Tuesday, 8 April 2014

Nordic/Germanic Pagan Interpretations of Style I: A draft of a critique

[I am working up - very belatedly - the paper on Style I that I gave at the Leeds IMC in 2010.  I got a little distracted by critique of a particular school of archaeological thought on the topic. I present here, in draft from for discussion, that bit of the paper.  The footnotes etc are omitted because they aren't finished.]

Less satisfactory than the descriptive analyses, however, are the explanations of how, in Dickinson’s own phrase, ‘animal art gained its place in early medieval affections’.  Previous analyses have, in my view, tended to be constricted within an unsatisfactory and problematic conceptual matrix. 

One principal axis of this matrix is the cultural description of the art-style as ‘Germanic’.  The explanatory weaknesses of the term ‘Germanic’ become very evident when one encounters phrases like the following:
‘[Style I’s appearance] is marked by the sudden disappearance of all sea creatures, which up till then dominated Scandinavian ornament and represents the beginning of the Germanic interpretation of the animal world’ (emphasis added)[1]
This begs two crucial questions: ‘why then?’; ‘why like that’?  We are entitled to ask why, if the term ‘Germanic’ can satisfactorily perform any analytical work, the ‘Germanic interpretation of the animal world’ in art only ‘begins’ in the late fifth century, when Germanic-speakers had dominated the region for centuries.  Furthermore, why does this art take this particular form after (and indeed before) centuries within which the metalworkers of Germania Magna had proved more than capable of reproducing Roman models or otherwise producing coherent figures and interlace?  Appeals to a pan-Germanic cultural ethos get us nowhere in response to either question. 

That the notion of pan-Germanic identity and ethos emerged in the precise, contingent circumstances of sixteenth- to (especially) nineteenth- and twentieth-century German politics is surely now well established. The idea that all speakers of a Germanic language can be treated as culturally interchangeable can no longer be sustained.  Scandinavian origin myths among post-imperial peoples cannot be traced before the sixth century and, even then, among the Italian Goths alone, where it was clearly only one of a number of stories circulating.  Alleged Scandinavian origins for other early medieval peoples (like the Lombards or Burgundians) emerge later, in emulation of the Goths.  Yet, such appeals to ‘Germanic’ culture remain as common as ever.  A swathe of recent work on Style I has attempted to read it in line with changes allegedly occurring in Scandinavia at this time which produced the Germanic concept of the hall.

Another appeal to the pan-Germanic ethos is associated with the fairly widespread claim that Style I was a badge of a ‘shared Germanic aristocratic identity’.  When applied to decorative art, this argument is circular.  Style I’s popularity is explained because its ‘Germanic’ nature appealed to the ‘Germanic’ social élites who sponsored its production.  For the irrefutable reasons previously outlined, ‘Germanic’ is a label that can carry no analytical, or even descriptive, weight, either for the art or the people.  That a common élite identity existed amongst Germanic-speakers in the late fifth and sixth centuries finds no more support in any written data, such as exists for at least some of the regions this shared aristocratic culture allegedly encompassed.  This explanation has no empirical grounding whatsoever.  Therefore, the existence of a shared ‘Germanic’ élite ethos has to be argued in reverse, from the distribution and popularity of the art style.  And so on...  Furthermore, if ‘Germanic’ art only emerged in the later fifth century, in spite of the long-standing existence in Scandinavia of Germanic-speakers, then it would seem that, to be popular among a Germanic élite with a shared culture, that ‘shared Germanic culture’ could – logically – only have emerged at the same time, alongside the art.  How this might be possible across a wide area of diverse, frequently antagonistic peoples is not an issue that appears to have troubled students of Style I who espouse this argument.  If there were any area where a common, non-Roman, military identity might be employed to unify Germanic-speakers of diverse cultural, geographical and familial backgrounds, that area would not be Germania Magna but, ironically, the Roman Empire itself, where fifth-century politics were increasingly focused around military leaders of at least a claimed non-Roman origin.  It may in fact be that material culture did come to be employed in precisely that way to unify exactly that disparate élite group but, if it did, it was material culture originating in the Mediterranean, not the Baltic. 

The other component of the problematic analytical matrix is religious.  Style I imagery frequently continues to be read according to concepts drawn from Norse sources from (at the earliest and most optimistic) c.1000, half a millennium after the appearance of this form of decoration.  These texts are then supplemented by (and indeed read via) anthropologically-derived ideas of shamanism and tribal ritual of uncertain applicationIt must never be forgotten that the principal sources upon which any view of Nordic paganism is based, the Prose and Poetic Eddas, were both written down in earlier thirteenth-century Christian contexts, the former by Snorri Sturlusson.  The extent to which they represent earlier texts (written or oral), the faithfulness with which they do so, and how much earlier any such texts can be dated are all matters for guesswork, not fixed starting points for analysis. 

The ‘pagan’ axis of Style I interpretation nonetheless assumes that an image of pagan belief obtained from these texts can be applied, in detail, to artwork no less than 750 years earlier.  However, between the late fifth and the late twelfth centuries artistic motifs changed dramatically in the Nordic world.  Style I gave way to Style II, Style III, and then the array of Viking styles – Oseborg, Borre, Jelling, Mammen and Ringerike – before, with Urnes Style (in its early, middle and late forms of course), we arrive at the end of the Germanic/Viking Animal Art ‘era’.  Even at Urnes Style’s demise we still have fifty years or so to wait before Snorri wrote down the Prose Edda.  While Scandinavian decorative art between c.475 and c.1170 went through no fewer than nine changes (not counting Urnes Style’s subdivisions) of sufficient importance to necessitate a change of art-historical categorisation, the mentalité it represented is thus supposed somehow to have drifted along unchanged beneath this turbulent surface of actual, documented expression, eventually to be written down in effectively the same form as it had taken 750 years previously.  The difference between the supposed stasis of Scandinavian pagan belief and the dynamism of its artistic expression is made yet more problematic by the paradigm itself, which sees the emergence of Style I animal art (surely correctly) as symptomatic of important social and cultural change.  If that were the case then something similar surely lies behind the many subsequent changes of style.  Scandinavian archaeology, reveals important, dynamic change in social and economic structures throughout the second half of the first millennium.  Against this background, an entirely stable set of religious beliefs and practices defies credibility.
These problems are only exacerbated at the earlier end of the period.  If Style I’s content is religious, does its difference from earlier art mean that this religion only emerged c.475 and, if so, why?  Or, if the religious beliefs were older, why did art not manifest it before the emergence of ‘Germanic animal art’, or why did it represent it in such different ways?  These are the same problems as we encountered with the ‘Germanic’ construct, and they are equally fatal to the approach.  The religious axis takes us nowhere – in answer to the questions of ‘why then’ and ‘why like that’ – just as quickly as the ‘Germanic’.   

The strongest argument offered in favour of the approach is the depiction on Style I metalwork of episodes identifiable, with varying degrees of plausibility, as those mentioned in the Eddas.  There are two problems with this justification.  A fourth-century depiction of Christ at, say, the marriage at Cana might be recognisable as such to a twelfth- or thirteenth-century western European Christian.  We would quite wrong, however, to assume on that basis that Christian theology, practice and organisation from 1300 could be applied in any detail at all to the church of c.400, or even of c.900 and its artwork.  In that instance, moreover, a series of more or less canonical written texts existed, around which beliefs could be anchored to some extent.[2]  It is worth pointing out, furthermore, that the earliest runic texts actually to mention the gods of the Nordic pantheon come from slightly later than the period of Style I and frequently from areas far to the south, in ‘Alamannic’ regions.  One inadmissible justification for the use of later Eddic evidence is the assertion that structures would remain intact within traditions even if the details changed.  It is difficult to see what the basis for such a statement could be and, in any case, the ‘structures’ adduced are modern interpretations, rather than being explicitly described in the texts.  At the weakest end of the scale is the simple acceptance of the method, ‘faute de mieux’.  Tania Dickinson, one of the most careful and knowledgeable of Anglo-Saxon archaeologists, simply says that most schools of archaeological ‘theory’ accept the approach.  That will not suffice to exclude critical reflection on the issue.

These analytical axes are sustained, first by a raft of similarly evidence-free, mystifying ideas about the magic quality of the artefacts and then by ideas of ethnicity, migration and of a rigid Christian-pagan divide which are all inadequate to the task.[3] 

[1] Haseloff 1974: 12

[2] This means that, even if one could assume that the religious ‘texts’ of the Eddas (written or otherwise) had been transmitted, unchanged, for 700 years before being written down, we would be in no sense authorised to assume an unchanging accompanying theology.

[3] The most alarming element of what we might call the Nordic/Germanic Pagan’ school of writing about Style I is that it seems, intellectually, to be entirely hermetically sealed; discussion appears to be locked within a particular academic circle.  The sorts of questionable readings just discussed are repeatedly built upon, adding further assumptions and interpretations, which then in turn become building blocks for further work and so on, producing pyramids of dubious assertions.  With so much now at stake, in terms of academic status and power, there seems no way that a fundamental, scholarly critique of the approach can emerge from within this paradigm.

Wednesday, 2 April 2014

The Ostrogothic Military

[Here is a first draft, again in need of a thorough edit, of a chapter which have kindly (and somewhat surprisingly) been asked to write for a forthcoming Brill Companion to Ostrogothic Italy, on the army of that interesting realm.  It attempts to steer a middle (and I hope more interesting and subtle) course in a number of polarised debates: on 'the Goths: army or people', 'Hospitalitas' and the nature of Gothic ethnicity.  As ever, comments and corrections are welcome.  Thanks, too, to those who helped me track down the hardly-probative evidence for Theodoric's disbanding of the Roman Guard regiments.]

A gratuitous plug, but it does (probably)
Have Theodoric on the front.
The Ostrogothic Kingdom was created and destroyed by conquest and, throughout the realm’s short life, the army remained a central feature of its politics and society.  It is not surprising, therefore, that the Ostrogothic army has featured prominently in modern discussions of the kingdom.  A discussion of military affairs in Gothic Italy requires that attention be devoted to issues such as the Gothic settlement and its nature and the ethnic politics of the kingdom, which have been the focus of much, sometimes fierce, debate among historians and archaeologists in recent decades.  This chapter will be organised according to three main chronological phases: first, the army of the conquest; second the army during Theodoric’s reign as king of Italy; and finally the army of the Gothic wars.  This will permit the examination of change through time as well as allowing the analysis to deal with issues specific to each sub-period.  Although the Ostrogothic Italian kingdom spanned only three generations or less, it is important to remember that Theodoric’s was a long reign by any standards.  The important dynamics at work meant that the troops who accompanied him across the Isonzo in 489 were very much not those who undertook the military operations of his last years and entirely different from those of the Gothic Wars.

A: The Army of the Conquest

Theodoric’s Goths: Army or People?

The forces that Theodoric led across the Alps in 489 had developed out of more than one group of Goths.  Principally they originated in the armed following of Theodoric himself and in that of his namesake, Theodoric Strabo (‘the Squinter’).  Neither group can be considered as ‘the Gothic people’, regardless of how later sources, from within the Italian kingdom and outside, may have wanted to create that image.  The very fact that two such Gothic groups existed gives the lie to such a supposition.  Nor should we suppose that these were the only two such armed groups of Goths; they were simply the most numerous and, therefore, the most politically and militarily significant. 

Like the others, these bands originated in the instability that followed the fragmentation of Attila’s short-lived trans-Danubian ‘empire’ in the 450s.  In Attila’s polyglot empire, beneath a unifying Hunnic political identity, his subjects doubtless had several levels of what could be considered ethnic identity.  In a justly famous story, the East Roman ambassador Priscus met a Greek in Attila’s camp, but this ‘Greek’, it is clear, also fully regarded himself as a Hun.  It has long been pointed out that most of the Huns known to us, not least Attila and his brother Bleda, bear Gothic names.  Such material culture as can reasonably be associated with the Hunnic Empire makes most sense as emerging from a mixture of local Roman and Barbarian traditions.  After Attila’s death, civil strife broke out between his sons and other of his former commanders and troops.  Often depicted as a ‘rising’ of ‘subject peoples’, this conflict seems more reasonably described as a succession crisis in which those opposed to the Attilan dynasty adopted a non-Hunnic identity, bringing back to the surface those identities which had occupied lower levels during Attila’s reign but which, like the Greek ethnicity of Priscus’ interlocutor, had always been there.  Following the defeat of Attila’s sons at the Nedao, a somewhat bewildering array of ‘peoples’ came fleetingly into view in the wreckage of the Hunnic ‘empire’.  It is questionable whether some can even be considered as having a solid historical existence as peoples.  Only three named Skiri are known to us: Odovacar, his father and his brother.  On this basis it is difficult to decide whether Skirian identity ought to be considered ‘ethnic’ or familial.  Nonetheless, a successful family might attract enough followers for that kin-group identity to be spread and adopted and become an identity that operated in an uncontrovertibly ‘ethnic’ fashion.  After all, historians have become accustomed to describing all the people and culture of post-imperial Gaul between the late fifth and eighth centuries using a familial identity that originated precisely in Odovacar’s generation – Merovingian.  The families of the two Theodorics stressed their Gothic identity (‘Ostrogothic’ being a designation that originated among the Byzantines), just as other people with Gothic names and their followers had adopted, or continued to proclaim, a Hunnic ethnicity.  Others still made political claims based around Gepidic, or Herulian, or Rugian, identity.  Whether any of these factions ought to be considered a revival or reappearance of a tribe with a long pedigree seems questionable. 

One recent debate has been whether the Goths formed a ‘people on the move’, as in traditional Völkerwanderung interpretations, or, as in more recent works, a simple military force: an army.  This controversy is not capable of easy resolution.  Extreme interpretations, at either pole, are unsatisfying, not least because the terms ‘army’ and ‘people’ are both rather trickier to define than might be assumed.  Consequently, between the ‘polar’ readings things get considerably messier and conclusions more difficult to pigeon-hole as ‘army’ or ‘people’.  Nevertheless the issue is obviously of considerable relevance to this chapter.

Our sources mention that the Gothic factions (like, presumably, the others) had women and children in tow.  This has been taken as implying that they should be considered a ‘people’ on the move.[1]  This does not necessarily follow.  Roman armies took non-combatant women and children with them too, and so did most armies until well into the twentieth century.[2]  While this sounds a note of caution, however, it does not authorise us completely to disallow the interpretation of the Goths as ‘people on the move’.  Pursuing the ‘factional’ interpretation outlined earlier permits an intermediate course to be steered.  We might envisage a more inclusive social group, with women and children, but with young male warriors in the following of more established leaders nevertheless forming the most important element.  The issue of age has not fully been taken into account in discussions of the Gothic settlement of Italy (or, indeed, in that of other ‘barbarian’ settlements). 

After a long period of almost constant campaigning in the Balkans, in and out of official East Roman service, three consequences can readily be imagined.  One is the knitting of warrior bands into established, quasi-permanent bodies of men.  Living together year-round, practising weapon-use together and regularly fighting alongside one another, these would acquire most of the significant attributes of regular military units and the whole organisation those of a permanent army.  Indeed that seems generally to have been how the Ostrogoths functioned in Balkan politics in the 470s and 480s. 

The second consequence of the Goths’ career in this period, however, will have been the steady acquisition of wives and children, and doubtless other camp followers.  Paradoxically, then, as the Goths took on more of the form and functions of an army, they will also have become more varied in social composition.  The third consequence, all too-often forgotten, is that all through this time young warriors got older; mature warriors became old and possibly infirm.  Without an established place in Eastern Roman social, military and political structures, they could not settle down.  They had little other option but to continue to move and – as long as they could – fight with the rest.  This too made the Goths, even if originating, organised and functioning as an ‘army’, much more like a ‘people’ than most military forces.  Therefore, to see the force that headed for Italy in 489 as, by then, looking rather more like a ‘people’ than a normal army, one need not (indeed I do not) envisage Theodoric’s Goths as originating as a tribe or people that upped sticks and moved en masse.  Put another way, once the dynamics of the situation are thought through, even a narrowly military reading of the Goths’ origins and structure (like this one) must – in the end – see the force that arrived in Italy as something more socially variegated.  That fact must impact significantly upon how we understand Gothic settlement.

Italian Background

The Italian military background is also important.  The fifth-century decay of the western Empire produced crucial changes within Italian politics, as elsewhere in the West.  The loss of direct imperial control over Africa in the 420s and 430s was of critical importance.  The threat of seaborne attack from Carthage, meant that a significant force had to be stationed throughout the Italian peninsula, rather than just in the north, in a way that had hitherto not been necessary.  A key element of fifth-century politics was the increasing separation and rivalry between Italian and Gallic aristocracies.  Indeed, throughout the West, imperial politics came to focus upon different regional factions.  However, whereas the fourth-century Italian aristocracy had had little option but to accept the de facto shift of the imperial core to Gaul and the Rhine frontier, it now had an armed force to hand to help it ensure it commanded the centre of politics and patronage.  Control of the Italian army became crucial in peninsular politics, as Ricimer’s long period of dominance makes very clear.  Although unable to establish itself over the factions based upon the Gothic and Burgundian armies in Gaul, the Dalmatian army, or the Vandal forces in Africa, it nevertheless retained control over Italy itself, expelling the Gallic/Gothic faction in 457 and the (legitimate) Dalmatian claimant in 475 as well as fending off occasional attacks from African Vandals and trans-Alpine Alamanni.

Recruitment remained a problem, however.  The lack of effective control over fiscal resources much beyond Provence and the Narbonnaise in Gaul or beyond Tarraconensis in Spain greatly reduced the income of any Italian emperor.  Simultaneously, the peninsula became something of a political hot-house as the senatorial nobility, likewise cut off from properties and revenues abroad, found itself competing with lower-order aristocrats for honours, titles and patronage, especially at a local level where differences in wealth were now much reduced.  These factors made the government’s ability to levy troops as well as taxes more problematic.  It was preferable to use taxes to pay for the recruitment of soldiers from outside Italy, specifically from barbaricum and especially trans-Danubian barbaricum.  Such forces, at least initially, lacked local loyalties and were more easily employed as a coercive force.  In these circumstances it is unsurprising that the governmental resources used to pay the army were referred to as the fiscus barbaricus.[3]

Nonetheless, crucial dynamics were at work here as well as within the Ostrogothic army.  The remuneration of Roman soldiers had always involved land for settlement, and late Roman forces, as noted above, legitimately lived and sometimes moved, accompanied by wives and children.  Recruits, like Goths, get older, marry and settle down.  Military service was hereditary, so the children of such unions expected, in time, to follow their fathers into the army, which, over time, became as integrated into peninsular society and politics as any other group.  The soldiery that serially deposed Julius Nepos and Romulus ‘Augustulus’ probably contained significant numbers of men born and raised in Italy, even if serving in units with barbarian titles: second-generation ‘Italo-barbarians’.  

This discussion casts the confrontation between Odoacer’s and Theodoric’s armies between 489 and 492 somewhat differently from the clash of barbarian armies sometimes imagined.  Both sides shared important features but both originated in a very specific, fifth-century imperial context.  The similarities between the armies doubtless explain the drawn-out, long-indecisive nature of the struggle and the common changing of sides.  Nonetheless, Theoderic’s troops’ military experience and long practice in operating as units, was probably crucial to their eventual victory in the field.[4]  The similarities between the opposing sides may go some way towards understanding the nature of Gothic settlement in Italy.


The settlement of the Goths has probably been the most controversial issue concerning the place of Theodoric’s army in Italian society and politics.  It has been dubbed, misleadingly in some ways, ‘the Hospitalitas debate.’  The name hospitalitas (loosely, hospitality) came from a late Roman billeting law, describing the division of billets into thirds, the householder taking two and the soldier the other.[5]  Thirds also appeared in Procopius’ Wars, where he claimed that the barbarians were settled through the appropriation of a third of the land of Italy, and in Cassiodorus’ discussion of Gothic ‘thirds’ or ‘shares’ in the Variae.  That late Roman billeting law was used to divide the land of Italy, with one third going to the Goths, was long the more or less unchallenged understanding of the mechanics of barbarian settlement in Italy.  It fit well with then dominant paradigms, seeing the fifth century’s principal feature as violent barbarian conquest and viewing the barbarians as tribes searching for land.

This consensus was challenged in 1980 by Walter Goffart’s important book Barbarians and Romans.  Goffart changed the game in several ways.  He used the Italian evidence, rather than the Burgundian as had hitherto been more usual, to shape his general theory about how barbarians were settled.  The Italian data were more contemporary after all, if in some ways less detailed than the relevant clauses of the Burgundian Code.  Aquitanian Gothic and Burgundian settlements were separated from the documents that appeared to describe them by a long time and – evidently – more than one phase of settlement.  With Ennodius’ and Cassiodorus’ writings, one had a direct view of how barbarian troops were settled in a Roman province.  Goffart’s second, more famous, move was to place the settlement within the context of the mechanics of Roman taxation.  Broadly, he argued that the Gothic settlers in Italy were granted not ‘thirds’ of land but ‘thirds’ of tax-revenue.

Goffart showed that the Roman law of hospitalitas had nothing to do with salary, provisioning or settlement, but simply the temporary provision of shelter.   He dismissed Procopius’ testimony as parti pris, motivated by a desire to justify Justinian’s reconquest.  The Wars clearly manifest Justinian’s ideological campaign, claiming that the West had been lost to ‘barbarian invasion’ and thus required reconquering.  It seems reasonable to regard Procopius as distorting evidence concerned with ‘thirds’ to paint Theodoric in a bad light.  The reference to a third of the land may be no more than hyperbole, rather than having anything at all to do with the tertia referred to in other sources.  With Procopius set aside, Goffart turned to directly contemporary Italian evidence.  First of all there was rhetorical evidence in the form of Ennodius’ and Cassiodorus’ writings declaring that the Goths had been settled without the Roman landowners, especially the senatorial class, feeling any loss.[6]  It was difficult, said Goffart, to envisage such statements being made if Roman landowners had indeed just been stripped of a third of their estates.

Goffart then turned to more technical aspects of Cassiodorus’ Variae, especially to the terms illatio tertiarum and millennarius.[7]  The illatio tertiarum had previously been read as a tax paid by landowners who had not had had their estates partitioned in order to provide land for a Goth.  It was thought to equate with such a partition by being levied as a third of the revenue of all land.  Alongside actual expropriation, this would have made Ennodius’ and Cassiodorus’ statements (just referred to) crass in the extreme; it would have represented a serious – if not crippling – burden on the aristocrats of Italy.  The Italian aristocracy clearly retained its fifth-century prosperity under the Ostrogoths and it is hard to imagine this if their revenues had been reduced by this level.  Goffart suggested that the illatio was, rather, a third of the usual tax revenues.  This was set aside for the payment of the Goths, so the ‘third’ (tertia) referred to, not to a share of lands or an estate but instead to a fraction of fiscal income, diverted towards the Goths as their salary. [8]  In most earlier readings a millenarius[9] had been assumed command a thousand men, a chiliarch.  The term can indeed mean this.  Goffart, however, pointed out that a millena was a notional tax assessment unit in the Roman Empire, still used in Ostrogothic Italy.[10]  In imperial administrative practice units of this sort, in specific numbers and perhaps drawn from particular fiscal assets, were set aside for designated purposes.    For Goffart, a millenarius was a Gothic soldier paid with a millena of tax-revenue, which he collected from specified taxpayers.[11]  Gothic troops were also given periodic donatives and received other rewards from the king.[12]  Some recorded conflicts between Gothic soldiers and Italian taxpayers arose where the former attempted to convert a legitimate right to receive a salary in tax into the illegitimate ownership of the land on which that tax was levied.[13]

Goffart’s reading has considerable advantages.  Not the least is simplicity.  No longer did one need to envisage a horde of agrimensores touring the Italian peninsula, measuring up estates and their relative value before assigning measured portions to specific Gothic soldiers.  The state gained an independent coercive force – a standing army – and lost nothing; revenue collection was simplified.  It is difficult to gainsay are advantages.  Nonetheless, it is fair to say that most historians have remained unconvinced.  Some critiques have been scholarly and struck important notes of caution; others less so.  The most important issue is that, as formulated in 1980, Goffart’s thesis required readers to understand terra as having a specialist meaning as ‘fiscal revenue from the land’ rather than simply as ‘land’.  Many authors argued that this seemed rather forced.  In 2006, however, Goffart penned a rather more persuasive discussion drawing attention to the fact that even straightforward-looking references to land in the modern documents of the same general type as those investigated to look at barbarian settlement mean more than simple designated areas of the earth’s surface.  ‘Land’ comes with a web of diverse types of relations and obligations attached.  This ruled out any argument that simply proclaimed terra to be ‘unambiguous’, meaning ‘land’ as though ‘land’ were itself unambiguous.  In any case, Goffart’s argument had never simply relied upon tendentiously translating words like terra in new, technical ways.  It had accounted for a series of other relationships, largely ignored by anti-Goffartian works that proclaimed that ‘land is land’.

The main problem for Goffart’s critics (and there are many) is that the root of the traditional view was the common appearance of divisions into three in the Roman law of hospitalitas and in some texts that discuss barbarian settlement.  Whatever else one might have to say about Goffart’s studies, one thing is clear.  It is, to my knowledge, the one area where no historian has attempted to contradict him: Goffart decisively showed that the text in the Theodosian Code dealing with hospitalitas had no bearing on the issues dealt with in the texts that deal with fifth- and sixth-century barbarian tertia and the rest.  Therefore, even if one finds Goffart’s argument entirely unconvincing, it is not viable blithely to return to what he has recently called ‘expropriationist’ theses of the old style, based ultimately on that hospitalitas law.  Sadly, however, that seems to have been the option most commonly adopted.  Unwilling to accept Goffart’s ‘fiscalist’ revision, they have continued into a sort of interpretive hyperspace, where no reading actually has any grounding but where we can at least shut our eyes, block our ears and pretend to know where we are.  Yet, the imagined fixed points from which the anti-Goffartian view takes its bearings are, historiographically-speaking, spectres: reflections of something no longer there.

None of this implies that Goffart’s interpretation is unproblematic; his most recent contribution to the debate certainly does not ‘definitively’ settle the issue.[14]  Some preliminary ground-clearing is necessary.  We must be absolutely rigorous in keeping to the precise issue under debate and to the particular data relevant to that issue.  Evidence, for example, that the Goths owned land cannot be employed against Goffart’s view.  His book concerned the salary of barbarian settlers within the Empire and thus their relations with the state.  It discussed accommodation in that precise sense, not where or whether the barbarians owned land; the barbarians had to live somewhere.  We must also avoid the temptation to assume that a single mode of settlement applied in all cases.  Interpretation must relate to the particular documents under discussion and the specific social and political relationships described there, neither element of which need have universal implications. 

Similarly, context is crucial.  One cannot fill in the blanks from other barbarian settlements or, conversely, export Italian Gothic details to understand other arrangements.  The Gothic situation differed profoundly from those pertaining to the settlement of the Aquitanian Goths or the Burgundians in Sapaudia.  The system of accommodating the Goths in Italy was adopted when that barbarian army had just conquered and assumed direct government of the territory which had to provide for its payment.  The power-relations involved were therefore entirely different from those pertaining in Aquitaine in 418-9 or in Sapaudia in 442.  In both of the latter cases the barbarians were moved and installed by a Roman government which dominated them militarily.  Obviously, the victorious Goths are likely to have employed extant administrative arrangements such as may have been used in earlier settlements but we must not assume that all details must therefore have been the same.  The contrary seems more likely. 

A further piece of ground-clearing arises.  We need not assume, as do Goffart and his critics, that all the land of Italy was encompassed in the discussion of ‘thirds’.  The only text to talk in such terms is Procopius’ discussion of the background to Justinian’s reconquest.  If, as Goffart does (correctly, in my view), one rejects that testimony as, or as stemming from, Justinianic propaganda, one must logically reject it all.  One cannot pick and choose details from it.  The most one might say from it, as noted earlier, is that the fraction of a third might have come from the legal arrangements employed.  The documents discussing precise cases have no necessary implication of a universal, peninsula-wide arrangement.  All they need assume imply is that those relationships applied to sufficient lands or resources as were necessary for the Gothic army’s payment.  Indeed, all we need assume is that those relationships applied to sufficient lands or resources as were required to pay those Goths who were paid in that way.  There is no necessary implication that all Goths were remunerated entirely in the fashion discussed in the handful of relevant documents in the Variae.  Critics of the Goffart hypothesis have made the point before that it is unlikely that all Goths received the same payment, but they have usually done so on the entirely mistaken assumption that a standard salary, rather than a standard means of paying a salary, was implicit in Goffart’s argument.  The problem with the Goffart thesis lies in that very suggestion of a single means of remuneration. 

Nonetheless, Goffart’s reading of the nature of illatio, tertia, sortes and millenarii seems to me to be preferable to its alternatives.  There were, furthermore, late imperial Roman precedents for the general system suggested, and such a system had been used, it seems, for the payment of elite field armies, such as the Goths, in a general sense, were.[15]  A Gothic warrior would be paid by a draft on taxation.[16]  He would collect this from designated tax payers and, as Gothic status seems to have been more or less equivalent to service in the field army, one imagines this relationship would be inherited by his sons alongside the obligation to military service.  Most, if not all, of the key elements within this situation derived more or less directly from those existing in the late imperial army.  It is crucial, as Goffart correctly noted, to underline that the relationship between Goth and Roman was that of government official to taxpayer.  Beyond the legitimate payment of designated tax, it implied no other relative status.  A Goth may have been of a higher or lower standing than the Roman or Romans ear-marked to pay him his salary.   

The Goffart thesis’ limitation is its insistence that this single system entirely sufficed in all cases.  That requires complex and sometimes less convincing argumentation to explain away texts that do not fit.  It is simpler to propose that, while the system reconstructed by Goffart provided the army’s essential salary, it was not necessarily the only means used.  It has been argued before that different Gothic status groups may have wanted payment in different forms, rather than just in varying amounts.  The resources of the sacrae largitiones and res privata surely passed directly from Odovacar to Theodoric and these included landed estates and palaces as well as other sources of revenue.  At least one Gothic family (the Amals), then, received land upon which to live...  It is hardly implausible that Theodoric employed these resources, as the emperors had, to reward some of his followers.  The grant of imperial fiscal land on long-term, emphyteutic leases was reasonably well attested as a form of imperial patronage.  Theodoric, of course, had other – entirely traditional – resources that would fall within the sacrae largitiones and res privata.  The confiscation of the property of defeated enemies was a standard element of the fall-out from civil war.[17]  It is reasonable to see Odovacar’s senior supporters being expropriated in this way, their land taken by the fisc and used to reward Theodoric’s followers, or some of them.[18]  Contemporary sources talk of significant massacres of Odovacar’s men.  These men were probably paid according to a system much like that proposed by Goffart but they also, as noted, lived somewhere and that property would have fallen to the Gothic leader to retain or redistribute as he saw fit.  We can envisage at least some of Theodoric’s senior or favoured followers being remunerated with grants of land.  However, this has no bearing on the documents discussed by Goffart or the precise situations described in them; it has no relationship to the normal military salary was in Gothic Italy.

A considerable swathe of agri deserti (lands with no registered tax-payer) also existed.  The late Roman state had continued to reward retiring veterans with land and used agri deserti for this.  As these lands, by their very nature, yielded no tax, using this resource cost the government nothing.  Indeed, since they were now enmeshed in a system of military obligations, it extended the fisc’s resources.  Note, though, that, as agri deserti produced no fiscal income, they (like the confiscations just discussed) cannot have any relevance to discussions of sortes or tertia.  Some of the dynamics within the Gothic army come into play here.  Not all Theodoric’s men were warriors in the prime of life.  Some had campaigned with him for twenty years and doubtless expected to retire and settle down.  Others may have fought on into old age, or accompanied the army as infirm ex-warriors, simply because of the protection provided.  Such people would not normally draw an annual salary, plus periodic donatives, in return for military service.[19]  Land was a more appropriate reward for them.  Nonetheless, because Gothic soldiers had heritable status and obligations, lands used for these purposes were automatically entwined in issues of military service, especially when inherited by the next generation. 

Let us imagine an elderly Goth, an old companion of Theodoric and perhaps Triarius his father, being given an estate from the Italian agri deserti as a reward for long service.  This Goth, no longer militarily active, has a son, who is still in the army.  That son collects his salary from designated taxpayers according to the Goffart system; he is a millenarius.  When the old Goth dies, the son inherits his land.[20]  But, because the son inherited his Gothic status and military obligations from his father, that land is now seen as subject to military service.  The mature Goth now supports himself from two sources, both ultimately granted by the government: the ager (no longer desertus) and the salary from the millena/e allotted to him.  Imagine another Goth, a young man who joined Theodoric’s army in the course of its campaigns, with no elderly relatives to support.  He is paid, after the conquest of Italy, from designated millena.  As he grows older he marries an Italian woman and has children.  He may or may not buy land in this time, but when he retires he is rewarded, in Roman fashion, with a landed allotment.  The same features come into play as with the first Goth.  His sons inherit his identity and military obligation.  When they inherit the ager that land, like that of the older Goth discussed earlier, becomes part of a fiscal resource of a new type – land held tax-free, effectively in return for military service – and they have two sources of sustenance. 

None of the foregoing hypothetical reconstruction seems unlikely.  Note two things.  First, there has been no expropriation of any Roman landlord.[21]  Second, the Goffart interpretation of the standard means of furnishing a soldier’s salary remains entirely intact.  None of the reconstruction forces any revision at all of Goffart’s reading of the precise texts dealing with the illatio, with tertiae, or with millena/millenarii.  

Crucially, however, the system described contained within it the seeds of potential change through time.  Indeed recognition of change and the passage of time are crucial to a full understanding of the issue.  Thus, within a generation or so we have a situation where Gothic soldiers draw a salary not just from drafts on taxation, where land with attached military obligations has come into the equation.  This situation is indeed close to that which seems to be visible in Merovingian Gaul slightly later in the sixth century.  Indeed the growing connection between Gothic troops and specific lands and communities envisaged here is precisely the dynamic suggested earlier, whereby earlier barbarian recruits (such as Odovacar’s followers) had gradually become fixed within the Italian landscape.  The power relations also remain intact; the Gothic government retains a standing, salaried army while simplifying aspects of revenue collection and distribution, as Goffart imagined.  The advantage of this reconstruction is that it is dynamic.  Over time, with the cycles of life, salaried Gothic soldiers evolved into people settled in communities with their families, with local social ties (beyond those of tax-payer and tax-collector).  They nevertheless remained an essentially military element of society and politics.  This allows us to keep the Goffart interpretation in place and understand references to landed properties held by Goths without either having to explain them away or, alternatively, using them as a basis for rejecting Goffart’s thesis about salaries.

Goffart himself drew attention to another dynamic: the temptation for a Goth to transfer his right to collect a salary from a designated fiscal asset into the latter’s outright ownership.  This would completely change the relationships involved, making the tax-payer – whose status vis-à-vis the Goth’s was otherwise immaterial – into the Goth’s tenant.  Some documents in the Italian corpus seem to represent attempts to prevent, or to investigate allegations of, such abuses.  During weak, especially minority, government it is easy to imagine such abuses becoming more widespread and more successful.  I have argued that precisely this dynamic is behind changes in the nature of Frankish aristocratic landholding and power in Merovingian Gaul during a period of stress largely brought on by royal minorities around 600.  If we accept Procopius’ account, it may even be that this lay behind the Italian army’s demands to Orestes that led to the latter’s downfall though, as mentioned earlier, rejection of the whole story is probably the most consistent attitude to adopt.  Yet a further dynamic within the model is the purchase or acquisition by other means of landed properties that would be owned by the Goth in the usual way.  These, unlike any lands granted as remuneration for service and subsequently inherited, would not be free from the capitatio or other relevant taxes and obligations.  Goths might however want to extend their tax-exemption to all their lands.[22]  This, as again I have suggested for Gaul in the mid- and later sixth century, would be a source of conflict.[23]  Overall, it is important not to see the system used to settle the Gothic army after 492 as taking a single form or to imagine that the initial state of affairs remained in place and unchangeable throughout the kingdom’s existence.

B: The Army of the Ostrogothic Kingdom

The army in the governance of the kingdom

How to unify and govern Italy was the greatest problem facing Theodoric after his victory over Odovacar.  The power of Roman aristocrats, especially those below the level of the old senatorial nobility, whose authority was probably more intensive within specific localities, as well as the potential threat posed by leading Gothic families, aggravated the difficulties to communication and the exercise of power posed by Italy’s difficult physical geography.[24]  Theodoric’s relations with the army make a good case study of how he dealt with this problem.  In order to maintain his authority, the king had to scatter his forces throughout the peninsula in garrisons.  Yet this potentially made the problem just alluded to more acute.  A local commander (perhaps with a claim to nobility or even royalty just as good as Theodoric’s) might use his armed force, perhaps in alliance with the aristocrats of the region, to challenge royal authority.

One solution might be to ensure that Goths were not stationed, or did not carry out their military service, inmillenae were located, though whether such a solution was practical in Italy is doubtful.[25]  Theodoric seems instead to have made clever use of patronage and propaganda.[26]  The army appears to have been regularly assembled in the principal, northern royal centres: Pavia, Milan and Ravenna.  At these gatherings, Theodoric paid donatives (a supplementary salary in cash), rewarded those who had done well and punished those who had not.[27]  This enabled a continuous distribution and redistribution of royal patronage, not only in the circulation of offices but also in the geographical redeployment of his men.  This ensured that no single family or faction could build up an established local power-base.  Furthermore, it put Gothic families of noble or even royal status in clear competition for royal favour with their more low-born companions. 
Another gratuitous plug, but with a
helmet of (possibly) Ostrogothic
manufacture on thecover
the regions where their

When the army was assembled for these purposes it was subject to manifestations of royal ideology aurally, in the forms of speeches, panegyrics and so on, and visually, in the pictorial and epigraphic decoration of the buildings used.  The Senegallia Medallion demonstrates that some of the largesse distributed also carried Theodorician propaganda.  As manifested in Cassiodorus’ writings, these ideological productions stressed the army’s role as a pillar of civilitas and consequently the requirement for harmonious relations between Gothic troops and Roman civilians.   This output, of course, also stressed Theodoric’s status as the only legitimate king of the Goths, claiming (at least by the latter half of the reign) an ancient dynastic claim to royalty.  Competition for Theodoric’s patronage entailed subscription to this propaganda and ideology.  Consequently, an association with the king, or royal authorisation, trumped any other claim to legitimate authority in the localities.[28]  This process enabled Theodoric to undermine pre-existing social distinctions within the Goths and ensure that the royal writ penetrated the many, geographically disparate local communities of his realm.  Simultaneously, of course, it assured the army’s continuing function, in spite of increasingly complex and deeper-seated local ties, as an independent, state-controlled coercive force.

None of this meant that relations between the army and local society were uniformly harmonious, of course, any more than the imperial Roman army’s relations with civilians had been.  The Variae contain numerous instances of conflicts and complaints arising from the army’s behaviour.[29]  It is a repeated concern of Cassiodorus that Gothic troops not molest, harass or steal from the provincials in the areas where they are stationed or through which they are marching.[30]  One document compensates the provincials of the Cottian Alps for depredations committed by the army as it passed through the region en route to Gaul in 508.  Like Roman troops, while on service the Goths were supplied with food and other necessity by the fisc, in the form of the annonae.  For garrisons out on the kingdoms mountainous northern frontiers this was especially important.  Hungry troops could easily start to take what they wanted from their civilian neighbours.  As a result, Cassiodorus penned more than one document concerned with the rapid and effective payment of annonae.[31]


Reading the Variae, a rich source for the place of the army within Theodoric’s realm, gives a clear impression of the great degree of continuity from the late imperial situation into Ostrogothic Italy as far as military affairs are concerned.  Indeed, other than the fact that the army was made up of Goths, the Variae provide no a priori evidence that much had changed at all in this sphere.  Like the late Roman army, Gothic soldiers were subject to their own jurisdiction.[32]  It seems preferable to read the texts discussing the jurisdiction over Goths and Romans in this way rather than assuming that they refer to ancient Gothic tribal custom. 

A couple of texts suggest that one means by which serving Gothic soldiers were distinguished from civilians was (as in other areas) by their long hair (as capillati).  This was a survival from the late Roman military.  Whether this meant a particular hairstyle or was simply a reference to the typically hirsute appearance of serving soldiers (cp. the French poilu) is unclear.  The heavy chlamys also continued to be a notable sign of military authority.[33]  It may also be that the army was a key institution in maintaining outward signs of Gothicness.  A possible role in male socialisation will be discussed later but the late Roman army had long espoused signs of barbarian identity, real or invented.  Its jargon had incorporated Germanic terms, for example and the long hair of the capillati might have been another sign of ‘barbarian chic’.  The army had been a bastion of the Arian creed in late imperial Italy as well.  For these reasons it was well suited to maintain the means of signifying Gothic identity, such as Arian belief and use of the Gothic language, at least for specialised technical terms.

How the army was organised is not entirely clear.  Theodoric is said to have disbanded the Roman guards regiments as useless ceremonial units.[34]  However, although the comes domesticorum vacans was certainly an honorific rank, the text cited to support the claim says the opposite.[35]  The Variae refer to domestici and scholares.[36]  Royal bodyguards are referred to in the accounts of the Gothic Wars, albeit called by atticising Greek terms (hypaspistai, doryphoroi).  The reference to the horse- and foot-guards as domestici patres equitum et peditum, which perplexed Hodgkin, may point at an important structuring element of the Gothic army, to which I will return. [M. Maxime Emion has pointed out to me that patres (rather than partis or other variants) may simply be a happax in one of the mss. My thanks to him.]

As well as the guard regiments, the late Roman army had been organised into a field army (the comitatenses) and regiments of frontier troops (limitanei or ripenses).  Whether such a division continued to exist in Gothic Italy is unknown.  It has been argued that the Variae illustrate the existence of limitanei but the reference cited does not support the suggestion.[37]  Troops were certainly stationed on the frontiers in various forts, and Theodoric adopted traditional Roman vocabulary to refer to their function of keeping out the barbarians.  The Variae, however, give no hint that these were recruited from a different part of the population from the field army.  The term miles is sometimes used for these troops, and Goths are not mentioned.  Goths are more often mentioned in the exercitus, when it is on campaign.  This might support the suggestion, which would make sense given the more ‘barbarian’ composition of the late Roman field armies.  However, the formula for the appointment of the duke of Raetia, a frontier province, makes clear that milites are simply enough soldiers, serving in the exercitus, and contrasts them with Romani and provinciales.[38]  Nonetheless, Roman aristocrats had raised and commanded local defence forces during the fifth century – Cassiodorus’ great-grandfather was one such[39] – and it is likely that city garrisons included local Roman as well as Gothic soldiers.  Some sort of distinction may then have existed.

The ethnic component of the army has been one of the more debated elements of the Gothic military, especially since Patrick Amory proposed that Gothic identity was no more than a sort of professional appellation, founded in the ideology of the late Roman Empire; to be a Goth was simply to be a soldier.[40]  Amory’s view of ethnic identity as a kind of free choice has been forcefully criticised by Peter Heather, who argues that the Goths were very much a people, with an ethnic identity grounded in a class of freemen.[41]  Amory’s hypothesis of entirely fluid ethnicity is too extreme, but Heather’s primordialist view is much too crude.

At the heart of the controversy is the failure by both sides to appreciate two points.  First, ethnic change should not be envisaged as a straight swap, exchanging one real, monolithic identity for another.  Ethnicity is multi-layered, so that the process of ethnic change involved adding another level, not rejecting one’s entire ethnic identity and replacing it wholesale with another.  Different levels of identity can be reordered in importance according to the situation in which one finds oneself.  Over time an identity can come to be that according to which one normally acts and is categorised, without one necessarily ever abandoning the other identities.  We saw this process in action earlier, in the formation of Theodoric’s Goths themselves out of the wreckage of Attila’s empire.  The second point, related to this, is that the process whereby a person or, better, a family might change from primarily self-identifying as Roman, for example, to primarily self-identifying as Gothic, could take a long time: at least a generation and perhaps two or three.  This problem is accentuated by the short life of the Ostrogothic kingdom.  Although long, Theodoric’s reign spanned less than two generations.  The succession crises and instability that followed his death and, especially, the Gothic Wars that broke out in 535 (still only forty-six years after the Goths’ arrival on the Isonzo) are likely to have put a brake on these processes.  Thus it is hardly surprising that one cannot document clear-cut instances of complete ethnic change. 

Nonetheless, what is visible in the Ostrogothic evidence is the fact that the dynamics of such change existed, as indeed they did elsewhere in the post-imperial West. One index is the attestation of individuals with Gothic and Roman names.  It needs to be remembered that the addition of an extra name was hardly uncommon in late antiquity, especially when associated with a change of status.  Gregory of Tours added the name Gregorius when he entered the priesthood; his maternal great-uncle Gundulf doubtless took that Germanic name when he entered the service of the kings of Austrasia.  This was one means by which the gradual change of primary ethnic identification was brought about.  Amory also drew attention to the aristocrat Cyprian who had had his sons instructed in weapon-use and even had them taught Gothic.[42]  It is significant that this took place thirty years or so after Theodoric’s entry into Italy.  It seems clear that the competition for royal patronage and the advantages that went with military service were bringing even wealthy Italo-Romans to adopt a Gothic identity.  It is possible that service in local garrisons could result in the patronage of a senior Gothic  warrior, entry into a military household and thence inclusion in the exercitus, on the basis of which a Gothic identity might adopted and eventually become dominant.  Had the Amal kingdom lasted as long as the Merovingian it is likely that these dynamics would have had similar results to those observable there in the writings of Gregory of Tours. 

It is likely that the life-cycle played an important part in Gothic military service, as already intimated.  The Variae mention that adolescent Goths came of age when they were liable to serve in the army.[43]  Plausibly this took place at fifteen.  Cassiodorus mentions the training of iuvenes, who appear to be archers (saggitarii) and a mobilisation order commands the Goths to bring forth their young men.  This is where the mention of domestici patres takes on an added significance, possibly as a reference to older warriors.  Comparison with other post-imperial situations permits the suggestion that when he came of age a Goth learnt his trade either in the household of an older Gothic warrior or in units commanded by such veterans (like, perhaps, the archers of Salona).  ‘Adoption by arms’ was possibly important at this stage and would further bind military communities.  Merovingian comites had their followings of pueri; it may be that the domestici in attendance on Theodoric’s officials ought to be seen in the same way.  Clearly, they were paid salaries by the fisc.  At some point they may then have graduated to more established units of milites, with a salary drawn in the way outlined above.  Finally, at a certain age, they may have married, acquired lands and settled down, becoming older warriors called out only for specific campaigns but training their own military households.  This proposed system looks superficially ‘primitivising’, making the Gothic military resemble the Zulu army’s division into married and unmarried impis.  In fact it is consistent with a range of evidence across post-imperial Europe.  Even the late Roman army, where twinned regiments of iuniores and seniores might imply similar careers for Roman soldiers.[44]  The distinction between the doryphoroi and hypaspistai of Belisarius’ guards (whatever their actual designation) suggests a similar life-cycle-based career within a regular army.  The suggested role of the life-cycle within Gothic military service adds to other dynamics proposed in this chapter to underline change through time and the evolution of military identities and systems of remuneration.

Theodoric was careful to ensure his armies were well equipped and supplied, as the imperial armies had been.  Cassiodorus’ writings contain numerous references to the upkeep of proper military camps, the regular provision of annonae and the supervision of armourers.  The king also took a close interest in making sure that his cities were properly fortified.

Archaeological Evidence

It is often thought possible to determine the areas where the Gothic army was settled from the archaeological record.  The distribution (figure 1) of particular types of metalwork, usually from a inhumations wherein the dead were interred with such objects as grave-goods, has been thought to reflect the areas of Gothic settlement.  Such an interpretation cannot stand in any simple fashion for a number of reasons.  First, most of the material in question (largely feminine in any case) cannot necessarily be assigned to a place of origin that would authorise its designation as ‘Ostrogothic’.  Second, archaeological material does not have an ethnic identity in any case so that, even if such material demonstrably in the Gothic homelands north of the Danube, one would not know whether the person interred with the objects was a Goth who had accompanied Theodoric to Italy, or descended from one such.  Third, perhaps most importantly, the material is found in very small numbers.  If the costume associated with these objects had come to be thought of as Gothic, it must nevertheless be the case that not all Goths were buried in this fashion.  Thus the rite cannot simply
Figure 1: Finds of 'Gothic' Material culture
reflect Gothic settlement.

The context of such isolated finds is, consequently, crucial to our understanding.  The finds, as mentioned, mostly originated as objects deliberately and publicly deposited with the dead.  Although, as shown in figure 1, there are about fifty sites in Italy and Dalmatia containing such burials, usually only one or two such graves are known from each site.  Some are from urban cemeteries, notably at major centres like Rome, Ravenna, Aquileia and Milan and frequently associated with churches.    We must ask why some people were buried like this but the vast majority was not.

If these artefacts were possibly associated with Gothic holders of political and military power  their display in the public ritual of burial must be significant.  There are weapon-burials and other furnished inhumations especially in peripheral areas of Italy in the pre-Ostrogothic era so the custom of displaying a dead person’s status in death was not new.  Nonetheless earlier barbarian troops, like Odovacar’s do not seem on the whole to have taken to manifest their ethnic identity in this way.  That the Goths did so must therefore illustrate in some way the impact of imperial collapse and Gothic conquest upon Italian social relationships.  Furnished inhumation was a public display.  It is possible that, in the suburban church burials with possible Gothic connotations, the audience of a burial was made up of the politically powerful.  In rural contexts, as perhaps, if the find does not represent a hoard, with the lavish burial of a woman at Domagnano (San Marino),[45] such an audience might have comprised other local landowners and lesser people. 

That women as well as (if not more often than) men were buried with these objects like this argues that the deaths of all members of these kindreds could be marked by such displays.  It also argues for a particular gendering of power.  The families employing the ritual clearly demonstrated the basis of their pre-eminence: their association with the Gothic holders of political and military power.  This could be linked with competition for royal patronage within local communities and among the political élite.  We must also, however, surely conclude that people adopting this costume in public rituals were not necessarily (and were possibly even unlikely to have been) incomers of Danubian origin.  Nonetheless, the fairly limited number of these burials shows that, while the death of a family member produced some stress, any threat to their local standing was not critical.  These displays, however, still speak of some of the tensions involved in establishing new local power-structures.  The distribution of finds is thus most likely to show the distribution of areas where such local stress and competition for power was most common.  In the light of the above discussion, it is highly likely that these included some of the areas where Gothic newcomers were settled, but clearly it need have absolutely no relationship to the overall distribution of such settlements.  The quality of the evidence, almost invariably discovered long ago in obscure and even dubious circumstances, is so poor that more detailed social and chronological analyses are impossible. 

What is important for the purposes of this chapter is the way in which, in however attenuated a form, it shows that the political and military power associated with the Goths found its way down to local societies and the power-struggles within them.  That the styles of objects which seem to have manifest a connection with Theodoric’s government were as often feminine as masculine might further support the suggestion that, however they were salaried, over time the Gothic soldiers and their families became a fixed component of such communities and their politics. 

It is difficult to say much from the archaeological record about the equipment of Theodoric’s soldiers.  Weapons are not a common element of the find complexes just discussed, not least because so many of them are female burials.  Such as are known are unremarkable: lance-heads, Lavish items of horse-harness confirm the indications of the written sources that cavalry were a key element of the Gothic army. 

Several fortifications are known, which were occupied in the Ostrogothic period.  Invillino (Friuli) is one of the best known and most thoroughly excavated.  Although no phase had a direct relationship with the Ostrogothic period, its Period III encompassed that era.[46]

What is clear is that the Ostrogothic army of Theodoric’s reign was a highly organised and efficient instrument.  Its Gallic, Spanish and Balkan campaigns were well-organised, well-led and usually successful.  Success breeds success, of course.  Warriors continued to join Theodoric and the repeated experience of victory made Gothic troops not only battle-hardened but confident. 

C: The Gothic Wars

The accounts of the cataclysmic downfall of Theodoric’s kingdom in the Gothic Wars provide us with much detailed, but far from unproblematic, data on the Gothic army in action.  A significant mistake in the use of Procopius’ account of these wars has been its use to shed light upon the nature of the Goths who had entered Italy in 489.  As I have been at pains to stress throughout this chapter, numerous dynamics were at work within the kingdom – and especially within its army – that made the armed forces of the 530s to 550s quite different from those of the 480s and 490s.  The features of ‘the Goths’ as they appear in Procopius’ narrative are highly dependent upon the working through of these dynamics. 

We must be careful with Procopius’ account.  Although filled with the sort of detail that military historians love – and which is generally absent in early medieval western Europe – it cannot be taken as a straightforward description, even if Procopius was an eye-witness to some of the events described.  The Wars are enmeshed in ethnographic stereotyping of very traditional classical form and Procopius himself wrote in learned attic Greek, striving to make his account fit with the great examples of the historical genre: Thucydides and Polybius.  Hence the appearance of doryphoroi and hypaspistai in Roman and Gothic armies.  Procopius’ writing – at least initially – was heavily imbued with Justinianic ideology about the rightness of the reconquest.  For these reasons the accounts of the Gothic forces, especially in the set-pieces of the siege of Rome, have to be handled with the utmost care.  Procopius mocked those barbarians who wanted to be Romans.  Thus the tragicomic accounts of incompetently deployed Gothic siege towers and Gothic generals who fail to note the allegedly decisive military difference between the two armies, spotted by Belisarius early on in the campaign: that the Romans have mounted archers and the Goths do not.  Some of these accounts are surely hyperbolic.  Procopius’ account of Gothic ‘oplitoi must surely also be heavily ironic.  Although the term was apt description of an armoured close-fighting spearman protected by a large round shield, the cultural baggage of the attic hoplite, civilised citizen-soldier par excellence, and its disjuncture with barbarian warriors besieging Rome would not have been lost on Procopius’ readership.  Procopius less critical attitude towards Totila may stem as much from the fact that Totila correctly played the role of barbarian warlord allotted to him by Graeco-Roman ethnography – unlike the comic philosopher-king Theodahad or the equally bumbling Wittigis, would-be poliorcetes – as from any disillusionment with Justinianic policy.

Close scrutiny of the account suggests that the two sides were actually very alike.  The possibility of a distinction between older and younger warriors, the former acting as officers for the latter, especially within bodyguard units, has already been mentioned.  Another shared feature seems to be the ability of the warriors to fight mounted or on foot, according to the requirements of the situation.  This fluidity, rather than a formal division into units of infantry and cavalry, is quite characteristic of the early medieval west.  That the Gothic army, as is very clear from Cassiodorus, was a well-organised, more or less regular army on the Roman model, rather than the barbarian horde often envisaged in Byzantine accounts or uncritical modern studies based on the latter, also brought the two sides closer together.  Indeed, given the predominance of troops recruited from beyond the frontier, the Gothic army may indeed have been considerably more ‘Roman’ than the Roman forces opposing them.  This sort of irony seems to be heavily played upon in Procopius’ account.  However that may be, the similarities between the armies certainly facilitated (as in Theodoric’s campaign of conquest) the changing of sides between the two armies.  Walter Pohl has drawn attention to the fact that soldiers in the opposing forces were barely distinguishable from each other. 

The somewhat dismal showing of the Gothic army in the earliest phase of the war probably attests to the political stresses of the previous decade or more and the lack of active campaigning in that period.  Most of the experienced Gothic troops will have been located outside Italy, in the Balkans (where they scored some important early successes against the invading Romans), in Provence and even in Spain, where it seems likely that they were involved in sometimes successful campaigning against the Franks.  On the other hand, they faced battle-hardened and confident veterans, used to success under Belisarius (even if frequently more by luck than judgement).  The dynamics of the earlier Theodorician period were reversed.  They would turn back again when the Gothic forces of Totila experienced a long and unbroken period of battlefield successes.  

In terms of equipment, the Gothic warrior characteristically had horse, sword and shield, as the written and archaeological evidence from Theodoric’s reign also suggests.  Some troops used bows, at least when dismounted, and spears were thrown from a distance as well as used in hand-to-hand fighting.  Totila’s instructions to his men to discard all weapons other than their swords (if Procopius is to be believed) made sound sense in the context of the battle of Busta Gallorum.  He wanted his army to charge quickly and directly into close-combat.  Any temptation to exchange missiles with the Romans, who had the advantage of numbers, especially in archers, would be fatal.[47] 

The effects of the wars on the Italian peninsula are well-known.  One was surely the arresting of any dynamics that might have ultimately have led to the sort of ethnic change visible in Gaul and Spain and which are attested in Theodoric’s reign.  Harder lines were drawn between Goths and Romans, even though those lines almost certainly had more to do with political choice than biological descent.  Most of the rank and file of the 520s will have been born and grown up in the Italian kingdom, making them a significantly different group from warriors born and raised within the peripatetic Ostrogothic army in the post-Hunnic Balkans.  It is unlikely that more than a handful of those who mustered in Theodoric’s last military assemblies, even patres domestici, had any clear memory of a life outside the seemingly stable confines of Romano-Gothic Italy.  It would be yet more mistaken to see the soldiers who opposed Belisarius’ troops, let alone those who followed Totila and Teïas against Narses, as shaped by anything other than late antique Italian, Provençal or Dalmatian culture and society, just like their ‘Roman’ or ’provincial’ neighbours.  Marriage doubtless blurred familial and other genealogical distinctions further.  Besides that, the processes discussed earlier had already led to Italo-Romans joining the army and perhaps adding a Gothic dimension to their own hierarchy of identities.  The Gothic army had always incorporated other groups, sometimes retaining an ethnic label,[48] sometimes not.  Byzantine deserters joined the Goths during the wars and doubtless also added Gothic to their identities.  Those who went back to the East Romans surely abandoned it again.  None of this implies ‘incomplete assimilation’[49] or hard boundaries between Goths and others.  For one thing, we do not know whether Roman soldiers who returned to Justinian’s armies were the same men as deserted earlier.  A Roman deserting to the Goths became, in some ways, a Goth, although the non-Italian and frequently indeed non-imperial origin of these troops will have continued to mark them out among their new comrades; given the Italian upbringing of most Goths,  it was easier for a Goth deserting to Narses to become, in some ways, a Roman.

The dynamics discussed throughout this chapter permit a more subtle reading of the ultimate demise of the Goths than that which has most recently been repeatedly championed.  It has been claimed that the account of the final defeat of the Goths shows that they were a ‘people’ with a defined identity founded in a large class of freemen with a direct link to the king, as well as a class of aristocrats.  The decisive defeat of a portion of the Gothic army, and the threat to wives and children posed by Eastern Roman military operations has been held to be sufficient proof of this contention. 

The conclusion, however, by no means necessarily follows from the evidence.  The revival of the discredited Germanist notion of a class of Königsfreie need not detain us.  That the Gothic armies were stratified and contained a larger body of rank and file than leaders should hardly surprise us; nor should the idea that the rank and file had a significant part to play in politics.[50]  I have suggested that numerous dynamics within the Italian kingdom worked to embed Gothic military communities within peninsular communities, society and politics.  The edges of these communities doubtless hardened during the wars, as we have seen, and it cannot be astonishing that the inevitable wives and children, whole families, of serving Gothic soldiers should have been more at risk than in the peaceful conditions of Theodoric’s reign.  It might have been considered safer to take them on campaign than to leave them behind, giving some Gothic forces a character resembling that of the Goths who entered Italy in 489.  That the serious and bloody defeats suffered by Gothic field armies (and navy) similarly has no necessary bearing on the nature of the Goths in Italy.  The destruction of its field army at Adrianople in 378 had rendered the Eastern Empire – with far greater reserves of military manpower than the Italian kingdom – effectively incapable of serious offensive military action for perhaps a decade.  The slaughter of the western field army at the Frigidus was decisive for the western Empire; it never had a sufficient breathing space to rebuild a substantial force of the same standard.  Troops can be replaced in numbers but not necessarily in quality, military experience and so on, and Procopius makes it clear that their limited manpower, compared with that of the Empire, was a worry that dictated Gothic strategy in the 540s and 50s.  The men who accompanied Totila in his desperate charge at Busta Gallorum or who died with Teïas in the cataclysmic battle of Mons Lactarius were doubtless the best warriors that the Goths could muster.  They had lost others still in the earlier disastrous naval defeat of Sena Gallica in the Adriatic.  That these defeats effectively ended Gothic resistance is considerably less surprising than the facts that it took three bloody engagements to do so and that some Gothic garrisons continued to hold out even then. 

The disappearance of the Goths from history after that point is easily encompassed within the sorts of dynamics discussed throughout this chapter, albeit in reverse, and indeed Goths continue to be attested in Italy after the completion of the ‘reconquest’.  Although primarily military in composition and function, the Goths had been more than simply an army when they crossed into Italy under Theodoric.  By the time of Totila’s and Teïas’ deaths sixty-odd years later, the Italian Goths had – unsurprisingly – changed in nature in many ways.  Their primarily military character had, however, endured throughout.  A kingdom created by the sword had perished by it.

[1] P. Heather.

[2] Halsall

[3] Cesa (1994b), p.310; See Variae 1.19 for its successor, the fiscus gothicus.

[4] The intervention of a Gothic army from Gaul was critical in breaking Odovacar’s siege of Theodoric in Pavia.  Whether this represented a sort of pan-Gothic cooperation is unlikely.  It may be preferable to see Alaric II following his uncle, Theodoric II’s, example and the Gallic faction chancing its arm in Italian politics in established fifth-century tradition.

[5] CTh 7.8.5 (dated 398)

[6] Ennodius, Epist. 9.26; Cassiodorus, Variae 2.16

[7] Goffart (1980), pp.73-80.  The loci classici are Variae 1.14 & 2.16-17.

[8] See Bjornlie, elsewhere this volume, for the straightforwardly fiscal connotations of the illatio tertiarum.  That the tertia in question related to the fiscal payment schedule simplifies the situation further still.

[9] Goffart (1980), pp.80-88.  The key text is Variae 5.27.

[10] See Variae 2.37

[11] Mommsen (1889), p.499, nn.3-4, had related millenarii to millenaeLot (1928), p.1003, and nn.5-6, thought millenarii were officers. Generally, however, it had been assumed that a millena was a fixed amount of land.

[12] See Variae 4.14, 5.26-27, 5.36.

[13] Goffart (1980), pp.89-100.

[14] W. Goffart, Administrative Methods of Barbarian Settlement in the Fifth Century: The Definitive Account.’ Gallien in Spätantike und Frühmittelalter Kulturgeschichte einer Region, ed. S. Diefenbach & G. M. Müller (Millennium Studien zu Kultur und Geschichte des ersten Jahrtausends n. Chr. Vol.43) (Berlin, 2013), pp.45-56.
[15] CTh 7,4.20, 22.

[16] That such a system for payment was employed in Ostrogothic Italy is demonstrated abundantly clearly by ET 126 and, especially, 144.

[17] Variae 4.32 says that the property of the proscribed belongs to the fisc.  The Edictum Theodorici specifies the fisc’s claim to take over the property of convicted criminals in some cases, where there were no heirs.  ET 112-3.

[18] Variae 1.18 seems to me to refer more easily to the distribution of expropriated land (and abuses of that situation) when Theodoric conquered Italy than to illegitimate claiming of tax-revenue.

[19] See Variae 5.36.

[20] The illegal retention (by his uncle) and management of the paternal inheritance of an adolescent Goth who is of sufficient age to perform his military service is discussed in Variae 1.38.  Here, the text could relate at least as easily to an inherited draft on fiscal revenue as to landed property.

[21] I am perplexed to see that Goffart continues to view my own modified version of his thesis as ‘expropriationist’, something it has very clearly never been.  Goffart, ‘Administrative methods’, p.48.

[22] Such a situation may well lie behind the situations described in Variae 1.26 and 4.14.

[23] Halsall, Warfare and Society in the Barbarian West, pp.000-000.

[24] Theodoric’s concern with effective and rapid communications is visible in Variae 1.29, 2.19, 4.47, 5.5, etc.

[25] The Burgundian Code (54.1) suggests that something like this was practised in that, smaller realm.

[26] This is well analysed by P. Heather,

[27] Variae 5.27: bonos enim laus malos querula comitatur.

[28] Edictum Theodorici 43-44 and 46 undermine the use of patronage to influence legal cases.

[29] Most clearly perhaps in Variae 4.36.

[30] Variae 3.38, 4.13, 4.36, 5.10-11, 5.13, 5.26, 6.22, 7.4.

[31] Variae 2.5, 3.41.

[32] ET 145.

[33] Variae 6.15. Cp. CTh 14.10.1.  The military identification of the donor/s of Variae 1.26 is suggested only by a reference to the soldier’s cloak (lacerna) in the last lines: ‘tribute is owed to the purple [i.e., here, the king], not to the military cloak.’

[34] A.H.M. Jones, The Later Roman Empire, 284-602 (Oxford, 1964), p.256; J. Moorhead, Theoderic in Italy (Oxford, 1992), p.254.  The statement is carelessly repeated, on the basis of these authorities, by Halsall, Warfare and Society, p.000.

[35] Procopius, Secret History 26.27-28, says that it was Justinian’s officials who disbanded these corps, which had been generously left in place by Theoderic, in spite of their uselessness.

[36] Variae 1.10, 7.3.

[37]Wolfram (1988), pp.316-7, referring to Variae 1.11 claims that the troops (milites) commanded by Servatus, dux of Raetia, ‘cannot have been Goths’.  Wolfram does not stop to explain why, but Heather (2003), p.118, n.89, nevertheless builds on his foundations (and mis-cites the source) to allege that Servatus is ‘said to have led limitanei (i.e. inferior quality troops)’.  Variae 1.11 makes no reference to limitanei or even Romans.

[38] Variae 7.4.

[39] Variae 1.4.

[40] P. Amory

[41] Above all P. Heather,

[42] Variae 8.21.  A full fluency in Gothic seems to me to be less necessarily implicit in Cassiodorus’ statement than a competent command of army-Gothic argot.

[43] Variae 1.38.

[44] See also Napoleon’s Old, Middle and Young Guards.

[45] Bierbrauer (1994b), pp.194-202.

[46] V. Bierbrauer, Invillino-Ibligo in Friaul.  Die römischer Siedlung und das spätantik-frühmittelalterlische Castrum (Munich, 1987).

[47] Eighteenth- and nineteenth-century commanders similarly ordered troops to attack with unloaded muskets when an advance was to be pressed briskly with ‘cold steel’.

[48] Like the Gepids of Variae 5.10-11.  It should be born in mind that late imperial units frequently bore ethnic titles.  The Gepidic origins of many at least of the troops discussed should not be doubted but one ought not to assume that these Gepids were any more a ‘people’ on the move than the regiments of Franci, Alamanni or Parthi of the late Roman Empire, similarly redeployed with their wives, children and camp followers.

[49] One should note the highly conservative political connotations of phrases like ‘incomplete assimilation’.

[50] It seems to me, again, to be politically revealing that the suggestion that the Gothic rank and file did not blithely do what their officers and social betters told them should be represented as a surprising and defining feature of Gothic society.