Moreover, it is not evident whether Oedman actually had the opportunity to examine the seal himself, or if his account is based on someone else's description of the item. However, due to the many dubious facts presented by Oedman, there is every reason to treat this information very cautiously.
The earliest written documents that mention Marstrand in a way that could indicate that it had become some kind of urban centre date from the years 1293 and 1310, respectively. In 1293, a iusticiarium regis, a royal deputy, is mentioned in two related treatises in which the town councils of Stavoren and Kampen sought alliances with the Wendish towns due to complaints from King Eric of Norway (r. 1280-1299). The king was upset by merchants from the towns, because of their interference in domestic legal matters. At Pentecost the previous year, the merchants had captured some pirates near Marstrand. The pirates were brought before the king's deputy, who sentenced them to death for their alleged misdeeds (D.N. V, Nos. 21 & 22). It is possible that both the trial and the subsequent execution took place on Marstrand, although sources are somewhat ambiguous in this respect.
In any case, the mention of a royal deputy does not necessarily prove the existence of a town. The deputy may in fact have been the person in charge of the castle or royal estate that likely existed on the island at the time, and the trial may have been held either there or in the friary.
The second document -a complaint regarding a shipwrecked vessel in Kalvsund, a natural harbour or anchorage located within the Öckerö archipelago, south of Marstrand -was issued to King Haakon V by King Edward II of England (r. 1307-1327). The vessel in question belonged to a merchant named William de Toller from Grimsby, and was loaded with, among other things, Baltic grain, when it foundered en route to England in 1309. According to the complaint, the cargo of the ill-fated vessel was taken ashore and seized by the men of Peter Dyre, who reportedly was the balliuo (bailiff ) of ville de Malestronde (D.N. XIX,No. 472). In Scandinavian medieval Latin, the word uilla could be translated as 'town', but also, for instance, as 'farmstead' or 'village' (Odelman 2002, p. 569). Thus, ville de Malestronde do not necessarily relate to a town or even a village. In this context, it could probably allude to a manor or royal estate as well. Like the iusticiarium regis mentioned in 1293, Peter Dyre may have been the bailiff or castellan, as suggested by Claes Krantz (1919, p. 37). In fact, they may even have been one and the same person.
With the exception of these three documents, there is little evidence to suggest that Marstrand had actually become a town before the middle of the 14th century. Residents are featured for the first time in 1363, when Marstrand dwellers Albrekt Gire and Peter Volkersson witness a promissory note issued between Henrik Berendsson, burgher of the town of Harderwijk, and Gotskalk Skarpenberg, burgher of Lübeck (R.N. VI,No. 910;SDHK,No. 8331). Both documents are sealed with their owners' respective personal seals (Homeyer, 1870, pp. 386 & 390, Tafel XIII, No. 72 & XVI, No. 321). The names of the two residents indicate German descent, which shows that foreigners were able to become town citizens (Helle 2006, p. 86). Documents listing institutions, offices and occupations that could imply the existence of a town are otherwise rare in this period. This is despite the fact that Marstrand occurs in a relatively large number of documents from the late 13th and 14th centuries. However, even in some of the later sources, Marstrand is mentioned exclusively as a portus, or 'harbour'.
A circumstance that, indirectly, may support the idea of a foundation during the 14th century is the decline of the town of Kungahälla, approximately 20 kilometres to the southeast. Kungahälla, which was likely founded already in the late 11th century, for a long time remained the only proper town of the Southern Viken area. Archaeological excavations, conducted in the 1980s and 1990s, have shown that the settlement developed rapidly from the mid-12th century and up to the beginning of the 14th century, after which it began to diminish. This may indicate that it gradually became obsolete, due to changed economic and geo-political conditions. Surely, this decline was in part prompted by factors such as the late medieval agrarian crisis and the bubonic plague (cf. Spufford 2000, pp. 200-208). It however appears as if Marstrand soon came to supersede Kungahälla as the principal trading town of the region, whereas the newly founded Bohus fortress, located further inland, became the new administrative and fiscal centre (cf. Hasselmo 1980, p. 28;Andersson 2001, p. 189;Helle 2006, p. 127). For an interim period, it is even possible that Marstrand served as an 'outport' to Kungahälla.

Conclusion
The inception of ummeland voyaging in the 12th century, caused by the gradual silting-up of the western entrance of the Limfjord strait, could be likened to a major, permanent, detour. As such, it had farreaching implications on the contemporary transport geography of north-western Europe, as it forced seafarers to take a more northerly route around the dangerous Skaw. Moreover, it coincided in time with a major upswing in the seaborne long-distance trade of bulk commodities, which, in turn, was facilitated by the introduction of new, and gradually bigger, seagoing cargo vessels, commonly known as cogs. It is highly likely that ummeland voyaging was an important stimulus in this technological development.
For much of this ummeland traffic, the sheltered natural harbour to the immediate east of the island Marstrand, in the southern part of Viken, came to serve as a nodal point due to its strategic position and favourable conditions. Much of the earlier research concerning the emergence of the town of Marstrand has focused too narrowly on the herring fishery as the main explanatory factor, thereby missing out that Marstrand, first and foremost, provided a safe haven for ships in transit. This primary function seems to have persisted long after Marstrand had become a fully developed town with all that that entails. The urban development of Marstrand should thus be understood mainly from a seafaring perspective, which makes the question of economic hinterland largely irrelevant, at least initially. This, however, does not exclude the possibility of a large-scale herring fishery being part of the equation as well.
Marstrand's urban transition is generally assumed to have taken place in the first part of the 13th century. However, a 'close reading' of historical and archaeological sources do not support the idea of a population centre on the island before the turn of the 14th century at the very earliest. In fact, it is not until the second part of that century that there is more pronounced evidence of an existing town. This allows for a slightly different interpretation of Marstrand's development than is usually suggested. To begin with, there is no evidence to support the claim that the town was preceded by a fishing hamlet. Instead, the first establishment on the island was likely a castle or fortified royal estate. It was probably built during the reign of King Haakon Haakonson, sometime between the years 1226 and 1263, as a direct response to ummeland voyaging. To the king, this stronghold would have been an important means of enforcing economic and military control of Norway's southern border areas.
A Franciscan friary was built a few decades later, between the years 1277 and 1291. It was likely founded by Duke Haakon Magnusson, who later became king of Norway. By that time, seafaring had increased sufficiently to make the friary economically viable. The friars themselves were possibly involved in different activities related to this traffic. Mariakyrkan, Marstrand's first and presumably only church, was built as an integrated part of the friary, but also served foreign merchants, sailors and other residents, temporary and permanent, on Marstrand as well as from the surrounding islands. Further settlement on Marstrand is likely to have developed gradually. Of course, this does not necessarily imply that the town developed in an 'organic' and uncontrolled manner. However, since written sources are unlikely to provide any further details on the matter, future research will have to depend largely on archaeology.
Acknowledgements: Many people have read and commented on this paper in its different stages of completion. For this, I am of course very grateful. I am also greatly indebted to Magnus Krook, for helping me with the translation from Latin to Swedish of a number of important medieval diplomas, as well as for acting as a sounding board on various linguistic matters. My thanks also go to Auður Magnúsdóttir, who has helped me in a similar way with questions relating to Old West Norse sources. English was revised by Gwilym Williams.
Staffan von Arbin is a PhD candidate in archaeology at the Department of Historical Studies, University of Gothenburg. E-mail: staffan.arbin@gu.se