Although the military perspectives are fundamental, I believe that other aspects of the forts should be considered too. The lack of forts in the plains of southern Scandinavia shows that they were not essential to the warfare of the time. Forts were absent in areas with large and densely settled villages, but they were built in areas where settlement consisted either of single farms or of small villages and loosely organized larger villages (see Fallgren 1993; Riddersporre 1995). Forts were thus constructed in areas where the intimate and everyday interaction of the large villages of the plains did not exist. The forts thus seem to have met a special social need in areas where settlement and land use were more divided (see Wall 2003). Apart from this social dimension, the ritual and cosmological features that some scholars have detected in the forts must be reinvestigated. But the military and religious elements need not be viewed as being as separate as they are often claimed to be; they are more likely to have been different aspects of the function of forts. Instead of demilitarizing the ringforts, I think that the character of war itself should be redefined.
Outlines of an early Scandinavian martial culture
Only in recent years have war and warfare been more thoroughly debated in archaeology (for instance, Randsborg 1995; Nørgård Jørgensen and Clausen 1997; Carman and Harding 1999, Jørgensen et al. 2003; Otto et al. 2006, Vandkilde 2006, Hedenstierna-Jonsson 2006; Holmquist Olausson and Olausson 2009). Many important aspects of war have been underlined, such as its social and political consequences and the ritualized nature of much pre-modern warfare. Still, war as a historical phenomenon is an elusive activity. In some cases we can find the actual sites of warfare, such as battlefields (Schlüter 1997), burned hillforts (Olausson 2009:57), or burned and destroyed hall buildings (Herschend 1997). In other cases, traces of violent and bloody acts of war are found, such as mass graves of the slain (for example Thordeman 1939–40; Kjellström 2005). For the distant past, however, warfare can chiefly be traced by indirect means, and this applies particularly to Iron Age warfare. Apart from hill forts, Scandinavian warfare can, above all, be studied through weapon deposits, boathouses, and ringtun.
A good window on the culture of war is undoubtedly provided by the weapon deposits. These consist of weapons, riding equipment, personal objects, clothes, and slaughtered animals deposited in shallow lakes (Fabech 1991b; Randsborg 1995; Ilkjær 2000; Jørgensen et al. 2003; Nørgård Jørgensen 2009). Weapons were deposited in lakes, bogs, and fens in the Late Bronze Age as well as in the early Middle Ages, but the large finds of often deliberately destroyed weapons all belong to the Iron Age, starting with Hjortspring from about 350 BC (Randsborg 1995). Around thirty sites of weapon deposits have been found, most of them in present-day Denmark and southern Sweden, although there are two known sites on Öland and a few in central Sweden (Fig. 30). Thus, the weapon deposits are primarily located in the agricultural districts of southern Scandinavia, which lacked hillforts. A recurrent pattern is that weapons were deposited in the same place on repeated occasions, but often at long intervals. About fifty deposits at the thirty sites are known for the period 200–600, and only three large deposits are known from 350 BC to AD 150 (Fabech 1991b; Kaul 2003; Ilkjær 2003).
Figure 30. Selection of weapons and other objects deposited at Ejsbøl near Haderslev in southern Jutland during the fourth and fifth centuries. (After Ørsnes 1988, table 48.)
Complete weapon deposits, representing war booty of fully equipped armies, have only been found in a few places from about AD 200 to the late fourth century or about AD 400 (Nørgård Jørgensen 2009). One example is Illerup Ådal in central Jutland, where roughly 10,000 weapons were deposited on just one occasion at the start of the third century. The find may represent a defeated army of an estimated two or three thousand men (Ilkjær 2000). Extensive, but not complete, weapon deposits have been discovered in several locations from the first century to about AD 475. Selected smaller deposits comprising mostly weapons can be dated from the beginning of our era to about AD 600. From this time until the twelfth century only single weapons were deposited in water (Lund 2009; Nørgård Jørgensen 2009).
The spectacular war booties from AD 200–350/400 have been interpreted as expressions of wars on a larger scale and a more centralized military organization than before. The early deposits, such as Hjort-spring, contained fewer weapons and more diverse and personal martial equipment, reflecting small bands of individually equipped men (Randsborg 1995). From around 200, on the other hand, weapons became very standardized, which suggests that they were made in a few ‘weapon forges’ and distributed from there to a more professional and trained group of warriors (Ilkjær 2000; Kaul 2003).
It has been possible to reconstruct the organization of these groups of warriors, above all thanks to Jørgen Ilkjær’s detailed analyses of the weapons from Illerup (Ilkjær 1997, 2000; cf. Randsborg 1995). He is able to sketch a military hierarchy consisting of three different levels of warriors – ‘private’ foot soldiers, mounted ‘officers’, and mounted ‘commanders’. The armed warriors (89 per cent) were equipped with mass-produced lances, spears, and shields, while the officers (9 per cent) more often carried individually designed swords, shields, and belts with bronze mountings. The commanders (2 per cent) were equipped with artfully designed swords, shields, belts, and equestrian gear with gilded silver mountings, and some of them also had chain mail and decorative Roman helmets of silver. It is also to this military elite that objects with animal ornamentation and runes can primarily be linked. A similar tripartite army organization was found in the Roman army (Pauli Jensen et al. 2003) and can also be traced in the Pictish war organization in north-eastern Scotland (Alcock 2003:149).
Several scholars have studied the origin of the defeated armies using more personal items such as combs, strike-a-lights, and wooden vessels (Randsborg 1995; Ilkjær 1997, 2000; Pauli Jensen 2009). This allows us to glimpse the outlines of different lines of conflict in the Scandinavia Iron Age. Jørgen Ilkjær asserts that the defeated armies in the second century came from northern Germany, in the third and fourth centuries from northern Germany, western Sweden, and southern Norway, and in the fifth century from Öland, Gotland, and eastern Sweden. Ilkjær believes that the weapons were deposited close to the scene of the fighting, and consequently he sees the origin of the weapons as the origins of military attacks directed against Denmark. This large-scale scenario with a succession of different sources of aggression against Denmark has been criticized, however, because it rests to an excessive extent on some selected Danish finds. If the Swedish weapon deposits had also been included, the picture would presumably have been more complex (Nicklasson 1997:159 ff.; cf. Nordqvist 2007). In addition, there are smaller finds of weapon deposits along the south coast of the Baltic Sea, which could further complicate the picture (cf. Müller-Wille 1999, 2001).
Ilkjær’s analyses are important, but they can be criticized on one decisive point, which means that the entire question of the direction of aggression must be reappraised. He views the repeated deposits of weapons in the same place as a military pattern of recurrent aggression against Denmark. At a conference in 2001, I proposed that the Scandinavian weapon deposits instead should be compared with war booty in Greek, Celtic, and Roman areas, being inspired by Ida Östenberg (from a manuscript later published as Östenberg 2009). These ideas were taken up by some colleagues (Jørgensen 2001; Lund Hansen 2003a), but have also been criticized (Herschend 2003:312 ff.; Fabech 2009). Since I have only briefly touched upon them elsewhere (Andrén 2007a), I will expand the arguments here.
Figure 31. Marble frieze of Roman war booty from the second century AD, found in Ephesus. (Photo by the author.)
In Greek sanctuaries such as Olympia and Delphi, at sacred Celtic sites, and in Roman triumphal processions (Fig. 31), war booty from distant campaigns was assembled (see Baitinger 1999; Cunliffe 1997; Östenberg 2009). This war booty varied considerably according to time and place, but could comprise captured enemies, weapons, personal belongings, parts of ships, treasure from conquered regions and symbols of conquered land, such as trees and animals. Many of these aspects are present in the Scandinavian weapon deposits, such as weapons, personal objects, ships (Ørsnes 1988; Ilkjær 2000; Rieck et al. 1999), treasure in the form of golden arm-rings (Hagberg 1967a, 1967b), and possible symbols of conquered land, such as numerous sticks and branches of trees (Rieck et al. 1999). The recurrent weapon deposits in the lakes of southern Scandinavia must therefore be considered as following a ritual pattern, with clear contemporary continental parallels. Besides, many of the shallow lakes with weapon deposits were used for other ritual deposits, such as pottery, animal bones, and humans, before as well as after the weapons were thrown into the lakes (Pauli Jensen 2009; Monikander 2010). The Scandinavian sites could thus be seen as holy places to which parts of the war booty were brought on different occasions. The location of weapon deposits thus tells us nothing about where the fighting took place. This means that the battles could have been fought anywhere between the sites of the war booty and the place of origin of the defeated armies. The lines of conflict were presumably much more complicated than in Ilkjær’s reconstructions. It is therefore perfectly possible that the densely populated and economically dominant southern Scandinavia in many cases may have been the point of departure for the attacks (cf. Nicklasson 1997:159 ff.), and that battles could have been fought in western Sweden and southern Norway, or on the islands in the Baltic Sea and in eastern Sweden.
Another expression of the war culture during this time is found in the boathouses or naust along the coast of Norway (Myhre 1997), which were probably based on Roman models (Grimm 2001). Around 150 large boathouses from the Iron Age and Middle Ages have been registered, above all in western and northern Norway (Fig. 32). The medieval boathouses are scattered along the coast, often right beside the places that were central in the medieval naval organization of the ledung. The boathouses from the Iron Age can be dated to the period c.150–700, but unlike the medieval naust they are located in several smaller concentrations along the coast. They therefore represent a network that is separate from the medieval naval organization (skeppslag). Bjørn Myhre takes them to be expressions of a less stable naval organization, with boats and warrior retinues connected to changing chieftains or commanders. Along the 250-kilometre-long coast from Lindesnes to Bergen alone, Myhre claims that 90 boathouses, for vessels of 15–30 metres length, may have been used during the period 150–700 (Myhre 1997). The large number of naust means that a very large navy with warriors could be mobilized from just a small part of the Norwegian coast. The boathouses thus represent a new military organization on a large scale, similar to the warrior units behind the contemporary south Scandinavian weapon deposits.
The Norwegian coastal area is also the setting of the particular ringtun or tun sites (Storli 2000, 2001; Stylegar and Grimm 2003). The word tun refers here to a cluster of houses placed around an open space in a circle or oval with one or two openings. In several cases the gables of the houses open out towards the yard (Fig. 33). The layout and design of the houses show several similarities to the buildings in the Ölandic forts, but, unlike these, the tun sites lack walls (see Stenberger 1933:260–1). Around 25 sites are known, from Jæren in the south of Norway to Lofoten in the far north. Like the ringforts, the tun sites are usually placed in the outland between different settlements, although closer to other ancient remains. There are often large cooking pits, fire-places, and large burial mounds. Several tun sites are moreover linked to the place-name element -leik, which means ‘play’ but also ‘training for war’. The finds are few and modest, but the sites were often used on different occasions over a long time. The general time framework for the tun sites is between AD 200 and 850.
Figure 32. Two boathouses at Nes on Karmøy. (After Myhre 1997:170.)
The interpretations of the tun sites are as numerous and varied as those of the Scandinavian ringforts and hillforts. They have been interpreted as forts, barracks, military exercise grounds, amphitheatres, and overnight lodgings for boat crews, but also as villages, sites of legal assemblies, and sacrificial sites (Storli 2000, 2001). Since the end of the 1970s, political interpretations have dominated, especially when it comes to the north Norwegian tun sites (see Johansen and Søbstad 1978). In the most recent scholarly treatment, they are regarded as places of assembly for the warrior retinues of the aristocracy (Storli 2000, 2001; Stylegar and Grimm 2003). Since the sites are located in outland, Inger Storli interprets them as neutral meeting places, where the landowning elite could gather on equal terms. She views them in connection with drinking ceremonies, which were necessary for creating a sense of community in a band of warriors not held together by bonds of kinship. The tun sites were abandoned when more supra-regional leadership was established, in Rogaland in the sixth and seventh centuries, and in north Norway at the start of the ninth century. The age of the ringtun and its probable military functions show once again that the time around 200 was an important turning point for the Scandinavian culture of war.
Figure 33. Ringtun on Gimsøy and at Åse on Lofoten. (After Storli 2001:94.)
Studies of Scandinavian martial culture have been characterized by national divisions in the debate, which follow modern political boundaries and archaeological specialities. The weapon deposits have above all been studied in Danish archaeology, the forts have chiefly been a Swedish, Norwegian, and Finnish concern, while the boathouses and tun sites have been discussed almost exclusively in Norwegian archaeology. Like several other scholars, I believe that these martial phenomena should be viewed together, since they are all expressions of the same war culture in Scandinavia during the period 200–600/700 (Näsman 1988, 1997; see Nicklasson 1997:170). In concrete terms, the weapon deposits, forts, boathouses, and tun sites can be regarded as different aspects of the same conflicts in Scandinavia (Fig. 10). The forts in Sweden and Norway as well as the Norwegian boathouses and tun sites were the direct origin of some of the defeated armies found in the Danish weapon deposits (cf. Engström 1991:275; Stylegar and Grimm 2003).
War as ritualized action
The ringforts and hillforts, the weapon deposits, and the boathouses as well as the tun sites underline that warfare was an important and constant part of Iron Age Scandinavian society, above all during the period 200–600/700. Therefore, it interesting that in one of the few contemporary descriptions of Scandinavia, written by the Ostrogoth historian Jordanes c.550, recurrent wars are emphasized. He writes in his Getica that five peoples in southern Scandinavia living in the fertile plains ‘are disturbed there by the attacks of other tribes’, that four tribes are of ‘a race of men bold and quick to fight’, and that three tribes ‘live like wild animals in rocks hewn out like castles (castelli)’. The term castelli has been interpreted by some scholars as a reference to the contemporary forts (Lund 1993:281 ff.).
The character of this warfare is not described in any detail. However, analogies with Britain and Ireland after the Roman period can show how war was a constant feature in society, but also that different forms of warfare existed. Based on anthropological studies of war (for example Haas 1990), Guy Halsall has examined statements about war in the Anglo-Saxon Chronicle and other historical works (Halsall 1989; 2003). He shows that war and violence were endemic in Anglo-Saxon England (450–1050), involving everything from private blood vengeance to large-scale war. In the same way, Leslie Alcock has studied war in the Celtic parts of Britain and Ireland, above all in Scotland (Alcock 2003:119 ff.). In both areas it is possible to see a clear distinction between ritualized war and large-scale war.
Ritualized warfare involved small-scale acts of war, such as annual plundering campaigns and raids conducted by aristocratic groups. Through these martial acts, political leaders could gain access to prestige goods such as slaves, cattle, and horses. On the other hand, the conflicts did not change the basic balance of power. In ritualized war, there were clear norms, which meant that the combatants could jointly choose the place of battle, that parts of the plunder could be returned, and that the fighting could be replaced by tribute and concluded with an exchange of hostages, the swearing of oaths, and the establishment of ties of friendship. Generally speaking, the parties were expected to fight fairly, based on a kind of warrior honour (Halsall 1989). In Ireland, Scotland, and Wales it was part of the initiation of a new king that he and his followers should mount a cattle raid on a neighbouring kingdom. In early medieval Welsh laws, it was furthermore regulated that a king was allowed to plunder outside his kingdom once a year for a maximum of six weeks (Alcock 2003:119 ff.)
The large-scale wars, on the other hand, lacked the code of honour involved in ritualized combat. The major wars were big and bloody, and could mean that entire population groups were killed and replaced by others. The conflicts could change the balance of power, by shifting the boundaries of political units and allowing the rule over an area to be wholly or partly taken over by another king. In the whole Anglo-Saxon area, large-scale wars are mentioned roughly every three years, but for each individual kingdom a large-scale conflict occurred roughly every twenty years (Halsall 1989). In Scotland, major campaigns and battles occurred at roughly similar twenty-year intervals in the period 570–870 (Alcock 2003:131 ff., 180–1).
Ever-present war meant that there was a warrior culture, as is clearly expressed in Old English and Irish literature. This warrior culture is evidenced in the poems of praise to warriors and their weapons, and also in descriptions of how quarrels between warriors could suddenly arise. Otherwise, it was a tendency in Britain and Ireland that the kings gradually tried to restrict small-scale ritualized war, such as feuding and plundering, through legislation and tributes.
Despite the constant presence of war, there are no finds of weapon deposits and virtually no forts in Anglo-Saxon England during the period. In contrast, the Celtic regions have many forts. In Scotland alone more than a hundred forts have been dated to the period 200–850 (Alcock 2003:179 ff.), that is, exactly the same period as many of the Scandinavian forts. In Britain there is thus the same contrast as in Scandinavia, between flat plains without forts and more forested and mountainous areas with forts. This shows that Scandinavian weapon deposits, forts, ringtun, and boathouses were not necessary expressions of war, but were used as markers in special contexts. They may be viewed as special manifestations of the warrior culture of the time. They thus constitute a kind of social and ritual filter, through which war can be detected. In the plains of southern Scandinavia, with large and densely populated villages, war was expressed through very large weapon deposits in a few select places. In the more forested and mountainous areas in the rest of Scandinavia, with single farms and more loosely organized villages, war was instead expressed through many forts located within short distances.
As in Britain and Ireland and in many other parts of the world (Otto et al. 2006), the actual warfare in Iron Age Scandinavia may have been large-scale wars as well as ritualized wars. Indications of large-scale wars are above all the very large weapon deposits from AD 200–400. Ritualized warfare may have been expressed in the smaller selected weapon deposits from the first century to around 600. Ringforts and hillforts as well as boathouses and tun sites could possibly be connected to both forms of warfare.
In the much later Icelandic literary tradition, there are several echoes of ritualized warfare in Scandinavian society (Price 2002; Nordberg 2003). Hints of predetermined and marked battlefields are found in the Poetic Edda, for example in references to small islands or fields marked off with hazel rods. In several skaldic poems and Icelandic sagas there are mentions of berserks (berserkir, meaning ‘bear shirts’) and ulfhednar (ulfheðnar, meaning ‘wolf skins’) as groups of specially initiated warriors dressed in animal skins (Simek 1993:35, 338–9; Lindow 2001:75–6; Näsström 2006; Hedeager 2011). Exactly what the terms berserks and ulfhednar meant is a matter of controversy (Simek 1993:35; Näsström 2006), but most scholars agree that they designated some kind of ecstatic warrior role associated with the war god Odin. The terms also show how war was associated with ‘the wild’ and especially with savage and feared beasts of prey. The earliest reference is found in the skaldic poem Haraldskvæði, composed in the late ninth century. However, warriors like these are depicted as early as around 600, on matrices for making helmets, at Torslunda on Öland (Fig. 34). Remains of bearskins in graves are also known from the whole of Scandinavia in the period 150 BC – AD 1050, although they appear in female as well as male graves (Petré 1980; Ström 1980; Nicklasson 1997; Hedeager 1997, 2004, 2011).
Figure 34. Bronze matrices for producing helmet plates, dated to c.600, found at Björnhovda in Torslunda on Öland. The two matrices illustrate a warrior between two bears and a ‘wolf warrior’ together with a one-eyed warrior with a helmet (Odin?). Both images call to mind the concepts berserkir and ulfheðnar. (Drawings by Bengt Händel in Sandwall 1980:24–5.)
Other expressions of the wildness of war are hints from the much later Icelandic literary traditions that initiation rites for warriors comprised meals of raw meat from wild animals (Schjødt 1999b; Näsström 2001a; Sundqvist and Hultgård 2004). The acts of war themselves were also perceived as wild, since slain enemies in skaldic poetry are conventionally described as feeding wolves, eagles, and ravens. The wildness of war is moreover evident from many of the martial names for men, based on the names of wild animals -arn (‘eagle’), -björn (‘bear’), -ormr (‘serpent’), -svín (‘wild boar’), and -úlfr (‘wolf’). Several scholars have pointed out the striking similarity between these men’s names and contemporary animal ornamentation, depicting eagles, serpents, wild boar, and wolves (Werner 1963, 1966; Müller 1968; Beck 1986; Hedeager 1997, 2004, 2011; Nordberg 2003; Domeij Lundborg 2006). However, the personal names based on wild animals are older than the animal art, since a name such as Widuhundar (‘wood hound’, probably a metaphor for ‘wolf’) is attested in a runic inscription from Himlingøje on Sjælland as early as the early third century (Krause and Jankuhn 1966:i. 32–3). The berserks and ulfhednar, together with the men’s names associated with wildness, have been linked by scholars to war bands as well as some form of warrior shamanism (Erä-Esko 1965; Müller 1968; Haseloff 1981; Roth 1986; Hedeager 1997, 2004; Gaimster 1998:48 ff.; Price 2002; Sundqvist and Hultgård 2004).
The violence of war and the associations with wildness may explain why the objects in the war booty were so deliberately destroyed. The particular deposits in the shallow lakes can be regarded metaphorically as traces of ravenous beasts of prey that ate up the booty, leaving only tattered remains. In the same way, the placing of the many ringforts and hillforts in the outland between settlements can be perceived as wild places appropriate for the equally wild wars. The forts were thus mentally in the same wild landscape as the men in bear shirts and wolf skins, and the men bearing names of beasts of prey.
War and cosmology
Judging from the late Icelandic texts, warfare in Scandinavia above all seems to have had mythological associations with Odin (Hedeager 1997, 2004, 2011; Schjødt 1999b; Price 2002; Nordberg 2003). However, the recurrent figure of nine at Ismantorp, Eketorp, and Gråborg points towards more general connections between war, rituals, and cosmology. The same numerical symbolism is found in graves, artefacts, and ritual sites from at least the Pre-Roman Iron Age to the Viking Age, that is, from about 500 BC to AD 1050 (see for instance, Äijä 1999; Rundkvist 1996, 2001; Svensson 2012).
The clearest parallel to the Ölandic ringforts are circles of standing stones (Sw. domarring) that marked cremation graves from about 500 BC to about AD 550 (Fig. 35; Jaanusson 1966; Lundström 1976; Leube 1979; Hyenstrand 1984:73 ff.). Stone circles are known in southern and eastern Norway, in southern and central Sweden, on Öland, Gotland, and Bornholm, and along the south coast of the Baltic Sea in present-day Poland and Mecklenburg-Vorpommern. They normally consist of an uneven number of stones, mostly five, seven, or nine stones, but the number varies regionally. In most provinces of Sweden, the most frequent number is seven stones, although circles with nine stones are common in all areas. Stone circles with nine stones, however, predominate on Öland and in Skåne as well as in Mecklenburg-Vorpommern (Leube 1979; Hyenstrand 1984:73 ff.).
Figure 35. Circle of nine standing stones at Norrby in Hallsberg in Närke. Apart from the usual number of nine stones, this circle is also constructed in relation to cardinal points. The tallest stones are erected in the north, whereas the shortest ones are placed in the south. (Photo by the author.)
On Öland, all excavated stone circles are dated to the period AD 1–550 (Arne 1938). They are mostly built up of round boulders, lying directly on the ground or placed on three or four small underlying stones. In cases where the stone circles have a central stone, on the other hand, it is a standing stone (Arne 1938). Kerbs linking the stones in the circles are mentioned in earlier antiquarian sources, but today they are not preserved or visible at any site. The form of these stone circles shows great structural similarities to the Ismantorp fort. The central pole, the nine gates, and the ring-wall at Ismantorp are paralleled by the central stone, the nine boulders on their underliers, and the kerbs of the Ölandic stone circles. Even the houses in Ismantorp can be said to correspond to the area in the stone circles where charcoal and burnt bones of humans and animals were scattered inside the ring of stones. Since the stone circles are widely distributed chronologically and geographically the recurrent figure of nine at Ismantorp, Gråborg, and Eketorp must refer to a spatial design that is otherwise mainly known from grave contexts. In view of the placing of burnt human bones in stone circles, this depiction must be connected to ideas about humans and their place in the world (see Bergström 1979; Andrén 2004).
Another parallel to the nine gates at Ismantorp as well as the stone circles is knobbed or ribbed bronze rings (Rundkvist 1996, 2001). These bronze rings are often decorated with nine knobs or ribs. In present-day Sweden, about sixty rings are known in regions facing the Baltic Sea from the late Pre-Roman Iron Age to the Merovingian Period (about 150 BC – AD 750). However, more than half of the rings come from Öland. They are dated to the Migration Period (400–550) and all except four are furnished with nine or eighteen knobs or ribs. One ring with nine ribs was actually found in the ringfort of Lenstad.
Martin Rundkvist, who has investigated the rings, underlines the similarities to the stone circles. Judging by graves in central Europe, this kind of ring was worn by women, hanging in their clothes. Given the recurring number nine and the connection with women, Rundkvist interprets the rings as amulets used in connection with pregnancy (Rundkvist 1996). However, a large neck-ring of similar design has been found somewhere in the parish of Torslunda on Öland. This bronze ring is furnished with nine knobs decorated with silver inlay, and has been dated to the Migration Period as well (Holmqvist 1951:116). Therefore, the bronze rings may have had more general significance connected to the number nine.
We can glimpse how this world was perceived from the echoes in the Icelandic literary tradition, where the figure of nine was a recurrent unit for time, place, and transformation (Simek 1993:232–3; Gansum 1999:459 ff.). According to Vǫluspá 2, the ‘measuring tree’ was surrounded by nine worlds or preceded by nine ages (see Schøjdt 1992); Odin hung for nine nights in the world tree (Hávamál 138); the god Heimdall was the son of nine mothers (Hyndluljóð 35), who probably correspond to the nine daughters of the sea god Ægir; from Odin’s or Baldr’s gold ring Draupnir there fell eight rings every nine nights (Gylfaginning 48; Skáldskaparmál 33); and at the end of the world (Ragnarok), Thor, after his fight with the Midgard Serpent, managed to take nine steps before he fell down dead (Vǫluspá 56, Gylfaginning 50).
The figure of nine also occurred in rituals. Thietmar of Merseburg and Adam of Bremen in the eleventh century mention that large sacrificial feasts were held every nine years for nine days at Lejre and Gamla Uppsala (Nordberg 2006). At Gamla Uppsala, nine men and nine male animals of seven different kinds of animals were sacrificed (Sundqvist 2002, 2004). These descriptions have been questioned, but they can partly be confirmed by a much earlier rune stone from Stentoften in Blekinge, where the figure of nine is associated with animal sacrifices. This text from the seventh century states that ‘with nine bucks, with nine stallions, Haþuwulf gave a good harvest’ (Santesson 1989; Sundqvist 2002). The figure of nine can thus be seen as an expression of a kind of cosmological and ritual perfection, and connected to transformations from one stage to another (Brink 2004; Sundqvist 2010).
The structural similarities between stone circles, Ismantorp, and the mythical and ritual figure of nine means that the graves and the fort may be regarded as images of Old Norse cosmology. The stone circles and Ismantorp were thus concrete symbols in the same way as the three-pointed stone settings that I discussed in the previous chapter, the difference being that the three-pointed stone settings signified only the world tree while the stone circles and the fort expressed the worlds around the world tree. For Ismantorp this cosmological interpretation means that the post in the centre of the fort and the nine gates corresponded to the world tree at the centre of the world, surrounded by nine different worlds (see Sjöborg 1822:128).
The cosmic dimensions at Ismantorp are very clear, but it is perfectly possible that simpler forts likewise display associations with the Old Norse worldview, as Birgitta Johansen (1997) has suggested. The world was called Midgard, and according to Grímnismál 41 the gods made Midgard from the eyelashes of the primeval giant Ymir, and according to Snorri’s Gylfaginning 7 Midgard was a fortress. Ismantorp, like many other Scandinavian forts, thus had cosmic associations. Yet this connection between war and cosmos is not confined to the Norse worldview; it is also found outside Scandinavia.
Models and parallels
In view of the dating of Ismantorp and the other forts on Öland, it is natural to turn to the Roman Empire and the Roman army in search of models. Stenberger suggested early on that the distinctive Ölandic forts may have been imitations of Roman models (Stenberger 1933:255 ff.). Joachim Werner believes instead that the forts on Öland were inspired by late classical Byzantine fortified villages and strongholds (Werner 1949:267–8.), while Frands Herschend and Ulf Näsman draw attention to Roman models for a special portcullis gate at Eketorp (Herschend 1985; Näsman 1989). Some forts in the Mälaren valley are also believed to have been built in imitation of Roman models (Olausson 1997, 2009).
The significance of the Roman Empire in northern Europe has long been noticed in Scandinavian archaeology, yet recent studies show that the Roman finds do not reflect simple ‘imports’ or passive influences from the empire (Hedeager 1979; Lund Hansen 1987; Ekengren 2009). Instead, the Roman objects were used actively as markers by socially and politically dominant groups in Scandinavia. Beyond the artefacts, Roman links can be detected in a number of different spheres, not least in the culture of war.
Around 300, more than a hundred years after the Marcomannic Wars (166–72, 177–80), the Roman warfare and defensive strategy had changed (Johnson 1983; Le Bohec 1994; Southern and Dixon 1996; Bachrach 1997; Schnurbein 1997; Fischer 1999:138 ff., 2000:207 ff.; Mackensen 2000:213 ff.; Schmidts 2000:219 ff.). Before the wars, the defence of the empire was chiefly based on a fortified border with ramparts, forts, army camps, and large garrisons, planned to fixed principles. After the wars, the Romans instead gradually created a deep defence with forts and fortified towns in a wide border zone. Over time, the fortifications were increasingly adapted to the terrain and thus became irregular. At the same time, warfare changed from battles in open land to siege warfare around fortified sites. The changes in the Roman army and its warfare can also be detected in the structure of the Roman army camps. Most of the geometric rectangular Roman fortresses had disappeared by 300, although some forts of rectangular shape were still built during the fourth century, such as Deutz, Boppard and Alzey in the Rhineland and Fenenpuszta in Hungary (Fischer 1999:138 ff.; Oldenstein 2009). Instead, many smaller forts were built, with high walls and towers. These forts could be square, trapezoid, round, or oval, but they could also be totally irregular if they had to be adapted to the terrain.
The classical Roman camps or fortresses were built as large squares or rectangles with slightly rounded corners (Fischer 1999:48 ff.). They were divided into four parts by two main streets, cardo and decumanus. In the middle of the camp there was normally a principia, or headquarters building, where standards, ensigns, weapons, and the regimental treasury were kept. Grouped around the headquarters were the commander’s residence, barracks, stables, storehouses, field hospital, kitchens, and latrines. The fortresses were built according to the same principles throughout the empire, thereby creating a common spatial framework for the multicultural Roman army (see Schnurbein 1997). Unfortunately, little is known about the design of the late classical fort, and the spatial principles are unclear in most cases. However, at the rectangular Deutz the caserns were still built in rows, whereas the barracks were built along the walls in Alzey (Johnson 1983; Oldenstein 2009). This place later became a political centre in the Burgundian realm in the fifth century (Fig. 36). Alzey as well as many of the other late Roman forts can be regarded as models for Ismantorp and the other ringforts on Öland. It is not necessary to search for a single model, since the Ölandic ringforts should rather be viewed as local reinterpretations of the contemporary Roman forts.
Figure 36. The late Roman fortress of Alteium (modern Alzey) between Mainz and Worms. The fortress was established in 367–70, and is one of several late Roman rectangular fortresses in the Rhineland. It was used by the Romans for only about thirty years, then rebuilt twice before it fell out of use in the late fifth century. (Plan by the author, based on Oldenstein 2009:268.)
The meaning of the distinctive design of the classical rectangular Roman fortresses has been studied in detail by Joseph Rykwert (1988). Based on Latin texts describing the ceremonies for the foundation of towns and garrisons, he is able to show that the places expressed the Roman cosmology, with the division of the heavens into four parts, and the connection between heaven, earth, and the underworld. The places were moreover perceived as images of Rome, even though the capital did not have a regular plan. The role of representation is particularly clear from the foundation ceremonies, which everywhere in the empire were based on the myth of Romulus founding Rome. The rituals included taking omens from flying birds, purifying the area by fire, and marking the centre of the world with the aid of a pit, which also symbolized the opening into the underworld. From the pit, garrisons were planned and simultaneously demarcated from the outside world by a furrow ploughed with a team of oxen. The foundation pit in the centre has actually been found a few times, for example at the legionary fortresses at Nijmegen in the Netherlands and at Inchtuchil in Scotland (Pitts and St Joseph 1985:59, 78). The classical Roman camps, in other words, expressed how the military order was incorporated in a larger cosmological model, which applied all over the empire. The possible meaning of the late Roman forts is not known to the same extent. However, due to the close link between war and religion in the Roman Empire (Rüpke 1990; Gustafsson 1999), as well as the foundation of some new rectangular fortresses in the fourth century, it is highly likely that the older cosmological order survived in some form.
This association between cosmology, fortification, and military organization was not confined to the Roman Empire. The best-known example after the Roman Empire comes from eighth-century Saxony, as mentioned in the previous chapter (Palm 1948; Best et al. 1999; Vikstrand 2001). In 772, the Carolingian army conquered the Saxon fort of Eresburg, where according to several different sources there was a large wooden pillar in the fort, which Charlemagne had chopped down and replaced with a church. The pillar, which was called Irminsul (‘the great column’), should be regarded as a representation of a world pillar at the centre of the world. The Carolingian church no longer exists, having been replaced by a later church built at the highest point of what is today called Obermarsberg. The location of the church shows that Irminsul must have been placed at the highest point on the plateau where the fort stood. Eresburg was thus a heavily fortified stronghold but at the same time a representation of the world, with the world pillar erected at the highest part of the fort (see Fig. 11).
It is possible that the Slavic strongholds from the eleventh and twelfth centuries, which Stenberger used in order to interpret Ismantorp, may be viewed in a similar context. The Slavic forts were more permanently occupied than the earlier Scandinavian forts, and their central ritual functions were more pronounced (Słupecki 1994). At the same time, they were powerful strongholds with obvious cosmological associations. As we have seen, the fort of Rethra had nine gates, and, as at Ismantorp, these can be interpreted as symbolizing nine worlds. A narrow gate in the east led out to a nearby lake ‘which looked truly terrifying’ (visu nimis horribile monstrat, Thietmar of Merseburg VI, 23, in Słupecki 1994:52). In view of this wording, perhaps this gate can be perceived as a special ‘death gate’. The association between organization of war and cosmology is also clear from other forts, which Stenberger does not mention. In the main Pomeranian stronghold at Szczecin (Stettin) there is mention of a temple but also a large oak and a spring, which can be connected with the idea of a world tree with a spring at its root. And at the centre of the main stronghold at Pskov in western Russia, the remains of a large post have also been found, arousing associations with a world pillar (Słupecki 1994:72 ff., 152). Thus the association between cosmology and the organization of war reflected a general pattern in the Roman as well as in the post-Roman world.
Back to Ismantorp
The background to the ringfort must be sought in the clear changes in Scandinavian martial culture that took place from the third century onwards. Before this, armies had consisted of smaller groups summoning individually equipped men, but from around this time the armies became bigger and more hierarchically organized, with mounted commanders. This military elite controlled the mass production of weapons and probably also the more professional training of soldiers. Like other contemporary changes, the new military organization must have been based on Roman models. Similar to other Roman elements in Scandinavia, it was not a straightforward adoption, but instead a process of conscious selection and reinterpretation (cf. Ingemark 2003). Aspects of a Roman martial culture were probably introduced by Scandinavian soldiers who had served in the Roman army. But the new military order could scarcely have been established without local negotiations. The location of the forts in outlands between different settlement areas indicates that they were built on a collective initiative, and that the army leaders thus had only temporary power during the actual campaigns. The large finds of weapon deposits in similar ways reflected collective rituals after martial campaigns.
Viewed from this perspective, it is possible to understand the meaning and function of Ismantorp as one of the distinctive Ölandic ringforts used between 300 and 650. In Ismantorp it is possible to see how Roman models were incorporated and reinterpreted in a Scandinavian context around 300 (Fig. 37). The foundation ceremonies for Roman towns and camps were partly adopted at Ismantorp. The rituals included determining the centre of the world by digging a pit, which simultaneously marked the opening into the underworld. From this pit the towns and garrisons were laid out, after which the pit was filled and covered. In the middle of the open area in Ismantorp there was a corresponding pit, which was the geometrical centre of the fort, since the nine gates were oriented towards it. Unlike the Roman foundation pits, however, this pit was open for a longer period.
From their foundation ceremonies, the Romans planned square or rectangular towns and camps with two crossing axes. The builders of Ismantorp instead constructed a virtually round fort with nine gates. The Roman cosmology, with a quadripartite sky, was thus translated to an Old Norse worldview with nine different worlds surrounding its centre, each world being represented by one gate. The local model for this design was the older and contemporary stone circle with nine stones. As mentioned earlier, one of the small gates was connected to a house. This strange construction can be viewed in cosmic perspectives as the way to a realm of the dead, either Hel or Valhall, which both surrounded the world tree. The gate could thus have served as a Helvegr, a special road used only to carry people to the grave (de Vries 1956:194).
In the middle of the Roman camps there was normally a headquarters building, the courtyard of which was placed over the original foundation pit. At Ismantorp the pit was retained in the open area at the centre, evoking associations with the well of fate or Urd’s well, which according to Vǫluspá 20, and Gylfaginning 14 was located in the middle of the world beside the world tree (Simek 1993:236–7, 342). The post at the eastern end of the open area can be regarded as the representation of the world tree or world axis, whereas the small semicircular building may have staged the dwelling of the Norns, which according to Gylfaginning 14 was located beside the world tree. In the small house a seeress or a vǫlva may have been active, evoking associations with the mythical Norns. Several Roman sources mention seeresses as important to Germanic troops already in the first century AD (Simek 1993:356–7; Dobat 2009), and the exquisite relief brooches in two of the Ölandic ringforts have can be interpreted as marking the presence of high-ranking women with ritual capacities (Magnus 2004). No traces of large-scale rituals have been found in the centre of the fort, but the pit, along with the small building and the post, possibly provided the framework for other rites, such as the initiation of warriors and divinations before campaigns. The lancehead and the arrowhead deposited in small pits in the centre could thus be traces of such rites.
Figure 37. Reconstruction of Ismantorp ringfort, with possible cosmological references: the nine gates (red) and their orientation towards the pit in the centre (red lines), the house in front of the south-east gate (orange), the semicircular house in the centre (violet), and the post east of the house (green).(Plan by the author.)
In the Roman camps, all the other buildings necessary for the army were constructed around the headquarters. The buildings were mostly grouped according to their various functions, so that the stables were grouped in one place, the barracks in another, and so on. At Ismantorp there were likewise different types of houses around the centre of the fort, but they were not separated according to function in the same way as the buildings in the Roman camps. Some of the buildings at Ismantorp, like the buildings in the Roman army camps, must have been intended for the followers, who may also have included women.
Many of the Roman camps became permanently occupied, and moreover attached to a less regulated civil settlement outside the fortifications. Ismantorp and most other Scandinavian forts from this period lacked this feature. Most of them were built in the outland far from cultivable land, so they lacked the conditions necessary for becoming permanent settlements. In addition, they were mentally speaking in a wild or liminal landscape, which suited the wildness of the Scandinavian martial culture. Ismantorp, like many other forts, must therefore be regarded as a kind of military resource, built by the estates in the surrounding settlements.
The exact function of Ismantorp is difficult to determine, due to the general lack of finds. However, various martial functions are the most plausible. The place could have been used periodically for training warriors and for the initiation of new warriors. It could also have been the scene of feasting and drinking ceremonies intended to reinforce the bonds between warriors from different settlements, in the same way as the Norwegian tun sites functioned. The ringfort could have served as a place of assembly for external campaigns or as a base for defence. It may have been used in full-scale wars, but was probably used mainly for recurrent plundering raids and ritualized wars. It is therefore possible that livestock, horses, and slaves from raids could have been gathered and divided in places like Ismantorp. The impressive monument moreover signalled that the surrounding settlement and their estates were well organized in military terms. The fort could thus have had an outright deterrent purpose.
Since the unique ringforts on Öland are located at fairly regularly intervals (see Fig. 26), several scholars have tried to reconstruct large-scale military divisions of the island (Hofrén 1948; Näsman 1997). Such ideas are quite plausible, due to the parallels between the size of the forts and the size of the surrounding settlement (Fallgren 2008). However, these divisions should not be viewed in too territorial a perspective, by using medieval military divisions as analogies (see Näsman 1997). Therefore, the forts should also be regarded as potential military resources, built and used in relation to the changing power positions of the estates in the surrounding area, as an analogy to the clusters of early boathouses along the Norwegian coast. The ringforts need not have been used at the same time; they could have been activated according to the prevailing political situation within an area. Such non-centralized initiatives could thus explain the low level of activity in Ismantorp as well as the short distance of only five kilometres between Ismantorp and Mossberga.
The ringfort at Ismantorp can be regarded as a hybrid form, with Roman elements incorporated and reinterpreted in a new context. This new creation was and remained unique to Öland, since the island’s ringforts took on a highly distinctive character. There may be several reasons for this. The Roman-influenced hybridization of southern and central Scandinavia was not a uniform process, since circumstances differed in this politically and socially divided area. The different Roman phenomena should instead be regarded as individually adapted strategies by different local elites to demarcate themselves vis-à-vis the rest of the population, through Roman-inspired traditions and expressions. The Roman features could thus have taken on a highly local or regional character, and been given a wide variety of expressions. Densely settled ringforts were constructed only on Öland, boathouses and tun sites only along the Norwegian coasts, while Sjælland is particularly known for burials with rich Roman finds.
The special design of the Ölandic ringforts may also have to do with the fact that the island was strategically situated on the main sailing route along the eastern coast of the Scandinavian peninsula. The island represented a border zone between, on the one hand, a south Scandinavian area with large villages and weapon deposits, and, on the other hand, an east Scandinavian region with single farms, small villages, and hillforts. On Öland, there were both weapon deposits and ring-forts, but also large, loosely organized villages. This border location, which was also expressed through the styles of contemporary dress ornaments (Høilund Nielsen 1991), may have had the result that the island was heavily militarized (Näsman 1997; cf. James 1997). Perhaps part of Öland’s economy was based on the fact that the island did not just produce cereals and livestock, but also soldiers. In view of Öland’s strategic location, it is also possible that the densely settled forts, like the Roman camps, contributed to the creation of a common cultural framework for the Ölandic warrior groups, even if the individual warriors came from different cultural backgrounds (see Andrén 2011c).
Ismantorp and the other distinctive Ölandic forts were built in the fourth and fifth centuries, but they successively lost their primary functions in the sixth or early seventh centuries; the same can be said of many of the other Scandinavian forts. During the same period, collective weapon deposits were replaced by the depositions of single weapons (Nørgård Jørgensen 2009). The martial organization introduced in the third century was based on negotiations between different aristocratic families, and the military leaders probably only enjoyed temporary power associated with the actual campaign. From the fifth century, however, it seems that some of these war leaders gradually extended their power in a more lasting way. Signs of this shift in power are that a few forts became more permanently occupied in the fifth and sixth centuries, for example, Eketorp on Öland and Runsa in Uppland (Wall 2003; Olausson 1997, 2011). In this way, the power that the war leaders had only exerted during certain short periods became permanent. In the fifth and sixth centuries, the concentration of power thus led to a new form of leadership (Skre 1998; Näsman 2008; Olausson 2009). With this more permanent political elite, the military organization was to a higher degree based on retinues of warriors attached to a leader by oaths of personal loyalty (Nørgård Jørgensen 1997). At the same time, power and control over warfare was strengthened in the same way as in England and Scotland, so that the recurrent, chiefly ritualized raids gradually disappeared. In the long run, there was thus no longer a need for weapon deposits, forts, or tun sites where power relations between different aristocratic groups could be negotiated and renegotiated in more or less ritualized acts.
Ismantorp and Old Norse cosmology
Ismantorp was neither a place of passive refuge nor a central ritual site. Instead, it functioned as an occasionally used army camp inscribed with a cosmological meaning. The fort is a good illustration of how pre-Christian religion in Scandinavia was not necessarily a separate religious sphere but an integral part of people’s practice. This also applies to such a special world as the martial culture of the time. As cosmological representations, Ismantorp and the contemporary stone circles are the oldest known depictions of the Old Norse worldview, as it is known in later texts. Cosmologies need not necessarily correspond to systematic and well-ordered figures of thought, but general ideas about the world can become a clearly structured universe in special circumstances. I believe that this is what happened at Ismantorp and the other Ölandic ringforts. In the encounter with Roman culture and its distinct worldview, Old Norse cosmology became more explicit and more systematic than before.
Earlier hillforts may very well have had cosmological associations. A good example is the hillfort at Odensala in Uppland, where different deposits around the walls seem to have expressed an anthropocentric worldview. In the centre of the hillfort were only deposits of human bones, whereas animal bones were deposited further away from the centre, and cereals were deposited on the periphery (Olausson 1995). However, it was only in the encounter with Roman culture that such a worldview was systematized into something partly new. Ismantorp was a hybrid based both on Roman models and on local traditions in the form of stone circles. In the Roman encounter, ideas about the world that were expressed in burial contexts were used for the martial settlement of the Ölandic ringforts. This encounter between Roman and Scandinavian culture engendered some of the features that would later characterize the Old Norse tradition in the Icelandic texts.
Figure 38. Early picture stone in the fabric of the church at Bro. This part of the church was built c.1300. (Photo by the author.)