01 november 2007

Scandinavian armies in the Late Roman period

The types of weapons used by Germanic armies in the Late Roman period are well-known and well studied. Thanks to the excavations and subsequent publications of Illerup, Ejsbøl and other war-booty offerings in Denmark, we now also have a rather clear picture of functional aspects and internal organisation of Germanic armies. What is still lacking are more regional studies dealing with questions that the recent analyses of the bog-deposits have enabled us to ask, but which this material cannot itself provide answers to. An important question seems to me how the contemporary armies were recruited. While I don’t believe that there is necessarily only one answer to this question, the present study of Late Roman period weapon graves from South-East Norway suggests that in this area at least there is an interesting topographical

aspect to the distribution of weapon combinations and types indicative of the different levels of military hierarchy, as these can be reconstructed from the war-booty offerings.

Illerup place ’A’
It has been stated that ’a material of great value for the understanding of Roman period history in Norway, is now amassed in museum collections in Denmark as a consequence of the excavations in Vimose and Illerup’ (von Carnap-Bornheim & Ilkjær 1999:138-9). The background for this statement is Jørgen Ilkjær’s work with the Illerup material, and his conclusion that the material from the Illerup ’A’ offering, dating to C1b, originated from the Western part of the Scandinavian peninsula (von Carnap-Bornheim 1992; Ilkjær 1990, 1993, 1997, 2003; Ilkjær & von Carnap-Bornheim 1999).

Ilkjær and von Carnap-Bornheim were able to differentiate between three qualitatively different combinations of weapon equipment, and thus three different levels of hierarchy in the beseiged army, whose weapons and personal equipment (12.000 artefacts) had subsequently been deposited in the then lake at Illerup The first and most prominent level (level 1), the army commanders, had swords, shields, belts, and riding gear decorated with mounts made of gilded silver. The second level (level 2) consisted of officers with swords, shields, belts, and riding gear decorated with bronze mounts, while the third level (level 3), the regulars or infantry, had a combination of weapons different from the officers, most often a spear, a javelin, and a shield with mounts primarily made of iron. In Illerup ’A’, of which only c. 40 percent has been excavated, 5-6 sets of weapons associated with the uppermost level were found, against 35-40 associated with the middle level, and c. 350 with the lower (Ilkjær 2003:50). Furthermore, about a dozen sets of horse equipment were rescued (Ilkjær 2000:102), indicating an identical number of equestrians. From the somewhat younger Ejsbøl find (c. 300 AD) were rescued c. 200 sets of the combination lance/javelin/shield, some of which also included bow and arrow. 60 weapon sets also had swords and (weapon) knives. 12-15 sets stood out by having more exclusively worked sword grips, strap buckles and mounts, and 9 sets included horse equipment as well (Jensen 2003:571).

Ilkjær and von Carnap-Bornheim also suggested that there is a degree of correspondance between the internal organisation of the Illerup army and information supplied by Tacitus in the 1st century that the Germanic armies consisted of princeps (princes), comites (officers or retinue), and pedites (foot-soldiers). This tripartite organisation is also mentioned by Ammianus Marcellinus in the 4th century, thus suggesting that what we are dealing with is a long-standing structural feature of Germanic armies. The very homogeneity among parts of the Illerup ’A’ find, especially the lances, javelins and shields – that is, in the equipment belonging to the bulk of the beaten army – could indicate that these weapons were handed out by the army leaders from armouries controlled by them (see, for instance, von Carnap-Bornheim 1993; Ilkjær & von Carnap-Bornheim 1999:144).

Warriors and soldiers
The numerical relationship between the three levels of hierarchy in the Illerup find is interesting. There is a 1:8 proportion not only between levels 1 and 2, but also between levels 2 and 3. The latter relationship compares interestingly with the one between officers and foot-soldiers in the Roman army, where the smallest unit was a contubernium, a group of eight men led by a decurion and sharing a tent.

In a more recent contribution, Xenia Pauli Jensen et al. (2003) have discussed the Illerup find and the Illerup army from a more anthropologically informed perspective. Drawing upon a distinction argued by the anthropologist Andrew Sanders, these authors seek to distinguish between warriors, whose vocation is war, soldiers, whose profession is war, and conscripts, who are ordered into war. Typical for warriors are personally acquired weapons and localised weapon production. Soldiers and conscripts, on the other side, have standardised weapon types supplied by their leader or employer, and weapon production is centralised (Pauli Jensen et al. 2003:313). While conscripts in this sense do not seem to appear in the archeological source material until the Late Iron Age, Pauli Jensen et al. argues that the distinction between warriors and soldiers is indeed relevant for interpreting the 3rd century Illerup hierarchy. Thus, the army leaders, the princeps and the comites, seem to have had warrior status, while it is likely that the pedites were soldiers, as their weapon equipment appears very standardised (2003:313-14).

Both the Illerup and the Ejsbøl material seem to point to other divisions within the Germanic armies, perhaps of a different order than that between the three different levels of hierarchy, and perhaps related to tactical aspects of the armies. For instance, the number of horsemen in the Illerup army is considerably higher than the number of princeps, and, on the other hand, considerably lower than the number of comites, strongly suggesting that only a proportion of the officers were equestrians. This fits well with the written sources:

’The Germanic hosts which swept into southern Europe in the late second century BC were mainly foot-soldiers. A century later the warriors who formed the armies of Maraboduus and Arminius also fought largely on foot. The migrating peoples who passed into the Roman provinces depended overwhelmingly upon their infantry troops, relying for the most part on massed ranks of men. The use of mounted troops among most Germanic peoples was limited. … Julius Caesar employed German horsemen in his armies in Gaul … The cavalrymen who appear in the historical and archaeological record were mainly chiefs and their immediate retinues’ (Todd 1992:36-37).

Likewise, the pedites in both finds seem to consist of men with at least two different types of weapon combinations, with a smaller group displaying swords in addition to the numerically dominant lance/javelin/shield combinations. Another, lesser group is suggested by the presence of arrows in both the Illerup and the Ejsbøl finds.

There is no doubt that these were specialised, highly organised armies. The presence of cavalry and of archery testifies to the degree of dicipline and structure needed. Furthermore, the standardised weapons, the presence of Roman weapons and the sheer speed by which new Roman weapon types are adopted in Germanic society throughout the Roman period, points to there being a rather close relationship between Roman military developments and Germanic ones in this period. As pointed out by Pauli Jensen et al., there are profound differences between the tribal armies that can be reconstructed from the early war-booty offerings like Hjortspring (4th century BC) and the later ones, like Illerup, Vimose, and Illerup. However, there are also obvious differences between the contemporary Roman legion and the Illerup army. One striking feature of the Germanic armies was noted by Tacitus already in the late 1st century, namely that ’even iron is not plentiful with them, as we infer from the character of their weapons. But few use swords or long lances. They carry a spear (framea is their name for it), with a narrow and short head, but so sharp and easy to wield that the same weapon serves, according to circumstances, for close or distant conflict. As for the horse-soldier, he is satisfied with a shield and spear; the foot-soldiers also scatter showers of missiles, each man having several and hurling them to an immense distance…'. Still in the 3rd century Illerup army, lance/javelin and shield seems to have been the normal combination of weapons among the foot-soldiers, unlike in the Roman army (as for instance testified by Vegetius at the end of the 4th century, but with reference to ’olden days’).

Pauli Jensen et al. provides a likely answer to the problem, namely that the army organisation evidenced in Illerup ’A’ is less reminiscent of the Roman legion itself than of the supporting, auxiliary units (cf Albrethsen 1997:216): ’The explanation might be as simple as this: the Germans who fought on the Roman side in the Marcomannic wars and later on served as mercenaries in the period after Marcus Aurelius, have brought home with them their knowledge of the Roman military structure. The speed by which the Germanic armies adjust to Roman conditions – and vice versa – points to an interconnectedness, whereby each side continually adjust weapons and fighting techniques to the opponent’s newest innovations’ (2003:325).

The military capacity of Scandinavian society in this period is well documented also by other types of archaeological finds and monuments, be it boathouses (Grimm 2006; Myhre 1997), water barriers (Nørgård Jørgensen 2001), or hill forts (Engström 1991). Fortified setlements like the Early Roman period village of Priorsløkke in Eastern Jutland have been shown to have had defences in the form of pallisades well suited to resist attacks by mounted warriors (Kaul 1997).

Roman period weapon graves
Let us then move on to the weapon graves. There have been several attempts to correlate the weapon combinations and military hierarchy from Illerup place ’A’ with other types of archaeological source material, mainly Scandinavian weapon graves, not least by Ilkjær himself (von Carnap-Bornheim 1992; Gansum 2000; Ilkjær 1990, 1993, 1997, 2001a, 2001b, 2002, 2003; Ilkjær & von Carnap-Bornheim 1999; Solberg 2003:103-123; for a more critical view of the possibility of discussing military organisation based on burials, see Härke 1997). Naturally, much of the focus has been on the burials belonging to Ilkjær’s levels 1 and 2.

In Scandinavia the first weapon graves appear in the Late pre-Roman Iron Age, mainly, but not exclusively in parts of Denmark (Hedeager 1992:115). Although there are clear regional differences when it comes to leaving weapons with the dead (Hedeager & Kristiansen 1981:122f; Watt 2003), the practice as such from that point onward exists in many parts of Scandinavia right through to Christian times.

In Denmark, the number of weapon graves increases steadily in the Early Roman period (ER), when c. 7 % of all contemporary graves contain weapons. In the Late Roman period (LR), this percentage decreases to 2 % (Hedeager 1992:115). The samme pattern occurs in SE Norway as well, but somewhat delayed, as it were. Thus, while the weapon grave ritual is established in the ER, it does not peak until the LR (actually in periods C1a and C1b), only to decrease in the Migration period. In the latter period, most weapon graves are found in the Western districts of SE Norway, such as Vestfold and Telemark, but even there the number of graves are decreasing, even if the remaining ones are more elaborately furnished than in the preceding period.

As for regional differences, within SE Norway there is a core area with particularly many graves in the inland areas around lake Mjøsa and lake Randsfjord, centred on the districts of Hadeland, Toten, and Valdres. The number of Roman period weapon graves in these upland parts of SE Norway is matched by Gotland only in the Nordic countries. We will return to these districts below.

What do the weapon graves represent, status wise? The use of weapons in burial rituals is definetely a reflection of the militarisation of societies in the tribal zone beyond the limes. But are we correct to interpret the status associated with weapons and different combinations of weapons as military status? I think we are, and I believe the results from the present study strenghten this interpretation (below). This is suggested also by Hedeager’s study of the Roman period burials in Denmark, and more specific by her discussion of the place occupied by spurs in the graves. She points to anthropological studies (from Northern German cemeteries) which show that weapons are usually found with younger men, while sets with spurs but without weapons belong to older men (1992:160):

’In other words, weapons are particularly linked to the function of the active warrior. Second, warriors with spurs similarly lose the weapon set with age, but keep the spurs and the wealth; that is, they maintain a status which overrides the warrior function. Third, both weapons and spurs can be found in the graves of very young men, which leads us to conclude that the status/function marked by the right to carry both weapons and spurs is likely to be a matter of birth, rather than earned.’

These patterns are consistent with Tacitus’ description of Germanic practices (Germania, ch. 13):

’They transact no public or private business without being armed. It is not, however, usual for anyone to wear arms till the state has recognised his power to use them. Then in the presence of the council one of the chiefs, or the young man's father, or some kinsman, equips him with a shield and a spear. These arms are what the ’toga’ is with us, the first honor with which youth is invested. Up to this time he is regarded as a member of a household, after-wards as a member of the commonwealth. Very noble birth or great services rendered by the father secure for lads the rank of a chief; such lads attach themselves to men of mature strength and of long approved velour. It is no shame to be seen among a chief's followers. Even in his escort there are gradations of rank, dependent on the choice of the man to whom they are attached. These followers vie keenly with each others as to who shall rank first with his chiefs, the chiefs as to who shall have the most numerous and the bravest followers. It is an honor as well as a source of strength to be thus always surrounded by a large body of picked youths; it is an ornament in peace and a defense in war.’

There is, however, one aspect of the mortuary customs in the Roman period that argues against assuming a too direct and causal relationship between arming in life and weapon combinations in death. It seems that only one individual per family in each generation got a furnished burial, i.e. the first spouse that died, whether a woman or a man (Jørgensen 1988; Stylegar 2006). Most of the weapon graves ought to be sought in graves of this type. They are a way of showing status common to all families, but only once every generation. And not all families had access to the same resources, or the same weapons. Thus, while a male weapon-less grave does not rule out the deceased’s status as a warrior, a grave with weapons is indeed a reflection of the deceased’s military status. But it seems to me that the implications of this distinction are quantitative rather than qualitative, i.e. it does not ruin the possibility of reconstructing the relative size of different levels of the Roman period military hierarchy, only their absolute numbers.

For East Jutland it has been suggested that the distribution of Roman period graves with spurs indicate a political division of the landscape into a whole number of similar ’area moduls’, each of a size comparable to the later ’herreder’ (Hedeager & Kristiansen 1981:125f.). Lund Hansen, too, argues for a military interpretation of the weapon graves (Lund Hansen 1995). Starting from the richly furnished graves at the Himlingøje cemetery in Stevns in Zeeland, she points out that not only are weapons basically absent from Himlingøje; they are absent from a much larger area from Roskilde to Prestø. The weapon graves in Zeeland occur in a belt outside this area, and Lund Hansen interprets the weapon graves as belonging to a body of armed followers placed securely at some distance to the political centre at Himlingøje. But admittedly, her material is rather scarce.

Hedeager grouped the weapon graves into three different categories based on the number of weapon types in each grave (three, two, and one weapon, respectively), and she argued that this tripartite division reflected a military hierarchy (Hedeager 1990). She found that graves with a ’full’ set of weapons comprised only 10-15 percent of the weapon graves (1992:121). Her study predates the publications and analyses of the Illerup material, and the suggested military hierarchy, based solely on the grave finds as it is, represents a much coarser structure than the one argued by Ilkjær. Thus, to a certain degree all of Hedeager’s three groups can be found within each and every one of Ilkjær’s.

As for Norway, Roman period weapon graves have been the subject of a number of recent studies (Joki 2006; Eketuft Rygh 2007; Storli 2006). Thus, Storli has suggested that a relative distribution of weapon groups similar to the one argued by Hedeager can be discerned in Northern Norway. She too argues that individuals buried with a complete set of weapons represent a military elite (Storli 2006:88-89). As for the relative distribution of different weapon types, Solberg (2003:81-82) gives a brief summary of the Norwegian material as a whole.

What about SE Norway, then?

Weapon graves in SE Norway
The empirical basis for the present study is Late Roman period weapon finds from the modern-day counties of Østfold, Akershus, Oslo, Hedmark, Buskerud, Oppland, Vestfold, and Telemark. Altogether, c. 150 grave finds and stray finds with weapons that can be dated to C1-C3 are known from this rather extensive area. Most of the finds in question are listed in Ilkjær (1990) and/or Bemmann & Hahne (1994), while the rest are taken from the catalogue of the University museum of cultural history’s collection in Oslo.

Taken as a whole, graves with one weapon type comprise 41 % of the total number of graves. This numerical dominance of single-weapon graves fits well with Hedeager’s results based on the material from Denmark. However, graves with two weapon types comprise 27 %, and graves with three comprise 32 %. Thus, it seems that graves with a ’full’ set of weapons is over represented in SE Norway vis-a-vis Denmark. The ratio of graves with three weapon types from the Ejsbøl find is slightly over 20 %, thus somewhat higher than the grave material in Denmark, but still lesser than the number for SE Norway.

Looking at the material from the perspective suggested by Ilkjær, we find that 12 % of the weapon graves can be attributed to his level 2, against 88 % in level 3. For Illerup ’A’, the comparable ratios would be 9 % and 89 %, respectively. Only one burial in the study area belongs to level 1 – the C3 grave from Sætrang in Ringerike, Buskerud, with a bandoleer with silver fittings (Ilkjær cat. no. 735, cf. Fett 1937; Slomann 1959; figure 3). This single burial of course defy any statistics (it equals 0,7 %; just one more find would bring the percentage up to 1,3, and thus very close to Illerup’s 1,5 %). Thus, there is a great degree of correspondence between the SE Norwegian weapon graves and the Illerup find. When it comes to riding gear, 10 of the burials in SE Norway have spurs, equal to 7 %. The number for Illerup is 3 %, as it is for Ejsbøl.

If we look at the weapon combinations within each of the two groups (levels 2 and 3), we notice some striking differences between the two of them. In level 2, distinguished by shields, belts, and riding gir decorated with bronze fittings, 62 % of the finds (i.e. 13 finds) have sword, lance, javelin, and shield (and, in some instances, spur(s)). This seems to be the prescribed weapon combination for level 2, as the remaining 38 % of the finds consist of acccidental ’combinations’ each found in one instance. All level 2 graves have sword (except one find that consists of a shield boss only). Spur(s) occur in 29 % of the level 2 burials.

As for level 3, three different combinations stand out. The by far most common one, represented by 39 % of the finds (i.e. 47 finds) in this group, consist of a singular lance or javelin (and sometimes including a shield). Two other combinations – one with sword, lance, and javelin (and often including a shield) and one with lance and javelin (and sometimes including a shield) are represented by 22 % (27 finds) and 19 % (23 finds) respectively. The remaining ’combinations’ seem to be accidental and due to either bad preservation or the context of recovery, as none of them occur in more than three instances. Only four burials belonging to Ilkjær’s level 3 contain one or two spurs. 3 of these have a ’full’ set of weapons, while one contains only a lance.

To put it simple: Level 2 consist of two combinations: a full weapon set either 1) with or 2) without spur(s). Level 3 has three combinations: 1) a singular lance or javelin (and sometimes including a shield), 2) sword, lance, and javelin (and sometimes including a shield), and 3) lance and javelin (and sometimes including a shield).

In the Illerup ’A’ find, most level 2 sets consisted of ’full’ weapon sets, while the most common combination in level 3 was spear and javelin (and shield). As for level 2, our findings again correspond to the Illerup material. In level 3, the share of burials containing a sword is comparable to the Illerup situation (22 % in SE Norway; 28% in Illerup). Only one aspect of the SE Norwegian material associated with level 3 seems to diverge from the Illerup pattern, namely the dominance of finds with only one weapon type, either a lance or a javelin.

Some notes on source criticism are in order here. First, only a small fragment of the material stem from archaeological investigations. Most finds result from either accidental discoveries by people ploughing, digging, building roads etc, or come from barrows excavated by local farmers or other non-professionals. This is a general problem with Roman period graves in Norway (cf. Ilkjær 2001a:2-4). Strictly speaking, only c. 10 % of the 150 finds in our material result from scientific investigations. About half of these had a ’full’ weapon set, while the other half consisted of burials with either lance and javelin or a singular lance or javelin. Thus, the three main combinations (and only these) are represented in the material resulting from professional excavations.

Still, the dominance of single weapons might call for caution. It is not that it is unheard of in the Roman period material; in B1 Jutland and Funen, for example, a combination of a lance and a shield is very widespread indeed (Watt 2003:184-186). But if we look at Late Roman period graves from SE Norway in Bemmann/Hahne’s catalogue, single weapons account for almost 70 % of the finds from unknown contexts or not explicitly said to come from barrows or other grave contexts. Most of the remaining 30 % is taken up by a lance and a javelin found together. It does not seem unlikely that a portion of the single weapons actually represent specimens lost during hunting etc. In some instances this is a self-evident interpretation, and the finds in question are not included in this study. However, a look at Bemman/Hahne’s catalogue for Western Norway (i.e. the counties Rogaland, Hordaland, Sogn og Fjordane, and Møre og Romsdal) indicates that single weapons do actually form a separate weapon grave group.

There are relatively fewer Late Roman finds in this region than in SE Norway, but a seemingly clear pattern emerges when one look at the finds from more or less certain burial contexts: the combination lance and javelin is much more common in the Western districts, with twice as many finds as those represented by solitary lances or javelins. The situation is different with the stray finds, and here there is a 50:50 relationship between the two groups. Thus, is seems likely that a proportion of the single weapons either represent partially recovered burials, most likely of a lance and a javelin, or deposition contexts other than burials. But both these relative numbers are different from the SE Norwegian ones, and this further suggests to us that the combination lance and javelin is in fact more widespread in the Western districts than in the Oslofjord area, and that a substantial amount of single weapons in SE Norway actually represent graves with one weapon only (unless one wants to suggest that people in the latter region lost their weapons more often than their western counterparts).

Some of the single weapon finds in SE Norway probably represent only partially recovered graves, most likely graves with lance and javelin (and shield) combined. But it is not possible to state with any certainty the real relative size of the single weapon group versus the group consisting of graves with a lance and a javelin (Solberg states that lance and javelin is the most common combination in Norway, but this may be based primarily on the material from Western Norway, see Solberg 2003:81). Suffice it to say that the level 3 graves in SE Norway consist of three major combinations. The ’full’ weapon set accounts for 22 % of the finds, while the two other main groups are made up of single lances or javelins and lances and javelins found together. The relative weight of the latter two is uncertain, although together they account for c. 60 % of the total number of graves in level 3.

Let us now have a closer look at one sub-region in SE Norway, and see how the different levels of hierarchy and different weapon combinations appear on the ground, so to speak. Let us focus on the Upland districts.

The ’Opplandene’ region
Most of the weapon graves are in fact from this area, i.e. from the districts centred at the great lakes of Eastern Norway, the Randsfjorden, the Tyrifjorden, and the Mjøsa. Indeed, in Ilkjær’s catalogue (1990) of 202 Late Roman period finds from the whole of Norway, 62 are from the modern region of Oppland – more than twice as many as the second region. This remarkable distribution pattern was commented upon by Shetelig already at the Nordic archaeologists’ meeting in Copenhagen in 1919 (Shetelig 1920). In 1925, he wrote:

’Cremation burials with a rich equipment of weapons and often with imported bronze vessels appear sporadically, one and one, from the beginning of the Roman Iron Age on both sides of the Oslofjord, and in the lake districts in the Uplands, at Tyrifjorden and Randsfjorden. … It is the older finds, from the 1st-3rd centuries, that have this rather widespread and even distribution across Eastern Norway. But during the 3rd, and even stronger in the 4th century, the burials of this type cluster very clearly in one single, limited area, in Hadeland with Toten and Valdres, the old historical Hadeland, while they gradually disappear elsewhere’ (1925:136).

Shetelig’s great idea was that the weapon graves in ’Opplandene’ were due to an influx of military specialists returning home from service in Roman or Germanic armies in Continental Europa, especially after the end of the Second Marcomannian War in AD 180. He especially linked this return to the name Hadeland, i.e. ’land of the warriors’ (Shetelig 1920; for the possible meaning of the landscape name, see Sandnes & Stemshaug 1997). After Shetelig, the Roman period burials in the area have been dealt with by others, primarily by Grieg (1926) and Herteig (1955), both of whom suggested different explanations than Shetelig for the predominance of weapon graves, as they argued that most weapon graves in the area stem from what today are wealthy farms, and thus that the weapon burial custom reflected the upper strata of a farming society. At least for Hadeland, a more recent study confirm that the weapon graves are indeed concentrated in areas that in Medieval times were occupied by bigger than average farms (Skre 1998:228).

Eketuft Rygh has recently argued that a new political structure was coming into being in ’Opplandene’ in the Late Roman Period, one that in the Early Merovingian Period is reflected in the so-called Åker complex (2007:98f). This is a likely background for the important part played by weapon graves in the former period. Thus, there is probably some truth to both traditional interpretations, i.e. both the military and the social one, to the extent that social hierarchy is reflected in military hierarchy, or vice versa.

The weapon graves have a relatively even distribution across the region, but still with clear clusters in Gran (Hadeland) and Østre Toten. Together, these two districts account for almost half of the weapon graves in the region. But there are other interesting facts about the distribution of weapon graves in this area.

First of all, there is a tendency for graves with swords to cluster in the vicinity of level 2 graves. This is particularly clear in Gran and in Østre Toten. A similar picture seems to emerge in Ringsaker, the peninsula to the north of the lake Mjøsa, but no level 2 grave is known there as yet. These same clusters seem to exist through the whole period in question. Within the clusters are some farms with a number of graves with sword, often spanning several generations. In some instances, more than one level 2 grave are known from the same farm. This is the case with Stabu in Toten and Egge in Hadeland. Thus, there must have been at least some stability and continuity in the recruitment of officers, as they seem to have been recruited from the same farms through generations.

Secondly, graves with lance and/or javelin (and sometimes shield) as the only weapon(s) also occur in more peripheral areas, and thus have a much more widespread distribution than the sword graves. This is evident in Jevnaker, Gjøvik, Furnes, and northern parts of Ringsaker, as well as in Land. Over large tracts of land, lance/javelin seem to be the only weapons accompanying the dead.

How are we to interpret this rather striking pattern? Let us look at the clusters again. While it is true that they encompass some of the best farming land in the region, and thus represent people of relatively high social standing, there are other areas which are similarly well suited for agriculture but which don’t have any clusters. So while there is obviously some correlation between clusters of weapon graves and the leading social strata, the relationship is not straight forward – at least not if ’leading strata’ means free odal farmers, as in Grieg’s and Herteig’s argument against Shetelig’s military hypothesis. There is, however, one kind of military organisation which seem to fit the pattern: aristocratic retinues could be the raison d’etre for the clustering of weapon graves in certain areas.

There is an extensive literature concerning the retinue or comitatus described by Tacitus and said to be widespread among the Germans (Germania, ch. 13-15; see for instance von Carnap-Bornheim 1992). The comites would consist of both cavalry and infantry (Kristensen 1983). They had taken a special oath which obliged them to assist their leader (princeps) in war as well as peace. In return they received maintenance, gifts and a part of the spoils in case of raids or other war-like acitivities. The princeps’ reputation depended on the number of brave warriors in his retinue, which again depended on his generousity and luck in war. In peace time the retinue was a heavy burden, economically speaking (Hedeager & Tvarnø 2001:105). Skre argues that ’the aristocracy in the Nordic countries had warriors attached to their persons and households from the Roman Period and well into the Middle Ages’ (Skre 1998:261).

A few words need to be said regarding the level 1 graves, that is, graves associated with the princeps. As mentioned before, there is only one level 1 grave in SE Norway. This grave, from Sætrang in Ringerike, belongs to a weapon grave milieu similar to the clusters in our study area. A comparable situation exists for one of the two other level 1 graves in Norway, namely the grave from Erga in Klepp, Rogaland. Of the five Swedish finds, it seems to be the case with at least two. Therefore, there is a case for arguing that level 1 burials too belong to the clusters, even if we don’t have any level 1 burials (yet) in Hadeland, Toten, or Hedmark. I would argue that we ought to expect level 1 graves within each of the clusters that have yielded level 2 and level 3 graves in these areas, i.e. in Gran in Hadeland, in Østre Toten in Toten, in Vang/Løten in Hedemarken, and perhaps also in Ringsaker, since there is a cluster of weapon graves containing swords in this district, even if no level 2 grave is known there. Looking at Eastern Norway as a whole, there are two more clusters of this kind: Valdres and Southern Østfold, but only the latter has yielded level 2 graves.

Thus, it doesn’t seem far-fetched to suggest that what we have before us in the distribution of weapon graves in the ’Opplandene’, is a topographical expression of the difference between the sword-possessing aristocracy and their followings on the one hand, and ’ordinary’ warriors on the other. The members of the comitatus were based on the princeps’ hall. The reason for the clustering of weapon graves in smaller districts, i.e. not solely on the farm containing the princeps’ hall, could be that the comites were granted farms that were part of the princeps’ estate for their maintenance. The princeps and his retinue, the sword users focused on the former’s hall, that is, were warriors in Sander’s sense, their vocation was war. We still have to account for the division between level 2 graves and other graves containing swords, and it is likely that level 2 equipment was associated with a special function within the retinue, for instance with standard bearers (cf for instance Rygge 1970 on swords with figure inlays), or with leadership positions within the retinue.

With this in mind, then, how were big armies like the one defeated at Illerup recruited? They was probably made up of several chieftain’s retinues, with each level 1 warrior commanding a retinue including both level 2 and level 3 men. It is interesting to compare with the situation in the Viking Age, where for instance Knut the Great’s thingmannalid was constructed in a similar way (Lindkvist 1990; cf. Andræ 1960). Heerhufen is the term used by Kuhn to denote armies put together for a major military campaign or other particular purpose (Kuhn 1956). If the Late Roman Period Heerhufen were made up of several chieftain’s retinues put together for particular purposes in this way, the widely diverging size estimates of contemporary and near contempory armies extracted from written sources, become more understable (see for instance Hines 1989; Reuter 1997; Wallace-Hadrill 1975). It follows from the arguments presented in the preceding paragraphs, that 10 to 12 clusters of the same kind as the 3 to 5 found in the ’Opplandene’ would be enough to gather an army a 1000 men strong.

But how do we account for the relative dominance of swords within the clusters. The similarities between the relative distribution of weapon types in Norwegian graves and in the Illerup find seem at odds with the weapon combinations dominant within the clusters in ’Opplandene’. This might indicate that Late Roman Period Scandinavian armies consisted not only of several chiefs and their personal retinues, but that a considerable number of pedites, mainly armed with lance and/or javelin, and shield were added. Many of these seem to have been recruited from a much wider geographical area than the comites. The foot-soldiers were probably recruited from the large and diverse group of free men, including independent farmers, loyal to the chief. The pedites were probably soldiers in Sander’s sense.

Conclusion
The excavation of the Illerup ’A’ war-booty offering and the subsequent publication of the finds, have yielded new insights into Late Roman Period military organisation. But only few studies have so far tried to utilise these insights to (re-)study other types of source material that can cast new light upon how Late Roman period armies were organised and recruited.

The present paper has dealt with weapon graves from SE Norway.

Altogether, c. 150 grave finds and stray finds with weapons that can be dated to C1-C3 are known from this rather extensive area. Taken as a whole, graves with one weapon type comprise 41 % of the total number of graves. Graves with two weapon types comprise 27 %, and graves with three comprise 32 %. Thus, it seems that graves with a ’full’ set of weapons is over-represented in SE Norway vis-a-vis Denmark.

12 % of the weapon graves can be attributed to Ilkjær’s level 2, against 88 % in level 3. For Illerup the comparable ratios would be 9 % and 89 %, respectively. Thus, there is a great degree of correspondence between the SE Norwegian weapon graves and the Illerup find. In SE Norway, graves associated with level 2 consist of two combinations: a full weapon set either 1) with or 2) without spur(s). Level 3 has three combinations: 1) a singular lance or javelin (and sometimes including a shield), 2) sword, lance, and javelin (and sometimes including a shield), and 3) lance and javelin (and sometimes including a shield). Only one aspect of the SE Norwegian material associated with level 3 seems to diverge from the Illerup pattern, namely the dominance of finds with only one weapon type, either a lance or a javelin.

Most of the weapon graves in the present study are from the ’Opplandene’ area, i.e. from the districts centred at the great lakes of Eastern Norway, the Randsfjorden, the Tyrifjorden, and the Mjøsa. The dominance of weapon graves in this region has been dicussed within Norwegian archaeology since the 1920s, with researchers at times arguing for a military interpretion, and at times for a social or economic one.

In the ’Opplandene’, there is a tendency for graves with swords, either belonging to level 2 or level 3, to form large, marked clusters. There is one such cluster in each of the old ’folkland’ Ringerike, Hadeland, and Toten. Graves with lance and/or javelin (and sometimes shield) as the only weapon(s) also occur in more peripheral areas, and thus have a much more widespread distribution than the sword graves. In the Illerup find, swords are associated with the upper and middle level of the military hierarchy, as well as with a minority within the lower level. In Late Roman Period burials in ’Opplandene’ the numerical distribution of weapon types is similar to the one from Illerup. But topographically speaking there seem to be clear differences between weapon graves with or without swords, as graves with swords are found first and foremost in clusters which also contain level 2 graves, i.e. graves associated with the middle level of the Illerup hierarchy. There is only one kind of military organisation which seem to fit this pattern: aristocratic retinues.

The distribution of weapon graves in the ’Opplandene’ is a topographical expression of the difference between the sword-possessing aristocracy and their followings on the one hand, and ’ordinary’ warriors on the other. The members of the retinue were based on the chieftain’s hall. The reason for the clustering of weapon graves in smaller districts, i.e. not solely on the farm containing the chieftain’s hall, could be that the comites were granted farms that were part of the chieftain’s estate for their maintenance.

The spear and/or javelin carrying foot-soldiers were probably recruited from the large and diverse group of free men, including independent farmers, loyal to the chief.

Big Late Roman Period armies like the one defeated at Illerup were probably made up of several chieftain’s retinues, with each level 1 warrior commanding a retinue including both level 2 and level 3 men. 10 to 12 clusters of the same kind as the 3 to 5 found in the ’Opplandene’ would be enough to gather an army a 1000 men strong.

It cannot be a coincidence that there is one cluster of weapon graves with swords in each of the old landscapes of Ringerike, Hadeland, Hedemarken and Toten, i.e. all of the central districts in the ’Opplandene’. One can hardly avoid the conclusion that these’folkland’ were in existence already by the 3rd century AD, at least as distinct social and military hierarchies.

Literature
Albrethsen, Svend E. 1997. Logistical problems in Iron Age warfare. In A. Nørgård Jørgensen & B. L. Clausen eds.: Military Aspects of Scandinavian Society in a European Perspective AD 1-1300. Copenhagen.
Andræ, Clas-Göran 1960. Kyrka och frälse i Sverige under äldre medeltid. Uppsala.
Bemmann, Jan & Güde Hahne 1994. Waffenführende Grabinventare der jüngeren römischen Kaiserzeit und Völkerwanderungszeit in Skandinavien. Studie zur zeitlichen Ordnung anhand der norwegischen Funde. Berichte der Römisch-Germanischen Kommission 75. Mainz.
von Carnap-Bornheim, Claus 1992. Die germanische Gefolgschaft. Zur Interpretation der Mooropfer der jüngeren römischen Kaiserzeit in Südskandinavien. In E. Straume & E. Skar eds: Peregrinatio Gothica III. Oslo.
von Carnap-Bornheim, Claus & Jørgen Ilkjær 1999. Import af romersk militærudstyr til Norge i yngre romertid. In I. Fuglestvedt, T. Gansum & A. Opedal eds.: Et hus med mange rom, vol. A. Stavanger.
Engström, Johan 1991. Fornborgarna och samhällsutvecklingen. In C. Fabech & J. Ringtved eds.: Samfundsorganisation och regional variation. Norden i romersk jernalder og folkevandringstid. Jysk arkæologisk selskabs skrifter 27. Højbjerg.
Fett, Per 1937. Setrangbeltet og andre belter fra eldre jernalder. Bergen museums årbok 1937, no. 7. Bergen.
Gansum, Terje 2000. Romertid og germanertid i den norske nasjonalhistorien. I J. Ilkjær: Den første Norgeshistorien. Tønsberg.
Grieg, Sigurd 1926. Hadelands eldste bosetningshistorie. Oslo.
Grieg, Sigurd 1932. Hadeland i oldtiden. In Hadeland. Bygdenes historie, vol. 1. Oslo.
Grimm, Oliver 2006. Großbootshaus – Zentrum und Herrschaft. Zentralplatzforschung in der nordeuropäischen Archäologie(1.-15. Jahrhundert). Berlin.
Hedeager, Lotte 1990. Danmarks jernalder. Mellem stamme og stat. Århus.
Hedeager, Lotte 1992. Iron-Age Societies. From Tribe to State in Northern Europe, 500 BC to AD 700. Oxford.
Hedeager, Lotte & Kristian Kristiansen 1981. Bendstrup – a princely grave from the Early Roman Iron Age: Its social and historical context. Kuml.
Hedeager Lotte & Henrik Tvarnø 2001. Tusen års Europahistorie. Romere, germanere og nordboere. Oslo.
Herschend, Frands 1993. The Origin of the Hall in Southern Scandinavia. Tor, vol 25. Uppsala.
Herteig, Asbjørn E. 1955. Bidrag til jernalderens busetningshistorie på Toten. Oslo.
Hines, John 1989. The Military Context of the adventus Saxonum: Some continental evidence. In S. Chadwick Hawkes ed.: Weapons and Warfare in Anglo-Saxon England. Oxford.
Härke, Heinrich 1997. Early Anglo-Saxon military organisation: an archaeological perspective. In A. Nørgård Jørgensen & B. L. Clausen eds.: Military Aspects of Scandinavian Society in a European Perspective AD 1-1300. Copenhagen.
Ilkjær, Jørgen 1990. Illerup Ådal. Die Lansen und Speere. Aarhus.
Ilkjær, Jørgen 1993. Illerup Ådal. Die Gürtel. Aarhus.
Ilkjær, Jørgen 1997. Gegner und Verbündete in Nordeuropa während des 1. bis 4. Jahrhunderts. In A. Nørgård Jørgensen & B. L. Clausen eds.: Military Aspects of Scandinavian Society in a European Perspective AD 1-1300. Copenhagen.
Ilkjær, Jørgen 2000. Den første Norgeshistorien. Tønsberg.
Ilkjær, Jørgen 2001a. Unterschiede zwischen Moorfunden und Waffengräbern in der jüngeren römischen Kaiserzeit. In B. Storgaard (ed.): Military Aspects of the Aristocracy in Barbaricum in the Roman and Early Migration Periods. Copenhagen.
Ilkjær, Jørgen 2001b. Centres of power in Scandinavia before the medieval kingdoms. In B. Arrhenius (ed).: Kingdoms and regionality. Stockholm.
Ilkjær, Jørgen 2002. Fjender og forbundsfæller i romertidens Nordeuropa. In M. S. Vea (ed.): Fiender og forbundsfeller. Regional kontakt gjennom historien. Kopervik.
Ilkjær, Jørgen 2003. Danske krigsbytteofringer. In L. Jørgensen, B. Storgaard & L. G. Thomsen eds.: Sejrens triumf. Norden i skyggen af det romerske imperium. Copenhagen.
Jensen, Jørgen 2003. Danmarks oldtid. Ældre jernalder 500 f.Kr-400 e.Kr. København.
Jensen, Xania Pauli, Lars Jørgensen & Ulla Lund Hansen 2003. Den germanske hær. Krigere, soldater og officerer. In L. Jørgensen, B. Storgaard & L. G. Thomsen eds.: Sejrens triumf. Norden i skyggen af det romerske imperium. Copenhagen.
Joki, Heidi 2006. Idealet om krigeren. Våpengravene i Norge i perioden 150/60-310/20 e.Kr. som kilde til ideologi og militær organisasjon. Master Thesis, Bergen University.
Kuhn, Hans 1956. Die Grenzen der germanischen Gefolgschaft. Zeitschrift der Savigny-Stiftung für Rechtsgeschichte. Germanistische Abteilung 73.
Nørgård Jørgensen, Anne 2001. Sea defence in the Roman Iron Age. In B. Storgaard ed.: Military aspects of the aristocracy in Barbaricum in the Roman and Early Migration Periods. Copenhagen.
Jørgensen, Lars 1988. Family burial practices and inheritance systems. The development of an Iron Age society from 50 BC to AD 1000 on Bornholm. Acta Archaeologica 58, 17-53. København.
Kaul, Flemming 1997. Priorsløkke and its logistic implications. In A. Nørgård Jørgensen & B. L. Clausen eds.: Military Aspects of Scandinavian Society in a European Perspective AD 1-1300. Copenhagen.
Kristensen, Anne 1983. Tacitus’ germanische Gefolgschaft. Copenhagen.
Lindkvist, Thomas 1990. Plundring, skatter och den feodala statens frmväxt. Organisatoriska tendenser i Sverige under övergången från vikingatid til tidig medeltid. Uppsala.
Lund Hansen, Ulla 1995. Waffenfunde und militärische Ausrüstungen der seeländischen Inselgruppe in der jüngeren römischen Kaiserzeit. In U. Lund Hansen et al. eds.: Himlingøje – Seeland – Europa. Ein Gräberfeld der jüngeren römischen Kaiserzeit auf Seeland, seine Bedeutung und internationalen Beziehungen. Copenhagen.
Løken, Trond 2001. Oppkomsten av den germanske hallen – hall og sal i eldre jernalder i Rogaland. Viking 64, 49-86. Oslo.
Myhre, Bjørn 1997. Boathouses and naval organisation. In A. Nørgård Jørgensen & B. L. Clausen eds: Military Aspects of Scandinavian Society in a European Perspective, AD 1-1300. Copenhagen.
Petersen, Jan 1957. Hedemarken i hedensk tid. In R. W. Halvorsen et al. Eds.: Hedmarks historie, 1, fellesbind. Hamar.
Pilø, Lars 2005. Bosted – urgård – enkeltgård. En analyse av premissene i den norske bosetningshistoriske forskningstradisjon på bakgrunn av bebyggelsesarkeologisk feltarbeid på Hedemarken. Oslo.
Reuter, Timothy 1997. The recruitment of armies in the Early Middle Ages: what can we know. In A. Nørgård Jørgensen & B. L. Clausen eds.: Military Aspects of Scandinavian Society in a European Perspective AD 1-1300. Copenhagen.
Rygge, Elizabeth Wiese 1970. Victoria Romana i Norge. Universitetets oldsaksamling årbok 1967-1968. Oslo.
Eketuft Rygh, Bjørn-Håkon 2007. Den siste reisen. De sørøstnorske våpengravene som kilde til kunnskap om krigerspesialisten i eldre jernalder. MA Thesis, Oslo University.
Sandnes, Jørn & Ola Stemshaug 1997. Norsk stadnamnleksikon, 4. ed. Oslo.
Shetelig, Haakon 1920. Et norsk folkevandringsrike. Aarbøger for nordisk oldkyndighed og historie 1920. Copenhagen.
Shetelig, Haakon 1925. Norges forhistorie. Oslo.
Skre, Dagfinn 1998. Herredømmet. Bosetning og besittelse på Romerike 200-1350 e.Kr. Oslo.
Slomann, Wencke 1959. Sætrangfunnet. Hjemlig tradisjon og fremmede innslag. Oslo.
Solberg, Bergljot 2003. Jernalderen i Norge. 500 før Kristus til 1030 etter Kristus. 2nd edition. Oslo.
Storli, Inger 2006. Hålogaland før rikssamlingen. Politiske prosesser i perioden 200-900 e.Kr. Oslo.
Tacitus: The Annals and The Histories. Translated by A. Church. New York, 2003.
Todd, Malcolm 1992. The Early Germans. Oxford.
Vegetius: Epitome of military science. Translated by N. P. Milner. Liverpool.
Wallace-Hadrill, John M. 1975. War and Peace in the Early Middle Ages. Transactions of the Royal Historical Society, 5th series, no 25. London.
Watt, Margrethe 2003. Våbengravene – regionale forskelle inden for våbentyper og gravskik i Danmark, 100 f.Kr.-400 e.Kr. In L. Jørgensen, B. Storgaard & L. G. Thomsen eds.: Sejrens triumf. Norden i skyggen af det romerske imperium. Copenhagen.

Ingen kommentarer: