There are important differences as well. In Mosul the local elite expandedits influence during the civil war and tribal leaders came to dominate thepolitical life of the province until the arrival of the Hamdanids, themselvesleaders of a branch of the nearby Ban-f Taghlib. In Egypt, in contrast, thelocal Arab elite was destroyed by the civil war, leaving the country open to theoutside control which was to continue until modern times. In Mosul, the leadingfigures derived their strength from their tribal followings and land-holdings;their control of the rawibit was no doubt useful but not essential. In Egypt, onthe other hand, the wujiih derived their power from their control of the localjund which was paid by the government; when the caliph chose to disband thislocal militia, he also destroyed the power base of the local elite.

The picture which emerges from the study of these two areas is a complexone. In both Mosul and Egypt, however, we can see that the power of thegovernor was in fact severely circumscribed. His term of office was usuallyshort and he was dependent on the good will of local notables for militarysupport. Financial administration was sometimes in the hands of menresponsible directly to the caliph and the administration of justice was alwaysoutside his control. The leaders of the local community could make theiropinions felt through their control of important local offices and there weretimes in both Mosul and Egypt when local men were appointed as g

The key to the governors' weakness was that they were dependent ontroops raised locally and commanded by leaders of the local community, thewujith. Although chronicles mention the arrival of KhurasanI and Syrian troopsin the province from time to time 54 it would seem that most of these were intransit to Ifriqiya or were soon withdrawn. Certainly at times of crisis likethe 'Alid rebellion in Fustgat in 145/763, or the disturbances which led to thedeath of Misa- b. Mus'ab in 168/785, the governor had to rely entirely on locallyrecruited troops. This local militia could field an army of up to five thousandmen.55 They were paid an 'atd' or salary and sometimes created disturbanceswhen this was delayed.56 Though militant on the subject of their pay, theywere less impressive when faced with an enemy, especially when this enemywas composed of Muslim Egyptians with whom they had many links andinterests in common.

After the Sawad of Iraq, Egypt was the most important source of revenuein the caliphate and any decline in the tax-yield in the province was a subjectof acute concern to the government in Baghdad. Much of the policy of successivecaliphs towards Egypt and its governors must be viewed in this context.Sometimes the governor himself was directly responsible for the financialadministration of the province but from late Umayyad times, a separatesdhib al-kharaj was appointed by the caliph, answerable directly to him. Onoccasion the sahib al-khardj could be a more powerful figure in the provincethan the governor. Under Hisham, 'Ubayd Allah b. al-Habhadb secured thedismissal of two governors he disapproved of and was in the end allowed tochoose his own.44 None of the 'Abbasid financial officials achieved this dominantposition but they were still important figures in the province. In the first yearsof 'Abbasid rule responsibility for taxation rested with the governor, but in141/758 the new governor refused to give the necessary damdn or guaranteethat the required sum would be raised. Al-Mansfir then put Nawfal b. al-Furatin charge and the wdli effectively lost control of the fiscal administration.45Thereafter it would seem that the financial affairs of the province were run byseparate officials. The sources are not full enough for us to follow the processin detail 46 but it would appear that it was not until the appointment of 'Isa b.

The wujih of Fustat seem to have been a more homogeneous group than thetribal leaders of Mosul. There is very little mention of Arab tribal strife inthe history of early Islamic Egypt. This was at least in part because, despitean attempt during the reign of Hisham to settle Qaysis at Bulbays in theeastern edge of the Delta,39 almost all the Arabs in the province belonged tothe Yamani group, but it may also have been because the Arabs were a fairlysmall minority among the largely Coptic population and tended to keeptogether. This also seems to reflect a different balance between town andcountry. As we have seen, in Mosul, many of the local notables had close linkswith tribal groups in the surrounding country and they could and did call onthese groups to assert their authority in the city. In Egypt, in contrast, althoughthe wujiih certainly had estates in the country,40 these were probably cultivatedby the local Copts and could not be used as a source of military power. Thelocal Arab militia, not tribal follo

Egypt in the early 'Abbdsid period was a much more self-contained politicalunit than the province of Mosul.37 In Mosul events in al-Jazirah or Azerbayjanhad a noticeable effect on local politics, while Egypt was almost immune fr

It is difficult, even from a chronicle as full as al-Azdi's, to assess the relativeimportance of central government and local elites in Mosul at this time. It isclear, however, that the governor must have had to rely very heavily on localnotables for support and advice. Almost always he was a stranger and he hadprobably never visited the area before. He seldom, if ever, brought any troopswith him and was dependent for military support on locally raised men led byprominent local citizens. His terms of office was usually very short and veryseldom did a governor enjoy power long enough to build up any personalfollowing or power base in the area. Through their control of the office of qdpdi,at least for some of the time, and above all their control of the local armedforces, local leaders played a much bigger part in the affairs of their city thanwould appear at first glance.

Control of taxation was almost always in the hands of the governor, thoughour information on this point is far from complete. From 147-51/764-8 theqjd4, al-H.arith b. al-Jarfid, seems to have been in charge,20 while from176-80/792-6 responsibility lay with Minjdb, a maawul of the caliph,21 who wasprobably answerable directly to his master.

Finally, because of the significance of possession of land, the lack of acentralised political infrastructure, and the self-determination of localcommunities, political power was diffuse and indirect, based on influencenot control. Rule worked through an elite which stood at the apex oflocal communities, and bound them to the centre. This elite was largelydrawn from those communities, and even when it included newcomersdespatched from the centre, they had to put down local roots based onland and patronage. That is, although those with local power presentedthemselves as office-holders, to be effective they needed to build up localpurchase. Political power rested on brokerage and patronage, and con-tained a strong element of reciprocity.

Second, as a result of the centrality of land to politics, kings had onlya limited impact on local communities. Royal resources were essentiallylimited to the ownership of land, and were used to feed the royal house-hold, with any ‘surplus’ holdings being granted out, in a variety of forms,to aristocrats in return for political support. Although kings did maintaincontrol over tolls, there was no central finance system which tapped rou-tinely the agrarian surplus or funded activities beyond the royal house-hold. Where the Roman tax system survived, it did so in local hands, aslocal dues all but indistinguishable from rent. The absence of routinemechanisms for interaction between the political centre and the localitiesdid not, however, preclude considerable structural power on the part ofthe centre. This structural power was created by the ability of the centreto intervene in the localities on exceptional but essential occasions. Thecentre could thus manipulate and canalise essentially local processes toensure that its strategic goals – the recruitment and supply of armies, themaintenance of roads, bridges and palaces – were met. This is preciselythe point that those who argue for a maximalist view of early medievalstates miss. Rightly impressed by the organisational capacity of earlymedieval kings, they assume that it must have been achieved through thetypes of administration with which we are familiar, when in fact it restedon an entirely different style of consensus-based politics which workedthrough extant social mechanisms. The Carolingian capitularies need to

hatever its effects on the peasantry as represented by Shenoud orChrysostom, the emergence of these large estates is likely to have made apositive contribution to the wider economic development of the EasternEmpire at this time. In recent years, archaeologists and numismatistshave become increasingly aware of the extent to which the period fromthe fourth to the early sixth centuries in the eastern Mediterranean wasassociated not only, as already seen, with demographic expansion, butalso, more generally, with economic growth."'1 That this expansion ineconomic resources was associated with a concentration and rest-ructuring of landownership should not occasion surprise. For, asemerges with particular clarity in relation to the properties of the Apionfamily, the aristocratic estates of the late antique east were emphaticallynon-autarkic enterprises. Rather, production on the great estates washighly commodified: labour was rationally and flexibly organized, withworkers being directed between estate properties; a certain amount ofspecialization would appear to have characterized the holdings which the

From an imperial perspective, this obligation to reside in perpetuityon the estate resulted from the fact that the peasant had agreed to beenrolled (thus 'adscripticius') on the estate's tax register. Demandingthat the peasant, and his heirs, remain on the estate in perpetuityrepresented an attempt on the part of the imperial authorities tointroduce an element of fixity and predictability to the collection oftaxes. It was typically, although not necessarily, through the estate thatthe colonus adscripticius was meant to pay the imperial taxes to which hewas liable, a practice known as estate autopragia or self-collection. Theadscription of an agricultural worker and his family resulted from a prioragreement made between the worker and the landowne

In the absence of documentary sources, however, the most suggestiveevidence is yet again provided by the legal texts. On the basis of theextant papyrological evidence, it is apparent that by the sixth century themainstay of the agrarian workforce employed on the Egyptian bipartiteestates, or at least, that portion of the workforce resident in the estateepoikia, bore the imperial legal designation of coloni adscripticii, that is,agricultural workers legally bound to reside on the estate in perpetuity,an obligation also incumbent upon the workers' heirs and descendants.This is at its most evident from documents acknowledging the receipt byan estate employee of a capital item or loan furnished by the landowner,and the contracts of surety contained in the papyrological dossiers, inwhich the workers are typically styled enapographoi georgoi, the Greekequivalent of the Latin legal term. By virtue of the fact that thesedocuments established potentially actionable agreements, those draftingthem would have been careful to record the precise legal status of theparties concerned.105

The imperial legislation on agrarian patronage records that, as aresult, agreements began to emerge between individual peasants or ruralcommunities and powerful local patrons holding imperial office,whereby the patron would either agree to intercede on behalf of hisclient with the imperial authorities to negotiate a lower rate of taxation,or would take it upon himself to directly alleviate his client's fiscalburden. From such initial arrangements, the practice developed bywhich the patron and client conspired to pretend that the localpotentate, rather than the lesser landowner, was the legal possessor of theland on which the imperial taxes were levied. In the short term, thisprovided the client with the means of escaping the land tax; whilst in themedium term, it allowed the patron a chance to extend his own estates,by turning a fictional transfer of propert

The fourth century thus saw the emergence in the eastern Mediter-ranean of a new imperial aristocracy of service, the leading members ofwhich were enrolled into the senatorial order. Members of this new classare recorded in the sources not only as figures of extraordinary influenceand power at court, but also in the localities of the empire.82 Inparticular, from the mid-fourth century onwards, the legal sources bearwitness to the process whereby, not only in Egypt, but throughout theEastern Empire, the wealthier and more powerful members of this newelite forced aside their social competitors and won mastery of locallanded society, restructuring agrarian social relations after the mannerrecorded in the Egyptian papyri. This is at its most evident from theimperial constitutions on agrarian patronage (de patrociniis vicorum)recorded in the Theodosian and Justinianic Codes

As Keenan has noted, the prosopographical testimony of the papyriwould suggest that not only was the scale and structure of elitelandholdings undergoing a transformation during the fourth and fifthcenturies, but so too was the social character of the elite itself. Soon afterConstantine's defeat of Licinius in 324, one begins to find mention in thepapyrological sources of individuals holding imperial civil and militaryoffice bearing the name Flavius - the principal gentilicium, or familyname, of the Emperor Constantine himself. This title was apparentlyadopted by them upon their entry into imperial service.71 As the fourthcentury progressed, these Flavii can increasingly be seen to have come todominate the higher echelons of imperial and civic government.72Concomitantly, they appear to have emerged to the fore of local landedsociety.73 To Banaji, this phenomenon can be seen to have representedpart of a process whereby a 'bureaucratic elite was consolidating its socialdominance'.74 It is a history that one can see neatly encapsulated in therise of the Apion family. As noted above, from a fifth-centurybackground in imperial service, associated with the management ofimperial estates in the vicinity of their native polis of Oxyrhynchus,members of the family can be traced through to the sixth century andthereafter, holding some of the highest offices the empire had to offer,whilst the family's own properties expanded and developed in theneighbourhood of Oxyrhynchus and beyond.75 The Apion family wasthe embodiment of the emergent trans-regional landowning aristocracyof service of late Roman Egypt.

Certainly, by the mid-fourth century, the fiscal landlists preserved forHermopolis reveal the existence of a dominant stratum within local civiclandowning society, the members of which far outstripped in terms o

his process would appear to have gone hand-in-hand with anincreasingly marked concentration of landed wealth within Egyptiansociety. Families such as that of the Apiones in the sixth century weretrans-regional landowners, owning property not only in the vicinity oftheir home town, but also elsewhere in Egypt and beyond. Thus inaddition to Oxyrhynchus, the Apion family owned property near otherMiddle Egyptian cities such as Cynopolis and Heracleopolis, as well asurban property in Constantinople and Alexandria, and possibly evenagricultural estates in other regions, such as Sicily.58 By contrast,members of the civic elite of Egypt in the third and early fourth centurieshad tended to limit their possessions to the territorium of their nativecity.59 This process of concentration had a long antecedence. The socialand economic influence of larger landowners had been bolstered at thebeginning of the third century through the introduction throughoutEgypt of municipal-style government on the traditional Greco-Romanmodel.60 The enrolment of the dominant local landowners into the newcuriae had provided them with the opportunity to entrench and extendtheir economic interests through deploying their imperial connectionsand affinities against their neighbours and social inferiors.61 Diocletian'sreorganization of Egypt at the beginning of the fourth century hadfurther strengthened the administrative hold of the civic elite over thecountryside.

Large estates of the Apion type were not, one should note, a historicalconstant within the Egyptian agrarian economy. Whilst Rathbone hasrevealed the existence of a number of third-century large estatesstructured after a manner closely analogous to that of the Apionproperties some three hundred years later, the landowning elite of Egyptin the third century would appear to have been far more reliant thantheir sixth-century counterparts on the straightforward leasing out ofland: that is to say, members of the landowning elite were primarilyrentiers.55 By the time one reaches the fifth century, however, suchevidence as exists would suggest that directly managed large estates werebecoming increasingly common, indicating a fundamental restructuringof agrarian conditions. This is evident from the toponymic record, withthe word for village proper (kome) coming increasingly from t

The relative lack of understanding of the agrarian economy of the lateRoman west is, of course, a result of the paucity of extant sourcematerials for the period, and, in particular, a marked absence ofdocumentary evidence. Until the appearance of the late M

Nor is he very convincing onthe expenses of the state, the major scale of which in both the Romano-Germanic and theCarolingian periods is invoked to justify his hypotheses about taxation (pp. 122ff, 22 Iff). Headmits that secular city spending declined (pp. 128-9, 134-6), and, somehow, vanished (pp.

It is evident from this brief characterization that the Byzantineempire managed to preserve the dominance of the ancient mode,despite the continued existence of the feudal mode, until well afterIooo. The only problem that arises is one of categorization. One ofthe keys to the ancient mode is the city/country relationship, withthe city strengthened by its fiscal powers. But in the seventh andeighth centuries the urban society of the Byzantine east collapsed.The exact motors of this are almost totally obscure, but its collapsewas in large part the price paid for the survival of the state at theexpense of the civil aristocracies, the bases of urban life. Theirsurvivors concentrated themselves in Constantinople; the state aban-doned all pretence of taxing through cities, and organized the processdirect, with a centralization of authority far surpassing Diocletian's -on a rather smaller scale, however. The empire could be seen ashaving become one giant city-state concentrated on Constantinople.But we must recognize that we are here touching the very edge ofthe arena within which the concept of the ancient mode helps us tounderstand how a society works. The tax-raising basis of the central-ized Byzantine state fits quite as easily into a more eastern patternwith the (partially imitative) Arab empire of the seventh to tenthcenturies and, indeed, the Sassanians in Persia before it and consider-able variety of states after it, reaching through the Ottomans andSafavids to the present day. These states, like those of the Romansand Byzantines, had the problem of balancing tax-raising with land-ownership, but often over very wide areas and without the institu-tional mediation of cities. They were very often quite successful too.The problem of how these systems fit into the category of the ancientmode of production is scarcely worth posing: they evidently do not.But the differences between them and, say, the social formation ofthe Roman empire seem to be differences in degree, no

As argued above, the Byzantine aristocracy was pe

The first problem that has to be faced here in making comparisonswith the west is whether the two histories are strictly comparable.The fifth century saw the overrunning of every part of the west; theseventh century in the east at least left the Byzantines with Anatoliaand the Aegean. But, as already noted, the point about the Germanicstates is not that they replaced the western empire; it is that theyultimately failed to reproduce the state power of their Roman prede-cessors. The tax-raising state continued in the east both in theByzantine and Arab parts of the former unitary empire. Not onlythis, but the seventh to eighth centuries in Byzantium appear to showan eclipse of aristocratic power. The state patronized generals andtheir armies, at the expense of the local civil aristocracies; the latterthus lost their independent role to new state subordinates who wereinitially more reliable, and indeed more useful. The old noblefamilies disappear from our sources; it is not until the ninth and tenthcenturies that they (or, more likely, the new military landowningfamilies) return in the texts to trouble the smooth functioning of themechanisms of government. In the power-struggle between state andaristocracy at the moment of crisis, it was the aristocra

It is possible, even likely, that the chief social group that benefitedfrom the fall of the Roman state and the transition to feudal societywas the peasantry. It was not, at any rate, the Roman aristocracy;the new Germanic states had their own ethnic aristocracies. SomeRoman families changed their names and began to command arm-ies - that is to say, became German; but most others becamepolitically marginalized, with the noteworthy exception of those insouthern France, easily the least Germanized area inside the successorstates, where the Roman aristocracies maintained their hegemony formany centuries. The end of the Roman state was in the long-terminterests of the aristocracy as a class; but not always in those of theindividual families involved in its breakup. The peasantry, however,were almost certainly better off; the mechanisms of surplus extractionwere in the seventh to eighth centuries less efficient than they hadbeen in the fourth. Rent-taking in the empire was conditioned bythe fact that tax took much of the surplus; the aristocracy neededtime to catch up with the possibilities engendered by its absence.This assertion is, it must be said, totally speculative; but it wouldgo a long way to explain the apparent poverty of the early medievalaristocracy - and even, sometimes, of kings. People built smallerand cruder buildings, wore simpler clothing, bought fewer luxuriesfrom the east. I do not think this can be explained, as it often is, bythe idea that peasants produced smaller surpluses than under theempire; no economic or social mechanism has ever existed which canexplain why political changes can produce a permanent productivedecline on the part of a subsistence peasantry. What must havehappened is that peasants kept more of it for themselves. And thenot insignificant class of peasant owners which had survived the warsand patronage agreements of the fifth century will have found thatvery little surplus was demanded of them at all; instead, the Franksand Visigoths, at least, expected them to serve in the army them-

The institutional weight of ideology showed itself most grandlyon the level of central government. There is no doubt of the publiccommitment and ambitions of the Carolingian state, and of thewidely diffused hegemony of this central commitment at its heightunder Charlemagne and for a time under his son Louis the Pious inthe late eighth and early ninth centuries, emphasized by the strikingeducative and propaganda impact of the Carolingian renaissance andfuelled, it may be added, by the almost Roman wealth of the kingsat the point of their greatest military success. This commitment, thesense of the public nature of the state, of officialdom, of politicalresponsibility, is almost purely Roman, and it says much for theresidual authority of the Merovingians and the memory of sixth-century Frankish royal power (as well as the force of Roman valuesin the church) that it could have survived in Romano-Germanicnorthern Gaul, of all places, and even extended its influence toAnglo-Saxon England. (The Lombards had maintained it too withlittle difficulty, and in all the Carolingian empire it was most firmlyrooted in Italy.) The only Germanic feature of the state itself (andeven this had Roman elements) was its conscious links with all thefree men of the kingdom, nominally the descendants of the warriorsettlers of the fifth to sixth centuries; this was certainly a key to thestate's legitimacy, but not to its conception of its functions. TheCarolingian state obtained a wide measure of consent from its aristoc-racies at the point of its maximum success and for surprisingly longafter.30 But this success depended directly on such consent. Whenthe late Roman state lost consent, it eventually crumbled, but theprocess was long drawn out and highly mediated, for the statewas based on a direct economic process of surplus extraction. TheCarolingian state, however, was based on land, as were the upperclasses; a Carolingian noble's personal economic power was thusfounded on exactly the same basis as his king's. The only way thekings could exert power was through obtaining and trusting theloyalty of their aristocrats; they had to buy it. Initially they coulddo this in exchange for offices, which still brought a lot of status foraristocrats, along with land; but increasingly, when the incessantwars of the eighth to tenth centuries once again undermined the

s. To repeat points made earlier: the wars tipped the balanceby challenging the state head on; the state was less advantageous tothe aristocracy as a protector and a source of profit, and its ideologicalhegemony, as the natural and inevitable focus of political activity,was put into question. As landowning (the feudal mode) was alwaysthere as the most solid element in Roman society, the aristocracycould retreat into it. With even the aristocracy wavering, thepeasantry had the opportunity to react as well, underpinning aristo-cratic actions and inactions. By the time the Germanic settlementeventually came, the dominance of tax-raising was breaking up. Itmust, however, be emphasized that this is not an explanation forwhy the empire was replaced by Germanic successor-states; that wasprimarily a political and military problem (though the revenuesavailable for the Roman army, as well as the preparedness of Romanpeasants to serve, had something to do with it). It is, rather, anexplanation for why, when this did happen, the successor-statesfailed to take the form of Roman states in microcosm, as in theorythey could easily have done and as Ostrogothic Italy for a timeperhaps did.25 The Germanic hierarchies in each kingdom werecertainly romanized enough (in social, if not cultural terms) to haveaccepted such a system. It is because the tax-raising mechanisms ofthe empire, the basis of the ancient mode, were already failing thatthe Germanic armies ended up on the land. The German aristocraciesexcluded many members of the Roman aristocracy from state power,and therefore often replaced them as patrons; but they too establishedthemselves as a result, not as officials but as landowners. The impactof war had exposed the contradictions existing in the heart of Romansociety in the west, and one mode gained dominance over the other.The motors of such a conjuncture are not unknown elsewhere: Russiain I917 has parallels

The dominance of the state as the ancient mode was directlyexpressed through its organization of social stratification. The ex-ploitative force in the Roman state was the public power; status wasimportant precisely in that it regulated access to this power and thus,in its upper levels, to the resources of taxation - as well as, at theother end, the obligation to pay it. We have already seen the statecontrolling the latter through the tying of the peasantry; this certainlyslipped out of the control of the Germanic states. But the state's lossof control over aristocratic status makes the process clearer. In thefourth century, hierarchy and status were legally taut concepts,linked at their upper levels directly to state office-holding (or elsesenatorial office-holding, theoretically part of the state, but alreadyperhaps partially drifting out of governmental control). The cate-gories most directly linked to wealth on its own were extremelyvague (for example, honestior and humilior); it was the network ofofficial titles, the categories most useful to the state, that stratifiedaristocratic society. In the sixth century, outside Ostrogothic Italyperhaps, they did not. The complicated terminology for late Romanoffice-holding and senatorial hierarchy had disappeared. Gregory ofTours uses the word senator for any major Roman landowner. Eventhe very Roman-looking rivalries for city office that he describes inhis histories are expressed largely in terms of the power-relationsand patronage of landowners.23 Landowners were seeking office and

There is no doubt that the Merovingian kings levied the land taxfor a long time. Taxation and its problems are a common motif inGregory of Tours's works in the later sixth century, and in seventh-century saints' lives. Theudebert I's tax-collector, the Roman Par-thenius, was killed by the Trier mob at his king's death in 548.Chilperic I's attempts to exact new and increased taxes brought himan uprising in Limoges in 579 and (Gregory says) his children'sdeaths from plague in 580. Gregory himself in 589 defended the taxexemption of Tours that his predecessors won for the city, but thebishop of Poitiers in the same year found it necessary to get thePoitiers tax registers revised to rectify the over-taxation of widowsand orphans. Tax was, then, still perceived as heavy. It was alsouniversally unpopular. Bishops tried to get exemption for their cities,and abbots did likewise for their monasteries, usually successfully.The seventh-century saints' lives underline this: not only increasedtaxation excites the wrath of the saints, but any taxation at all.Already, however, the sketchy indications we have indicate that taxlevels had fallen dramatically from the Roman period, to under 10per cent of crop. Globally, the economic dominance of taxation hadvanished. And popular assumptions about the legitimacy of taxationwere utterly changed too; even a tax level as relatively light as thiswas unacceptable. The Merovingians were strong and taxed as longas they could, which is to say throughout most of the seventh centuryat least. But they could not hide the fact that taxation no longer hadany purpose except the exaggerated enrichment of the kings; thismust indeed explain its ever-decreasing legitimacy. There was barelyanything to spend it on any longer. The army was landed; theadministration (except the tax collection mechanism itself) was rudi-mentary by Roman standards; the vast fiscal lands which the kingscontrolled were enough for their everyday needs. The only thingthat the tax system was good for was to give away in gifts, particularlyas exemptions to the church, for short- (or long-) term political gain.But in doing this, the Merovingians were already speaking thelanguage of feudal social relations. The land tax became simply onepart of the resources of the fisc, like an estate or a toll; the Merovingi-ans gave them away indifferently. By the Carolingian period all thatwas left of the land tax was a set of fragments wi

The new Germanic states were not yet feudal. The controllingoligarchy in each of the successor-states sought to maintain thefinancial mechanisms of the empire as far as they could. This at leastshows that no aristocrat who accepted German rule, more or lessreluctantly, could have done so with the expectation that this alonewould have meant the end of the tax-gathering functions of the state.Nearly all the states in the west in A.D. 500 levied taxes: the Vandalsin Africa, the Visigoths in Spain and southern Gaul, the Ostrogothsin Italy, the Burgundians and Franks in south-east and northernGaul. (By now we know nothing about Britain.) Such taxation wassuccessful according to the measure of the internal strength of the

empire. There is evidence, at least in Gaul in the460s, of actual political disloyalty by some major political figures.(Peasants too made this choice often enough for it to become a clicheof the period.) More often, however, the same result was producedless consciously, through the unintended consequence of regionalself-interest and factionalizing and through the surrender of politicalleaders to what now seemed the inevitable German victory in orderto protect their private interests. None of these reactions would havehelped the preparedness of the aristocracy to pay tax. The state grewweaker by the same measure. It was short of money from the verystart of the fifth century, but matters got worse. In 444-5, ValentinianIII, in one of the most indicative laws of the late empire, confessedthat the "exhausted circumstances and afflicted condition of thestate" made it impossible for him to pay the army, and that he feltit impossible to increase the land tax - he put a tax on sales, instead,though how much it raised is more doubtful. The Va

Contradictions do not necessarily get pushed to the point wheresomething breaks. Tax evasion in the east did not lead to the collapseof the state. The difference in the west was provided, as I have said,by the Germanic invasions. These were essentially an external,almost contingent force; but they cracked the structure of the state.Indeed, they defeated it militarily, at least in the Vandal conquestof Africa after 429 and the Visigothic-Frankish take-over of Gauland Spain after the 460s. The fifth-century wars kept the armysufficiently occupied to make mass tax evasion a politically practic-able activity too. But initially the barbarians caused a crisis ofideological hegemony, from which much of the rest stemmed. In theearly fifth century, writers for the first time begin to give the impres-sion that the duration of the Roman empire might be finite; hardlyever, even in the third-century invasions, had they done

The key to this is contained in the growth of private patronage,patrocinium. Patronage was an old relationship everywhere in theMediterranean, but as a serious problem for the late Roman state itbegins to crop up in our texts from the 360s onwards, in Egypt andSyria, and then in the 440s in Gaul. Peasants were beginning to enterthe clienteles of rich neighbours in order to avoid having to pay theever-increasing taxes. A whole section of the Theodosian Code dealswith this process; landowners offering such patronage are to be liablefor the unpaid taxes, and the relationship is made void. This we readin six laws dating from 360 to 4I5; the last (a law for Egypt) finallyconcedes the reality of patronage, but, a little hopelessly, insistson tax-paying nonetheless. An oration of Libanius from the 380sdescribes villagers in Syria actively seeking out military protectorsto avoid tax-paying (and in fact in the case of Libanius' own tenants,to avoid paying rent too, though this is a different process, thesupersession of one aristocratic elite by another). These are all easternexamples, not western; they show that this particular crisis was notconfined to the west. For the west we have Salvian, writing a religioustract against the times in the 440s in Gaul. Salvian rails, in strikingrhetoric, against the inequalities of taxation. The poor have to facemore supplementary taxes than the rich, and are the last to benefitfrom rebates and the cancellation of arrears. Taxation forces men,even the educated, to flee to the barbarians or to the Bacaudae. And,still more, it forces the poor to give their property to the rich inexchange for patrocinium, protection against tax-paying, and receiveit back as tenants; worse yet, they then find they are still liablefor the taxes. This text is not the less clear for being rhetorical:independent peasants are prepared to become tenants rather thanpay taxes. They do so, presumably, on the assumption that theirpatrons/landlords are going to be powerful enough, whether insidethe state or outside it, to evade these taxes. When this is not true,the peasants' hopes are greatly deceived, for they end up paying taxas well as rent, but this last twist is less important than the majorpoint: that rent-paying is to many peasants preferable to tax-paying.This is not surprising if Jones's tax:rent ratios are applicable here(as by the fifth century they probably are), for tax actually takesmore than rent. But at the least it means that at a moment of relativecrisis in Gaul (for there was war in Gaul throughout this period,though not going too badly for the empire), peasants as well a

The particular vulnerability of the ancient mode lay in its relation-ship to private landed property, in this instance the feudal mode,though the same problems had posed themselves less drastically inthe period of the rise of the slave mode. The state gave considerablewealth to those who controlled it, thanks to taxation, but in aneconomic system as undeveloped as the ancient world even at itsheight, there was not much that could be done with this wealthexcept put it into land. As the rich obtained land, however, theyalso obtained tax liability. Their private interests as landowners werethus in contradiction with their interests as rulers and clients of thestate. If their lands were large, their private interests outweighedtheir public ones. And although the financial resources of the statewere still a powerful focus of loyalty through their potentialities forenrichment, the direct commitment to private ownership of propertytended to be a firmer force than the more mediated opportunitiesoffered by control over state resources. The rich began systematicallyto evade taxation. The structures of the feudal mode were, in otherwords, more solid than the rival structures of the ancient mode, for

espite the weight and centralization of the state, it did not existon its own as the sole focus of public power and wealth; it was firmlyanchored in the cities of the empire. The empire had always been acellular structure based on cities and their territories (and creatingthem where necessary, in Gaul or Britain, for example). Earlyimperial municipia were in theory sovereign, with their own localsenates (or curiae) and tax-collecting mechanisms, and with theirown local aristocracies and public building programmes and localpatriotism. These cities and their elites dominated their rural territ-ories in the economic as well as political aspects that formed theancient mode of production. All Diocletian and his successors didwas to partially regularize and vastly increase the taxes that suchurban elites took, sometimes at the expense of the elites themselves.The members of the curia, the curiales or decuriones, were stillresponsible for tax-collecting (except for central government superin-dictions), and had to underwrite uncollected taxes. They complainedoften; and modern tears have often been shed at the plight of thecuriales, ground down with relentless taxation. Such tears are out ofplace; many curiales actually did quite well out of tax-collecting, inwhich the opportunities for self-enrichment were so vast, despite thedangers posed by the central government tax-collectors, who coercedthem and cut into their profits. But the central collectors were few.The empire was large; the state could not collect most taxes exceptthrough civic officials. And though cities had lost their position offinancial and political independence, an ultimately superstructuralchange, they were still the financial exploiters and foci of their ownterritories, and much money stayed in cities as a result of tax-collecting, both unofficially and officially. It is this urban focus forsurplus extraction that is the clearest sign that it is still useful to callthe taxation process the ancient mode of production. Each city was

Taxation dominated the economy and was the economic foundationfor the state. Nothing in the late Roman economic system escapedthe state's embraces. Long-distance commerce, for example, wasvery largely dependent on the state as a customer, as well as beingstrongly conditioned by regulations and often by requisitions thatserved the state's interests. Jones showed this very clearly, and hisanalysis, though underestimating the size of late Roman commerce,still stands. Commerce and the state continued to keep a closerelationship, that is to say of state dominance, up to the Carolingianperiod and well beyond; state patronage could always bring a mer-chant far greater wealth than anything as ordinary as commercialprofit.1

The relative quantity of taxation varied, though after Diocletianit was always high. But the quantitative dominance or near-domin-ance of taxation as a mode of appropriation of the surplus must byitself have integrated the late empire into a single social formation,despite considerable regional differences. Taxation coexisted withother modes, certainly - we have just seen its close coexistence withrent, the feudal mode - but it soon outweighed them. And this isseen even more clearly in qualitative terms; tax, and through it thestate, came to dominate the whole structure of the economy. Thesocial relations of production were aligned not with the interests ofthe landlord but with those of the state. This is best shown by thestate's concern to tie peasants to the land. Landlords had tried tosubject tenants in this way in the early empire, through debt bondageand forced renewal of leases, probably with some success despitethe intermittent hostility of the state. (The hostility was perhapssurprising since the same problem often arose on state lands.) Whenit was in the state's interest to force peasants to remain where theywere and get taxed, it did so through massive bouts of legislation.Not that peasant proprietors and more independent coloni were oftenin practice bound by such laws; as with the similar laws tying artisansto their professions, there was widespread evasion. But there is nodoubt about the seriousness of the attempt made by the state, at leastat its height, to exercise control over the most subject peasant strata.It may even be that the fourth-century state sometimes exercised morecontrol over the lives of dependent peasants than some landlords didthemselves. Exactly how this would have affected the work-processis less easy to determine. The state certainly exacted labour service,which was only very rarely required by landlords in the Romanworld. But in general the effect may have been slight. It is importantto remember that, apart from the slave mode, all exploitative pre-capitalist modes are based on peasant agriculture; the work processof the peasantry, and even their productive forces, are not necessarilyaffected by changes in the appropriation of the surplus (and thusthe social relations of production), although the whole mode ofproduction will be different if these do so change. As we shall see,

5. The Antaeopolis figures show us tax assessments in kind andmoney totalling between a quarter and a third of average gross yields,and thus between half and two-thirds of the whole surplus normallydue from tenants in Egypt (50 per cent being the commonest rent,the landowner paying the taxes out of it). Tax is thus equivalent toup to twice as much as rent. At Ravenna the tax:rent ratio is explicitin the text, for the landlord had to collect both and pass the tax on;the ratio comes to 57:43. This is a lot of tax. Landlords in the sixthcentury were getting to keep under half the surplus. Exactly howrepresentative the figures are we of course cannot say, but Italy andEgypt are certainly not among the provinces where rent is regardedas having been light; even if tax was higher in Egypt than elsewhere,as is possible, the relationship between tax and rent should not beregarded as unusual. The quantitative dominance of tax-collectingin the empire, even where it was in opposition to rent, is as clear asit is ever likely to be, given the figures usually at our disposal forthe late empire. Independent peasants paid at least the rates outlinedin these texts (the Antaeopolis figures are for owners and tenantsalike) and often more, for the Ravenna text is for partially privilegedchurch land. At such levels, over a quarter of gross yields will havegone in tax - at a guess, often over half the surplus (that is, afterseed and subsistence), and certainly over Ioo per cent in bad years.The figures are exceptional, but there is no reason not to take themseriously; some tax may never have been paid, but, equally, corruptcollectors in other places are known to have extracted more than thetheoretical norm. However, we must ask when these tax levels camein. The Antaeopolis register, if the date is right, precedes the taxrises which were intended to pay for the wars of Justinian's reign,and probably represents a tax level typical for some time. On theother hand, early fourth-century tax levels, at least in Egypt, wereprobably somewhat lower. The rise of an already high land tax tothese remarkable levels almost certainly began to occur in the laterfourth century with the beginning of the wars, and in the case ofEgypt, the growth in the population of Constantinople; perhaps onlythen did tax begin actually to surpass rent.14

in.13 The land tax was at first takendirectly from all free cultivators, or from their landlords if thecultivators were slaves. The institutional process was thus totallydistinct from that of rent-taking, even where the cultivator was atenant. Only from the 370s, perhaps, did tenants begin to pay taxthrough their landlords if they owned no land independently; in thefifth century such tax-paying through owners, rather than possessors,of land became generalized.

The weight of taxation in the late empire is well known, and oftenused as a standard formula in discussions of why the empire fell.But taxation was not just heavy and extremely burdensome; it wasthe basis of the state and the key element in the whole economicsystem, the institution that determined the direction of the economyand defined the dominant mode of production, which can still becalled the ancient mode. It has been said that the ancient modedominated the grouping of modes current in the late republic, onlyto be displaced by the slave mode in the period from the first centuryB.C. to the second century A.D.; if so, it now dominated again. Indeed,as we shall see shortly, despite the centralizing tendencies of the lateempire, tax was still raised through individual cities. But tax-raisingin complex societies can rarely exist in a vacuum; other modesof exploitation tend to coexist and their correlation is of crucialimportance. The correlation and its dominance by tax in late Romecan and must be analysed in a number of ways, for we have to seehow the late Roman social formation was constructed from it if wewant to understand how it fell.

The starting-point for our analyses is the late empire, the so-called"Diocletianic state" of the late third century onwards, the great age,the final triumph, of the Roman state. We start, that is to say,when the slave plantations of the first century had already virtuallydisappeared, though some may have continued here and there.10Instead, de

The feudal mode, the other one that concerns us, has in muchtraditional Marxist analysis been seen as based on serfdom and thecoercive political authority over tenants constituted by the seigneurie;Hindess and Hirst regard this as too narrow, and show, rightly inmy view, that feudal relations are represented simply by tenantspaying rent to (or doing labour service for) a monopolistic landownerclass; such landowners will always, while the system is stable, havethe non-economic coercive powers necessary to enforce their control,whether informally or through their control of public or privatejustice, but these powers do not have to be formally codified in theseigneurie to exist. (The authors present all this as a revolutionaryinsight, though it has long been perfectly well known to medieval-ists.) It should not be necessary to add that feudalism here hasnothing to do with military obligations, vassalage or the fief.6

out that of Barry Hindess and Paul Hirst in their book, Pre-CapitalistModes of Production. These two authors draw a distinction betweenthe ancient mode of production and the slave mode which we willfind useful, and they also widen some standard definitions of thefeudal mode. The ancient mode, in its most traditional ideal type(in the mythical early Roman republic, for example), was non-exploitative, and characterized by the control of a city-based citizenbody over the immediate countryside; the citizens were privateowners, but they co-operated in their control of the public landedwealth of the city. As Rome expanded, two developments occurred.The theoretical egalitarianism of the city broke down and the slavemode began to displace the free-owning peasantry, reaching by thelate republic its classic form, the plantation slavery of Cato andColumella which dominates the sources for the agrarian history ofthe second century B.C. to the second century A.D. But also, as Romeconquered the countryside and cities of Italy and the Mediterranean,the ancient mode itself changed in type, becoming an exploitativemode; the public wealth of the city, initially in land, came to be intribute or tax taken from proprietors in the subject countryside and,in the case of Rome itself, other subject cities. This graduallydeveloped into a wholesale taxation network, with the old city/country relationship as its inner structure, as we shall see. It is thisnetwork that I will call the ancient mode in its class form. It will bea key to my analysis of late Rome

notes. Bloch pointed out the tremendous increase inthe numbers of slaves during the great wars of the fifth to sixthcenturies A.D., but showed how they were not enrolled in the tradi-tional slave plantations characteristic of Augustan Italy; these slavesbecame tenants. At some point the plantations had broken down andslaves were put out in tenant plots; as the position of free tenantsdeclined, serfdom was born from the fusion of these two socialgroups; feudalism ensued. Broadly, this analysis is quite correct; butit presents, or seems to present, some problems. In particular thewhole pattern keys in very badly with what is known and generallyaccepted about the rest of late Roman history. If feudal socialrelationships already existed by A.D. 300, then what was the lateRoman state? If the latter was not feudal, as it does not seem to havebeen, then what filled the gap, and how? Moses Finley, who oughtto know if anyone does, has declared himself defeated: "I am unableto fit late antiquity into any neat series of stages", he says at the endof his most recent book, but "slave society did not immediately giveway to feudal society".4 Finley's picture of the slow crisis of slaveryfits in interestingly with recent Italian work on the slave mode (oftenposed in explicit opposition to him) to provide a firm picture of oneside of the problem, namely what happened to slavery in the secondto third centuries A.D.; but we cannot look at this here. The mostimportant result for us is instead that the slave mode can be left outof our arguments; there is no reason to regard it as having beengreatly prominent in the late empire at all.5 For what replaced it,what is principally needed is a tighter analysis of the modes ofproduction of the ancient world

To go back to the empirical questions posed at the beginning of this paper, foraristocrats and peasants were these centuries a further episode in the continuinghistory of a society whose fundamental structures remained the same, or was thisan age of revolutionary change, disguised behind superficial continuities in thecultural and political façade? On balance it seems to me that the evidence pointsto revolutionary change. The cultural and political façade certainly continued inmany ways as before, but a combination of archaeological evidence that fails toshow any sign of a continuing late antique landed elite in Byzantium’s Anatolianheartlands with the evidence of legal texts that show a new and peculiar interestin the details of rural communities suggests to me that something fundamentalhad changed.

It is unfortunate that one can say so little about the archaeology of Byzantinerural communities, and this is where the lack of a system of obligatory rescuearchaeology in Turkey has its effect. By contrast, recent work in Spain has beenable to draw on the large quantities of information that rescue archaeology hasnow made available, and it suggests approaches to the agrarian history of theearly middle ages that may be relevant to Byzantium. As the central state fellapart in Spain in the early fifth century, so the villas that were estate centres forthe landed elite were abandoned, cities decayed and the complex late Romaneconomy began to unravel. What eventually emerged, and this is where the dataproduced by rescue archaeology comes to the fore, was in many areas of Spaina dense network of villages. The archaeological evidence is quite clear that noestate-owning elite of any kind had their residence in these villages before theninth or tenth centuries. The villages were peasant communities that had emergedto service the needs of a rural world effectively top-sliced by the departure offormer elites. Eventually new power structures emerged in Spain, and the villageswere brought within them. Eventually too new elites emerged, and from theninth century these would gradually impose their control. But between the fifthand eighth century much of rural Spain appears from this evidence as a funda-mentally peasant society (Castillo and Guirado 2006).

The impression given by the Farmer’s Law is reinforced by the so-called LandLegislation of the tenth century, a series of imperial laws aimed at defending therural community from the encroachments of the powerful (McGeer 2000). It hasbeen rightly pointed out that in some respects the laws seem more concerned toharness the powerful than to protect the poor, and that we should read theselaws above all in the context of tenth-century politics (Morris 1976; Whittow1996, 342–3); but the fact remains that emperors and their closest advisorsperceived the protection of poor farmers and maintaining the integrity of theircommunities as an imperial duty essential to the well-being of the state. Thatcannot be read as simply a topos of Christian kingship. Kings in the Latin westhad no such concerns. Rather it reveals a distinctive feature of the Byzantine stateand the society it was attempting to rule

If the Roman landed elite had either retreated to Constantinople or joined thearmy and now maintained their position through service to the state, then thatwould have left Byzantium much closer to being a peasant society. That neednot imply an egalitarian society, but it would be a society where the primaryconcern of the tax-collecting state was no longer in relations with recalcitrantlandowners but with the affairs of much smaller scale farmers. It would be asociety where rural communities were at the heart of agrarian life, and bothpeasants and the state would have a strong vested interest in their protection andmanagement. From the peasant perspective the village community would pro-vide a context in which the small farmer could co-operate with his neighbours,find social and economic support and relate to the demands of the state. Fromthe state’s point of view such communities would provide one of the few obvi-ous means of keeping track of peasant families and forcing them to pay tax andcarry out whatever tasks were required. Landowning aristocrats may defy the statefrom time to time, but they are at least easy to find, relatively few in numberand they have no long-term interest in breaking their bonds. They want thegolden salaries and the titles that the state alone can give. Peasants are potentiallyharder to identify, have less invested in a particular place that will stop thememigrating and are traditionally masters of tax-avoidance. And such a concernwith village communities is exactly what one finds.

The legal texts that survive may not shed any useful light on the fate of thelanded elite, but seen from this perspective, they make a great deal of sense. Ofprime importance is the Farmer’s Law, a short legal text of 85 brief articles thatdeal for the most part with the relations between farmers in a koinotes, a termwhich literally means ‘community’ and is better translated as such rather than‘village’, since there is no necessary implication that this should be imagined as

There remains the possibility that in looking at Byzantine cities we are look-ing in the wrong place, and that to find the landed elites we need to explore thecountryside. These are early days, and the possibility remains open, but so farthe picture is not encouraging. Unlike Gaul, where the pollen evidence suggest

More work at Amorion, more work at other places such as Euchaita whereJohn Haldon has begun an intensive survey (Euchaita-Avkat project), will nodoubt provide further evidence for this sort of activity in the Byzantine period,but so far nothing has emerged to suggest the presence of a prosperous landedelite, continuing to dominate urban society as they had in the past. Even at Amorionthere is nothing to suggest the sort of local elite with money to spare and thewish to display their status that were paying for new churches and mosaics inthe contemporary Levant. Amorion looks like a fortress town, thriving on thepresence of the army, rather than the residence of a landed elite.

Most impressive of all are the current results from Amorion, which lies onthe Anatolian plateau, about 300 km south-east of Constantinople and 150 kmsouth-west of Ankara. An insignificant provincial city up to the fourth century,Amorion prospered in the fifth and sixth, with new churches and a set of majorcity walls built at imperial expense that enclosed some 50 hectares. Because ofthese walls and its strategic situation on the main road linking Constantinopleto the eastern frontier, Amorion became the capital of the Anatolic theme inthe seventh century and an important fortress. Before excavations started theassumption was that Byzantine Amorion was limited to the small mound inone corner of the late antique site, but it has now become clear that this was notthe case. Amorion continued to occupy the whole of the late antique area; thechurches remained in use, as did the baths. Coins and ceramics also show thatthe city remained a bustling monetarized settlement (Lightfoot and Lightfoot2007; Lightfoot 1998).

Other sites show more going on. At Ephesos, for example, a large fortifica-tion wall was erected, probably in the seventh century. It includes only the corearea of the ancient city and the hill behind, but at 3.5 m thick and running over2.5 km it is an imposing structure, and it proves that when needed there wereresources available. The same point could be made for the walls of Pergamon,Ankara, Kütahya or Amastris, and indeed for many other sites across Anatolia.Ephesos also remained an ecclesiastical centre. Its name appears in the lists ofbishops, and the Metropolitan of Ephesos was present at the seventh- andeighth-century councils. The bishop of Sagalassos came too, but at Ephesos theywere able to maintain two of the city’s great late antique churches. That of St

More decisive is the evidence of archaeology. Foss’ work showed that theurban culture still thriving in sixth- and early seventh-century Anatolia came toan end thereafter, and much of what has emerged more recently tends to confirmhis views – which it should be stressed were not that most cities were abandoned,but that they had become poorer, smaller, less complex places; in other words,they had been abandoned by the elites who had made them thrive in the past.Most cities were certainly not abandoned and many remained significantmilitary and ecclesiastical centres, but some were reduced to very little. A goodexample is Sagalassos in south-west Turkey, on the edges of the Taurus close toLake Burdur, and just over 100 km north of the Mediterranean at Antalya,which since 1989 has been the focus for a major on-going excavation. The cityhad been an important pre-Roman and Roman site, with the usual array ofimpressive monumental buildings, as well as a notable pottery industry thatcontradicts any assumption that trade in bulk goods can only be carried by sea.Up to the mid-sixth century the city prospered in a way that makes it a goodexample of late Roman urban continuity. Several churches were built and anumber of wealthy urban villas have been identified. A serious earthquake in theearly sixth century caused a great deal of damage, but the inhabitants were ablerebuild and maintain much of the city’s monumental appearance. Quite clearlyat this date the local landowning elite still had an interest in the site. But afterabout 550 Sagalassos fell into rapid decay. The site was struck by another earth-quake in the mid seventh century, and although there are signs of some occupa-tion after that, very careful work has produced nothing to suggest that the formervilla-dwelling elite were still in residence. At most, eighth-century Sagalassosconsisted of a church, a few simple houses among the ruins of the former city,and possibly a fort occupying the remains of the ancient sanctuary dedicatedto the emperors Hadrian and Antoninus Pius (Waelkens 2003a, 2007). To goback to Theodore Laskaris’ mouseholes simile, early medieval Sagalassos seemsto have been no more than a settlement of mice.

The case against such a continuous history in Byzantium and for the propositionthat the seventh and eighth century marked a significant break in the history ofthe empire’s elites can be made on a variety of grounds, historical and archaeo-logical. The historical arguments turn principally on changes in titulature, changesin naming patterns and an argument from silence. To deal with titles first, thelate Roman aristocracy had been defined by its senatorial rank, the outward markof which was a senatorial title, ranging from clarissimus at the bottom to illustrisat the top. Although the senate survived in some form until the very end of theempire in the fifteenth century, by the eighth century at the latest the traditionalpattern of senatorial titles had been superseded or devalued, and replaced by anew titulature based on court and military office. The change in naming patternsis that from the traditional nomenclature used by the Roman aristocracy whichinvolved the use of three names, a first name (praenomen), a family name (gentilicium)and surname ( cognomen), to the use of a single name, and a fairly restricted poolof names at that. The argument from silence is simply that unlike texts fromlater periods which often draw attention to someone being well born or of noblelineage, there is very little in the surviving sources for the seventh or eighthcentury to suggest that the aristocracy at this date was conscious of its ancestry(Haldon 2004, 187–92, 221–31; Settipani 2006, 20, 35–50; Cheynet 1996, 269–73).

Striking in its own right, but also serving to make the case for continuity inGaul more compelling, is the evidence from Spain. There over a wide area ofthe northern Meseta have been found a remarkable body of slate tablets,scratched with among other things the estate records of Spanish landowners whofarmed the region between the sixth and eighth centuries (Velázquez Soriano2000, 2004; Santonja and Velázquez Soriano 2005). These texts currently fall inan archaeological vacuum that parallels that of Gaul. The villas that had beenestate centres up to the fifth century seem to have been abandoned, and there isno evidence that landowners were now living in towns; and yet the tabletswithout any doubt show business as usual. If the late Roman elite continued inSpain, why not Gaul too? And if Egypt and the Levant, and Spain, and Gaul,why not where the empire actually survived, in Byzantium too?

Perhaps more surprisingly, a not wholly dissimilar picture has also begun toemerge from some parts of the Latin west, and in Framing the Early Middle AgesWickham makes it a key part of his argument. Unlike Egypt or the Levant, thereis no doubt that in the west by the sixth century the prosperity and complexitywhich the ancient economy had been able to deliver had come to an end, buteven so there is good evidence at least in some areas for the continuing survivalof the Roman elites, not just as ecclesiastics on the familiar model of Gregory ofTours, but as landowners continuing to run their estates very much as theirancestors had done before. Key to this conclusion is the world of great estatesand peasant farmers that thanks to the survival of monastic documents becomesvisible in northern Gaul from the seventh century. Where did these estates comefrom? It is possible of course that they were a new phenomenon created byFrankish incomers, and it is true that there is currently a gap between the lastevidence for late antique villas in the fifth century and the first documentaryevidence for these estates in the seventh. But surveys and pollen analysis showthat the fields of northern Gaul were never abandoned, and it is hard to see howthe complex estate structure visible in the documents could have come into placeso quickly if Roman structures had not survived (Wickham 2005, 178–84). ForWickham, if you want to see what a world looks like where the Roman orderreally did come to an end, then you need to look at post-Roman Britain. Therethe economy retreated to a basic simplicity not known since the early Iron Age,and Britain was not to achieve levels matching its Gallic neighbour until theeighth century at the earliest. Unlike Gaul, there is no evidence for the continuityof the late Roman elite, and in that difference Wickham would want to see thekey to their different economic histories (Wickham 2005, 306–15).

The same can now be said for the countryside too. Until the 1990s this pointwas rather obscured by the influence of Georges Tchalenko’s work. His seminalstudy published in the 1950s on the villages of the limestone massif in northernSyria between Antioch and Aleppo had revealed the late antique period as oneof burgeoning rural prosperity, and he argued, effectively only on the basis ofbuilding inscriptions, that this came to an abrupt end in the seventh century(Tchalenko 1953–8, i, 422–38). Tchalenko’s material was re-examined by GeorgesTate in the 1980s, but only to bring the end forward to the mid sixth century,and put the blame on a Malthusian crisis rather than the Arabs. The basic ideaof a world coming to an end was unquestioned (Tate 1992, 335–42). However, bythe early 1990s the results of new fieldwork carried out in Jordan and Palestinewere coming into print, and these projects were based on extensive survey tech-niques and the analysis of pottery scatters rather than simply the planning ofsurviving structures. What they showed, and the results have been confirmedsince then by a host of other projects throughout the Levant, was that Tchalenkoand Tate were wrong. Rural prosperity continued unabated, in some cases flour-ishing even more after the Islamic conquest than before (Walmsley 2007, 31–47,107–12). Landowners who had been doing well in the sixth century continuedto do well in the seventh and the eighth. The late eighth-century mosaics dis-covered in the church of St Stephen at Umm al-Rasas in Jordan in 1986 is onlythe most spectacular example of a phenomenon attested throughout the region(Piccirillo 1992, 232–43; Bowersock 2006, 65–88). The late antique Christian

Cities such as Pella, Gerasa, Scythopolis and Caesarea, once cited as evidencefor the catastrophic impact of the Islamic conquest, now seem to show thereverse. This is not to imply that any of these sites was frozen in some sort ofclassical time-warp; rather the point is that these sites had been thriving urbancommunities in the sixth century, and in the seventh they seem to have carriedon evolving on much the same lines as they had been doing for a century ormore. On none of these sites did the Islamic conquest mark a dramatic break(Wickham 2005, 613–26).

Much the same point could be made for Palestine and Syria. Until the 1990sit was commonly assumed that the Islamic conquest brought with it a rapiddissolution of the existing order. Some opinion had it that that order was alreadyundermined by the mid-sixth century, a victim of plague, war and other catas-trophes, but there was little disagreement that the seventh century marked a newworld. However, the arguments cited in support of these views tended to becircular. The period saw the end of the ancient world; therefore any otherwiseundatable objects that belonged to a period of prosperity must be dated to before600. In turn the lack of evidence for seventh-century prosperity was proof thatthe old order had come to an end, and so it went on. Over the last fifteen years,thanks to new knowledge and new attitudes to the period, much of this has beenproved wrong, and in town and country what now seems much more strikingis the evidence for continuity (Walmsley 2007).

In 1990 I stopped my discussion in the seventh century. But the question hereis: could one actually take this argument further? Did the late Roman elite havea similarly continuous history in Byzantium through the seventh and eighthcentury? In some parts of the former Roman empire this is quite clearly whatdid happen. In Egypt the late antique social and economic order seems to havecontinued regardless. Some individual families, such as the Apions, famousthrough their estate archives discovered at Oxyrhynchus, failed to survive theIslamic conquest, but overall the impression is that much the same body oflandowners continued to oversee much the same body of farmers and labourersdoing much the same sorts of things as they had in the past. Landowners con-tinued to be responsible for local government and the payment of taxes; the chiefdifference was that the tax now went to an Arab governor acting in the name ofthe Caliph rather than a Greek-speaking governor acting in the name of theemperor. The fact that Islamic rule brought higher taxes and the first recordedpeasants’ revolts in Egyptian history may hint at new pressures in the system,

The last twenty years’ research has if anything tended to reinforce the caseI made then for the strength of the late Roman elite. Peter Heather in his com-mentary on the orations of the fourth-century rhetor, Themistios, has made thefundamental point that what made his extraordinarily successful career possiblewas the need shared by successive emperors to sell imperial policy to the thou-sands of landowning aristocrats without whom government was effectivelyimpossible (Heather and Moncur 2001, ch. 2). If such families lost faith in theregime, or worse still, as happened in the fifth-century west, lost faith in theempire, then the emperor would rapidly be shown to have no clothes. Themis-tios carried conviction because he was one of them: a provincial landowner,educated as they were, and a pagan. At that date the latter was an importantqualification for the spokesman of even such Christian emperors as Constantiusand Theodosius. In practice this was not a bureaucratic state (the numbers werealways very small for the millions of peoples and hectares they tried to rule), butlike most premodern states one in which local elites were induced by a combi-nation of reward, threat and shared self-interest to serve the regime at the centre.Jairus Banaji has taken this further by showing how the empire’s adoption of agold-based coinage reinforced the economic and political power of an aristocraticclass now paid in gold (Banaji 2001), and Peter Sarris has shown how the ten-sions between the emperor and this gold-salaried and landed elite lay at the heartof sixth-century politics. Procopius’ portrayal of Justinian as a dangerous innovator,subversive of the established aristocratic order, reflects exactly this aristocratichostility to an emperor attempting to reinforce the power of central government.Sarris goes to on propose a reading of sixth-century politics where the emperor’sability to harness the resources necessary to wage war against the empire’s enemiesis being harmed by aristocratic reluctance to pay. Justinian made serious efforts toimpose the imperial will on the aristocracy, but in the end he and his successorsfailed. The empire did not face crisis at the beginning of the seventh century becauseof a decline set off by plague or climate change or the impact of war, but becausethe state was increasingly at the mercy of its aristocrats (Sarris 2006, 228–34).

In 1990 I published a paper arguing that, despite very important changes toRoman cities between the fourth and the beginning of the seventh century, inSyria, Palestine, Anatolia and Greece the elites that ruled them had a continuoushistory, adapting to new circumstances and surviving with their power intact(Whittow 1990). The familiar story of the decline of the curiales, members of thelocal city council, recruited from local landowners, was therefore something ofan illusion. These councils had certainly lost power, and membership hadbecome a chore to be avoided long before they disappear in the sixth century,but that did not mean that cities were any the less run by local landowningnotables. Their mode of operation had changed. They now tended to hold eccle-siastical office or have senatorial rank, both of which exempted holders fromcurial service, but otherwise the same sorts of people, estate owners as before,continued to exercise local power as they had done in the past (Whittow 1990).

The case for the paramount significance of the seventh century has to rest ona belief in fundamental change taking place in the empire’s socio-economic struc-tures, and this has been an area where Marxist historians have taken the lead.The disagreements aired between Chris Wickham and John Haldon in thepages of Past and Present and Journal of Peasant Studies during the 1980s were notabout continuity, both favour fairly radical change, but about how it should bedescribed. Did the seventh and eighth century see a shift to a different mode ofproduction? For Wickham, there is a fundamental distinction between rent andtax as alternative means of exploiting the peasantry. The former is characteristicof the feudal mode of production, the latter characterizes what he calls the tributarymode of production. The Roman empire was a tributary state; Byzantium wastoo, at least until the twelfth or thirteenth century. For Haldon, and for HalilBerktay, who also contributed to this debate, tax and rent are in this contextidentical. Whether the peasantry pay tax or rent matters little. The mode ofproduction and the lived social experience is the same (Wickham 1984, 1985;Berktay 1987; Haldon 1989).

For most of the twentieth century it was conventional to call the empireByzantine from the moment when Constantine converted to Christianity andfounded Constantinople in the fourth century. To a generation that came to itshistory from a study of the classics, and for whom Rome meant Caesar, Augustusand pagan temples, not Constantine, Theodosius and Christianity, this was anatural point to draw a distinction. For archaeologists working on the pre-Islamic Levant and Egypt it still is. But for historians of the empire that survivedthe early seventh century it has become conventional to use the term Byzantineonly from that point on, and to use it with the implication that the seventhcentury marked a moment of decisive transformation in the empire’s history.Byzantium was not only a rump of the Roman empire that had preceded it, itwas a fundamentally different society, economy and culture.

They were dubbed “ ideal type” since besidesoffering schematic presentations of complex historical phenom­ena such as modern capitalism, they also encoded the individualand relative values (or Ideas) of the investigator who set themup. The innovation was daring enough as it stood, but Weber

Moreover, in T h e P r o t e s t a n t E t h i cWeber deployed an array of new methodological devices deriv­ing from other disciplines, in an attempt to cope with the essen­tially infinite nature of historical data. The best known, thoughby no means the most original, was the reduction or grouping ofhistorical phenomena into types: these might be either timelessand universal (and “ sociological” ), or else unique and genetic(and “ historical”

On the other hand, he believed, as much as any orthodoxhistorian, that present-day phenomena should be understoodin terms of their previous history. In the case of East Elbeanrural society this took him back 50 or 60 years, but in thecase of fundamental structures, such as law, the state, and reli­gion, he returned to classical antiquity and pre-ChristianJudaism. As a result, there is immense historical content inboth his empirical and his “ sociological” (or typological) writ­ings, which are all readily accessible to the historically mindedreader. In his methodological writings, too, Weber argued withexemplary clarity that, while the investigator should be lookingout for typical and recurrent elements in social phenomena,the unique element underlying these was irreducible. Inthinking thus, he was little different from many historians inthe post-Enlightenment mainstream who used the foundationsupplied by historically unique evidence as the basis for widerand even timeless generalizations (which might be called “ phil­osophical” or “ sociological” ), except that the balance betweenthese concerns was tipped rather more toward the latter. Touse Weber’s own words, he saw himself as a historian in the“ broad” if not in the “ narrow” sense.

A fundamental problem confronts us at the outset: in whatsense is it proper to speak of M ax Weber as a historian? Hisundergraduate and graduate studies were principally in law,while his university appointments (18 9 4 -19 0 2, 19 18 -2 0 ) werein the faculty of “ National Economy.” Furthermore, apart fromtwo dissertations that were the product of his early legalstudies, there is only one work of his maturity that might inany sense be called a work of historical research - “ Die prote-stantische Ethik und der Geist des Kapitalismus” (1904-05,revised 19 20 ; T h e P r o t e s t a n t E t h i c a n d t h e S p i r i t o f C a p it a lis m ,1930) - and this, when published in the A r c h i v f i i r S o z ia l-w is s e n s c h a f t (a journal of which Weber was co-editor), wasdeemed to merit excuse as something exceptional to its (andhis) ordinary range. Nonetheless, there can be no doubt of thesignificance of Weber’s historical concerns. His primary focuswas indeed on the present, a premise that underlay a gooddeal of empirical research into the social life of his own day,and which readily explains his absence from any history faculty.

Evidently not all counts were in fact literate in Charlemagne's time. Themissi quoted above hoped the counts would read the written orders theywere sent, but foreseeing that they might not be able to, ordered that anycount in doubt over his orders, whether written or oral instructions, shouldsend a deputy 'who understands well' for refresher instructions. 'Read yourcapitularies and you will recall what you were ordered to do through thespoken word. '117 Each count was also told to 'make a list of any persons whoare rebellious or disobedient towards you', and either to send the listimmediately to the missus, or read it out to him when he came round oncircuit. 118 Constant interchange was envisaged by these missi: counts musthave needed their couriers. Some of Charlemagne's successors seem to haveexpected more literacy of their counts. Both Louis the Pious and Charlesthe Bald bombarded their counts with capitularies. Louis asked them, likethe missi, to produce lists of those liable for military service; 119 Charlesissued his counts with the same instructions, and also told them to produce

Missi were ordered to use documents for a variety of purposes, includingthe maintenance of communication both ways between centre and locali-ties. So that all missi might publicize them at local assemblies, those withcapitularies had to send copies on to those without them, ut nulla excusatiode ignorantia fiat ('that no excuse be made on grounds of ignorance'). 109Missi also had to bring written reports back to Charlemagne concerning theimplementation of the written instructions they had taken from his oralpronouncements. 110 They were told to bring him lists of names: of all thescabini, advocates and notaries whom they had appointed; III of all strangerswho had immigrated into their missatica, of their county of origin and ofwho their lords were; and of all 'bishops, abbots or others of our homineswho failed to attend the assembly' summoned by the missus. 112 Reportswere to be sent in about all who were unable to pay the army-fine for non-appearance at the host. 113 If the missi together with the local counts couldnot succeed in righting wrongs or bringing malefactors to justice, they wereto refer cases, with dossiers (cum brebitan·is suis) to Charlemagne him-self. 114 What was done with these documents when - if - they reached thecentre? There is no evidence of their preservation in a central archive.Having been 'brought' to court and served as aids to memory in oralexchanges between ruler and agent, the lists and records were perhaps then

The production of surveys by landlords can be linked in these instanceswith royal demands. But surveying was of direct benefit to landlordsthemselves. It has recently been argued that many extant ninth-centuryestate-surveys were products of 'private' rather than 'public' initiative. 85The difficulty of applying such categories becomes obvious when weconsider that the allegedly 'private' survey of the lands of the church ofRheims in 845 was made precisely because of the intersection of royal withepiscopal interests in these estates. 86 A crucial consequence of theincorporation of the church into the Carolingian state was that kings coulddemand the military services of men who held church lands as benefices.The importance of benefice-holders to Charlemagne's military capacity issuggested in a capitulary, probably of 807, which instructs missi and, intheir localities, vican'i, the subordinates of counts, to make surveys of 'allbenefices held by our men and by others' men', and to bring these lists to theking. The maintenance of benefices was then to be compared with that ofallods, in order to prevent benefices from being 'destroyed' or assimilated toallods, since only benefices kept 'correctly' were in a state to enable theirholders to perform military service at the king's command.s7 WasCharlemagne demanding access to the service of all benefice-holders, in

Charlemagne expected his missi to know how to draw up surveys: he sentthem models to imitate: et sic cetera breviare debes ('and this is how youmust write up your survey of the rest'). 67 How far were such instructionscarried out in the case of royal estates? Though Ganshof himself wassceptical in 1951, now, thanks to his own and others' subsequent work, thepatchy evidence for the making of inventories of Carolingian royal estates, ifnot in Charlemagne's reign, then later in the ninth century, has becomecumulatively impressive. 68 The production of such surveys makes historicalsense, too, in light of successive partitions of the royal lands betweenCarolingian heirs, and, in the early 840s, fraternal conflicts that necessarilyinvolved hand-outs of royal lands to win noble support. That was whyCharles the Bald's bishops advised him in 845 to send missi to 'make surveysrecording everything that had been in the royal service or held by royalvassals under your father and grandfather, and what now remains', so thatthe extent of recent losses could be gauged. 69 The contemporary authorNithard, himself a lay magnate, explains that in October 842, the three-waydivision of the impen'um between the sons of Louis the Pious had to bepostponed for lack of a complete (or up-to-date?) noticia, and that whileLothar tried to push through a division, nonetheless, his brothers' envoysinsisted that a thorough survey must be made first. 70 Contemporaryannalists confirm that during the following nine months, the realm was

The foundation of power, whether that of ruler, lay aristocrat or greatchurch, was efficient landlordship. Since power had to be devolved to thelocalities where the estates were, much turned on control of local agents: anextremely difficult business, even for kings. The first qualification of a 'good

The specific location of the written word in the Carolingian world is thekey to its function in the construction as well as in the deployment of power.It was not just the means whereby a ruler issued commands or standardizedthe conduct of legal business. It defined membership of the realm: the free,the faithful men in the widest sense, were those capable of receiving andusing written documents in public courts. It also defined a ruling cadre:those lay as well as ecclesiastical persons with direct command of Latin. Theroutinization of government presupposed the legitimation of its agents.Through active participation in literacy, that is, a capacity to write and readas well as simply use documents, the elite of the Carolingian world declareditself. Women shared its distinguishing trait: thus Eberhard left a law bookto his daughter, while his wife, in a charter she herself issued, declared thevalue of written documents in 'making [a property division] binding,according to law'. 59 Though gender prevented them from exercising publicoffice,60 it did not deny them privilege: women as noble holders, ortransmitters, of property rights, claimed through literacy membership ofthe governing class.

4 In the generation after Charlemagne, more members of the lay eliteabsorbed the values of their ecclesiastical colleagues. The noblewomanDhuoda advised her son that 'God is learned about through books.'55Bishop Jonas of Orleans pointed a moral for Count Matfrid: 'if men makesuch efforts, with the conduct of legal disputes in view, to learn so greedilyand to understand so precisely the prescriptions of secular laws which havebeen issued by mortals, how much the more should they strive to learn

The second kind of distinction to be made is between Romance-speakersand others. The bulk of our evidence for active lay literacy comes fromregions of Romance-speakers. But the survival of the two vast charter-dossiers of St Gall and Redon, both sited in non-Romance-speakingregions, shows that Romance-speaking was not a necessary condition forwidespread prag~atic lay literacy. 47 Other variables were crucial: whetheror not the region had once been part of, or strongly influenced by, theRoman Empire, and whether or not the Christian church had beenimplanted in the region long enough before the Carolingian period for thehabit of Latin documentation to have become established. Where thoseconditions were absent, as for instance in Saxony, little evidence survivesfor either passive or active lay literacy; but where those conditions werefulfilled, whatever the local spoken language, the charters and court recordsindicate that most of the free population, and some of the unfree too, werepassively, pragmatically, literate: that is, landholders and tenants habitu-ally used documents, and were even connoisseurs (as in the Tours case) oftheir formal traits. This must mean (and the researches of McKitterick andDavies have recently demonstrated for Alemannia, Rhaetia and Brittany)that notaries were very widely available in many parts of the Carolingianworld. Though in regions north of the Alps these notaries seem, for themost part, to have been clergy or monks, they could produce records fortransactions in which no church interest was involved. This must havebeen, like moneylending, a significant (and, in a small way, lucrative)service performed by clerics for local lay communities. 48 Only in Italy doesit seem certain that a lay notariat persisted.

Only a few members of the high aristocracy certainly wrote, as well asread and used, written texts. Some offspring of landholders of 'mediocre'status may have acquired basic capacity to read and write, and the survey ofthe tenants of St Victor, Marseilles, lists some half-dozen boys 'at school',presumably in training for holy orders.44 But not all schoolchildren weredestined for an ecclesiastical career: one set of episcopal statutes evensuggests that girls as well as boys received some instruction from parishpriests, possibly to prepare them for future godparental duties, for thesewere taken very seriously by the Carolingian church. 45 The mass of free (orfreedmen) landholders, however, whether owners or tenants, were passiveparticipants in literacy: they were regular users of documents of specificlegal types in specific legal contexts, and though they could certainly notthemselves have written the documents, and probably could not even readthem in full, they could recognize standardfonnulae, and display remark-able expertise on matters of formal correctness. In 857, for instance, at acourt held near Tours, in a case that turned on the validity of a title of sale,'nearly all the coloni [free peasants] gave witness that they had not seen [thedocument] being corroborated, nor were the names written down there

Measuring the extent of lay literacy in the Carolingian world in anystrictly quantifiable sense is, of course, impossible. But that need not deterefforts at more impressionistic assessments. Two kinds of distinction arehelpful here. First, we need to differentiate between, on the one hand,members of the ruling elite, that is, the high aristocracy, occupying highsecular or ecclesiastical office, and, on the other hand, the broader class ofmediocres and even some pauperes, free persons of lesser status anddependent on the political power of potentes. The former group may betermed the actively literate, those fully in command of Latin, able to wieldpublic power by conveying their will through their own written words; thelatter group were those familiar with procedures involving documents, ableto find notaries to draw up documents when needed, and experienced inwitnessing the transactions of their fellows - in short, such 'knowledgeableneighbours' as were referred to in a capitulary of 816: persons whose literacymay be termed passive or pragmatic. 4

For Carolingian churchmen, the written mode always coexisted with theoral mode. The Word was both medium and message. Whatever emphasisthe church put on writing, it always stressed, at the same time, that Truthwas revealed through speech. Preaching meant the oral explication of HolyWrit. Hincmar (c. 805-82) preferred to argue and prescribe et verbo etscnpto. For Regino-of Priim (d. 910), things worth recording were to belearned in chronicornm libn"s or ex relatione patrnm. 25 If access to literacymeant different things at different levels of the church's own hierarchy, sowith the laity, while the church's mission was universal, and all Christiansknew what holy books looked like from the outside, direct knowledge ofwhat was inside was hightly restricted. It was no part of the Carolingianprogramme to provide, as it were, access courses for pauperes, even if somepauperes might well benefit from having their names inscribed on achurch's list of registered poor, or (along with nobles) in a MemorialBook. 26 Churches were decorated with tituli, dedicatory inscriptions, sothat the faithful, physically separated from the liturgical books beyond thechancel rail, saw writing incised or painted on the walls above and aboutthem. These texts were explicated in sermons and pictures - so that, as Bedeput it, pictures could 'offer those unable to read, a sort of live reading (vivalectio) of the Lord's story'. 27 But 'without an accompanying text (inscrip-tio), how could you know if a picture of a woman with a child on her kneeswas the Virgin and Child or Venus with Aeneas?'.28 A minimal amount oftext was required as authentication. Only for the elite few was writing amedium of private spirituality aimed at self-control. For the rest, thewritten ~ord was a public medium of control imposed externally. It was in

How far were these outcomes intended? The reformers were not onlyinterested in accessibility. Operating in an extremely inegalitarian world,they assumed and exploited differentials in access to power. Latinity'spotential for restrictiveness was thus a recommendation. Within the churchhierarchy itself, full control of Latin was not for the lower clergy: it wasenough that they be able to read, or use as aids to memory, a small numberof basic texts. A full command of Latin, active as well as passive, wasexpected of the higher clergy, themselvespotentes , members of the magnateclass. It took special pleading and a lot of utilitas, that is, political andperhaps military capacity, to make up for a certain lack of ernditio in anepiscopal candidate. 22 Yet the phenomenal level of latinity acquired bysome aspirants to and holders of high ecclesiastical office was itself asignificant source and symbol of their authority, to be paraded in synodaldecrees and in letters and poems that also constituted a public discourse. 23The conditions under which these churchmen wielded their power imposeda particular kind of widened access to latinity: namely, that lay magnates tooshould be enlisted among the ranks of the actively literate - writers, andconnoisseurs of style, as well as readers. Handbooks of spirituality andprivate prayers produced by ecclesiasticalpotentes for lay potentes signalledan entente: the two wings of the Carolingian elite were to be united in

A crucial feature of the cultural context of Carolingian literacy was thechurch's commitment to the practice of the written word. Latin was theunique medium of orthodoxy in the early mediaeval west. Carolingianreformers did not choose to make it so: they received it as such fromChristian late antiquity. But though the medium had remained the same, itsfunction had changed. 21 In the fourth and fifth centuries, despite some

What the Carolingians did was to take up again the threads that linkedpatria and palace. Ninth-century land-surveys are inconceivable without acontinuous sub-Roman tradition of record-using landlordship. Carolingiannotitiae of judgements in county courts are inconceivable without asimilarly continuous tradition of the use of documents in legal proceedingsall over the so-called 'barbarian' west. Continuities with the Merovingianperiod are offset, on the other hand, by a huge increase in the coverage andvolume of the Carolingian evidence, especially from the reign ofCharlemagne onwards. The Carolingians operated on a different scale fromtheir predecessors. Of the capitularies printed in the standard MGHedition, only some 2S pages relate to the Merovingian period, over 700pages to the Carolingian. 13 The Leges, and the fonnulae, survive virtually

In the last few decades of the Merovingian period, up to c. 750, suchintrusions of central power became rare: not only had the links dwindledbetween the two levels of government, but the higher level had all butceased to function. This is why Ganshof saw a decline of the written word inadministration immediately before the Carolingians' advent. 12 There wereno late Merovingian capitularies, nor after c. 700 any more Merovingianroyal judgements, and very few royal charters. There were, however,plenty of non-royal charters and much activity in non-royal courts. In short,there was government, but it was government in non-royal places, at thelevel of the pain'a.

Butsome generalized economic demands were made by Carolingian rulers: ifgreater use of the written word accounts for the vast increase in the evidencefor economic activity in theCarolingian as compared with the Merovingianperiod, increased economic activity may be partly responsible for theincrease in the extant records. Rulers who want to control and exploitcoinage and markets, and to protect merchants while taking a cut of theirprofits, have obvious uses for written documents. The Carolingians,especially in the ninth century, resumed these aspects of Roman imperialpractice for a mixture of motives, including fiscal ones. to The issuing ofletters of protection for merchants, the maintenance of toll-stations, thelicensing of mints and markets, imposed considerable governmental

Medicine was another practical art studied in the schools. Doubtless,throughout the Carolingian world medical practitioners, clergy as well aslay men and women conversant with folk medicine tended the sick. In theninth century a learned tradition of medicine developed which wassupported by medicine's inclusion among the liberal arts and also byCarolingian legislation.166 Evidence for the study of medicine in the schoolsshows up, for example, on the fly-leaves of the unique ninth-centurymanuscript of Einhard's letters (BN lat. 113 79) or in the equally rare copy ofLupus of Ferrieres' correspondence (BN lat. 2858). Hundreds of medicalrecipes, dietary recommendations and medical explanations are scatteredthroughout other manuscripts along with numerous citations and lengthyextracts from Hippocrates, Galen, Oribasius, Dioscorides, Soranus, Alex-ander of Tralles, Theodorus Priscianus, Pliny, Quintus Serenus, CassiusFelix and Marcellus Empiricus.167 In the ninth century it was not unusual forecclesiastics and scholars of the mettle of Walahfrid Strabo, Lupus ofFerrieres, Abbot Dido of St Pierre-le-Vif or Bishop Pardulus of Laon to beknowledgeable about medical matters and, in the case of Hincmar ofRheims, to integrate concepts of disease and health into systems ofthought.1

Clear pedagogical concerns are also evident in what may have been themost significant development in computistical instruction in the ninthcentury. Some time early in the century, a Carolingian master gatheredtogether a series of excerpts bearing on astronomy from the second book ofPliny's Naturalis Historia. The selections discuss the positions and courses ofthe seven planets, the intervals between their circular orbits, their apsidesand, finally, their travels through the bands of the zodiac. The excerptsappear in thirty-eight manuscripts dating from the ninth to the twelfthcenturies and attest to the widespread interest in these sections of theNaturalis Historia. The Plinian texts also were available to Carolingianteachers in both the 'Three-Book Computus' and the 'Seven-Book Compu-tus'. The Plinian excerpts attracted the attention of teachers because they

The abundance and variety of arithmetical and scientific texts in Carol-ingian libraries and classrooms sometimes caused confusion and led todebate. In 809 a meeting of computists [compotistae) was held at Aachen totry to clarify the most vexing problems. The assembled experts were asked

The movements and characteristics of the heavenly bodies fascinatedCarolingian observers. Alcuin had to respond to Charlemagne's questionson the subject. Dungal too was called on by the emperor to explain theeclipse of the sun in 810. Lupus of Ferrieres and 'the Astronomer', theanonymous author of a Vita Hludoivici Imperatoris, were both intelligentobservers of astronomical phenomena.135 In the schools, instruction in thetechnically rigorous computus, specifically called for in the AdmonitioGenera/is, included arithmetic as well as knowledge of the courses of celestialbodies. The fundamental texts masters used included Victorius of Aqui-taine's Liber Calculi, Boethius' De Arithmetica, the works of Euclid, Plinyand Bede as well as numerous computistical manuals and tables andingenious mnemonic verses which helped students to learn and remembercomplicated relationships.136 One teaching text suggests that Carolingianstudents were the first in the west to learn the rules for adding positive andnegative numbers.1

Dialectical studies complemented the study of grammar and rhetoricsince all three aimed to understand words: written words, spoken words,words used in texts, in prayer, in argument and in speculation. The otherarts which formed the school curriculum concerned number. Carolingianstudents studied the mathematical arts for obvious practical reasons. TheAdmonitio Generalis required that young students learn to master thecomputus, the early medieval method of reckoning time and determiningdates. They would also have to know how to manipulate numbers when itcame to collecting the tithe and dividing it into fourths, when calculating theharvests from fields or the rents due from farmers. The vast buildingprogramme promoted in the Carolingian realms stimulated the study ofratio, architecture and geometry as evinced by the careful proportions ofsurviving buildings and by the meticulous draftsmanship of the Plan of StGall. The increasingly theoretical nature of ninth-century chant likewiserequired keen understanding of harmony and measure.1

This dialectical approach to theology was also broadcast by Candidus,Alcuin's pupil, from whose teaching a similar set of dicta survives. Extractsfrom the dicta appear in the Libri Carolini and in Benedict of Aniane'sMunimenta Fidei. It is in this context of growing interest in the technique ofdialectical inquiry that Fridugisus' De Nihilo et Tenebris must be seen.129Fridugisus was concerned to discuss the reality of negative qualities, aquestion that was assuming greater urgency as thinkers attributed negativepredicates to God. Fridugisus also broached the question of the pre-existence of the soul before union with the body and was rebutted byAgobard of Lyons who marshalled patristic and biblical citations againsthim.130 Agobard was not alone in questioning the results of the newmethodology. Theodulf of Orleans, Benedict of Aniane and Prudentius ofTroyes also reacted against what they thought was excessive reliance onsyllogistic reasoning — a defect they tended to ascribe to Irish masters. Butdialectic proved so useful in exploring perennial problems of Christiantheology that ninth-century debates on the Eucharist, the world soul andpredestination all depended on it. Indeed, in combating John Scottus

The recovery in the curriculum of works such as Aristotle's Categories and

The revival of rhetorical studies in the schools and of rhetoric in scholarlyand intellectual discourse in the ninth century was accompanied by intensestudy of dialectic and the application of dialectic to the problems intellec-tuals debated in the Carolingian realms. It is no longer possible to imagine along period of dormancy in dialectical studies between the time ofAugustine and Boethius on the one hand, and the eleventh century on theother.125 Nor is it possible to understand the fundamental nature ofdialectical studies in the ninth century only by saluting John Scottus' use ofdialectical reasoning in his works or by belittling Fridugisus of Tours'treatises on the reality of 'shadow' and 'nothing'. In the ninth century abroad range of masters and their students considered seriously the problemsof essence and universals and applied Aristotle's categories and syllogisticreasoning to theological speculation.

Study of the figures of speech and especially of the writings of late antiqueChristian authors enabled masters who wrote and scholars — an Alcuin, aTheodulf of Orleans, a Lupus of Ferrieres - to embellish their prose not onlywith rhetorical figures, but also with rhyme and parallelism. For moststudents, however, rhetorical studies amounted to the mastery of appropri-ate prose styles. The letters of Lupus of Ferrieres offered models that theymight emulate in their own correspondence. The master who used Lupus'letters in his teaching peppered the margins of the unique copy of the letterswith observations such as 'an excellent plea' {optima supplicatio), 'excellentencouragements' {optimae adhortationes), 'Note a humble and prudentrequest' (Nota humilem etprudentem implorationem).nx Collections of modelletters also prepared students to write all kinds of letters.122 Carolingianstudents learned how to write letters of condolence, how to describe a king,how to compose a debate between winter and spring, and how to writeletters and poems of praise and flattery. The results of this training in the useof stock phrases and exempla crop up everywhere in Carolingian literature

Grammar and the explication of secular and religious literature werefundamental to Carolingian schools and literary culture. Grammar was butone constituent of wisdom. Mastery of the first art opened the way tovigorous study of the other arts. When Carolingian masters taught aboutlearning, wisdom and the relation of the arts to each other, they abandonedthe schema of the traditional seven liberal arts canonised in MartianusCapella and in Isidore of Seville's Etymologiae, preferring instead that ofIsidore's Differentiae in which wisdom is subdivided among physics, ethicsand logic. Physics includes the quadrivial arts, arithmetic, geometry, musicand astronomy as well as astrology, the mechanical arts and medicine. Ethicsembraces the four principal virtues of prudence, justice, fortitude andtemperance. Logic, the study of words, consists of rhetoric and dialectic towhich grammar 'adheres'.117 The relationship among grammar, rhetoricand dialectic was indeed one of adhesion since they all focused on the propercomprehension and use of words. As Carolingian students studied grammarand secular and divine literature, they also began to learn how to manipulatelanguage through rhetorical and dialectical studies.

Any Carolingian student would have recognised Christian's themes fromstudy of the Carolingian commentaries on secular authors. In addition to thecommentaries on the grammarians, the ninth century produced glosses andcommentaries (partial and whole) on Virgil, Boethius, Sedulius and Martia-nus Capella among others. The commentaries usually begin with an accessus,or introduction, organised around the seven periochae: an account of theauthor's life; an explanation of the work's title; a description of the nature ofthe text; the author's intention in writing the work; the number of bookscontained in the work; the order of the books; and a summary of thework.115 The commentary then proceeded to explicate the text. Commentsranged from the lexical to the mythological and historical to the philosophi-cal and theological. The commentaries tended to have lives of their own andseem to have continually developed from year to year as the master taught.A written version of a commentary might enshrine a set of notes in atradition that was free to grow prompted by the demands of the classroom,the use of new texts or the incorporation of comments by other masters. Thecomplicated textual history of the ninth-century Martianus Capella com-mentaries and of John Scottus' Periphyseon testify to creativity and continualrevision on the part of Carolingian masters.1

Carolingian commentators clearly had students — bothyoung and adult - in mind. They avoided the allegorical approach for thehistorical and literal and aimed at brevity and clarity in their expositions.Originality was not one of their goals. Over and over they stressed in theprefaces to their works how dependent they were on the patristic giants whopreceded them. But when they excerpted, combined and rephrased theirexegetical predecessors, Carolingian masters did in fact produce new worksthat helped to make the ninth century the first great age of patristicscholarship.

Study of Latin authors, especially the patristic authorities, sometimes ledto the study of Greek which, during the third quarter of the ninth century,constituted one of the special achievements of the Carolingian renaissance.Continental knowledge of Greek was once thought to be an import fromIreland and limited to Irish masters such as John Scottus. But thewidespread evidence of wordlists and Greek terms in poetry and exegesissuggests that some Carolingian students were taught this second of the threesacred languages, enough at least to understand a Greek term encounteredin Jerome.1

Carolingian students inhabited several linguistic worlds. They never lefttheir first languages behind, be they rustic Latin or German, when theyembarked on the formal studies that aimed to prepare them to speak andwrite like the church Fathers. Hraban Maur quoting Augustine in the DeInstitutione Clericorum thought it prudent to remind his German-speakingstudents that when Augustine wrote 'our language' he meant Latin.98Learning Latin did not mean that students turned their backs on their firstlanguages. They continued to need to communicate in those languages andactively studied the vernaculars. Lupus of Ferrieres sent some of his monksto Prum precisely to learn German which he thought would be a usefulacquisition.99 As future priests learned Latin biblical, theological andecclesiastical concepts they also had to learn how to express them in thevernacular to the people. Bilingual Latin—vernacular glossaries helped themto explain their new learning in their first languages and doubtless proveduseful to missionaries and teachers as a springboard into the world ofChristianity and Latin learning.1

The study of Latin vocabulary went hand in hand with the study of themechanics of the Latin language, as the Lorsch master's lesson about thedifference between coepi and cepi suggests. Masters expanded the storehouseof their students' vocabulary in several ways. First, they glossed authorita-tive texts such as the Bible and the works of the school authors with simple

Elementary instruction in reading and speaking Latin was not limited toclassroom exercises. Daily participation in the liturgy and communalreadings reinforced the lessons of the schola. The Psalter was the first majortext students mastered. Singing and recitation of the Psalms embedded thesounds of Latin in their minds while at the same time initiating students tothe kind of moral formation Hraban Maur described in his De InstitutioneClericorum. The Psalter remained a lifelong companion of Carolingianscholars. When testing new quills out on the fly-leaves of the manuscriptsthey were copying, scribes often used as a pen-trial the phrase Beatus vir fromthe first Psalm. The Psalms came to mind almost automatically whenCarolingian authors wrote. Lupus of Ferrieres drew inspiration fromAugustine seventeen times and from Priscian and Virgil thirteen times eachin his letters, but it was the Psalms that came more frequently to mind inforty-four cases.90 Even in a relatively esoteric work such as the Exposition onthe Celestial Hierarchy, John Scottus recalled Psalm verses fifteen times,second only to his penchant for the Gospel of John which he quoted twenty-one times.

Hraban Maur's De Institution Clericorum illustrates well how secular anddivine wisdom complemented each other in the Carolingian classroom.85This handbook for the training of clergy obviously depicts an idealprogramme. Young men studying for the priesthood were expected toknow the liberal arts - grammar, rhetoric, dialectic, mathematics, arith-metic, geometry, music, astronomy and what Hraban called the 'philosophi-cal books'. At the same time, they received moral and spiritual training andprepared themselves to preach to their flocks. This is the programme ofAugustine, judiciously supplemented by Cassiodorus, Isidore of Seville andGregory the Great's Pastoral Care. But before students could embark ontheir studies, they hadfirstto master the elements of learning and religiouslife — reading, writing, computus and chant. Training in these basic skill

Despite the pervasiveness of the arts in the schools and in academicdiscourse, it would be misleading to compress the reality of Carolingianlearning to the liberal arts programme. First of all, on a theoretical levelmany rival schemata of the arts competed with each other. The question ofthe priority, order and relationship of the arts to each other remained avexing issue to masters throughout the Carolingian period and beyond.74Alcuin helped to 'Christianise' the arts by demonstrating how grammarcould serve as the handmaiden of theology. His description of the temple ofChristian Wisdom supported by the seven 'columns' of the arts put intographic form the relationship between secular and divine learning.75 So didthe notion of John Scottus approximately fifty years later when he taughtthat the arts come together as the tributaries of a stream to be united in thecontemplation of Christ.76 Such confident assertions were not without theircritics and throughout the ninth century nagging questions about theappropriateness of the arts to Christian education persisted. John Scottus'use of the art of dialectic to understand the issue of predestination wasroundly criticised by Bishop Prudentius of Troyes (c. 846—61).77 Agobard ofLyons was concerned that specialised study in music produced cantors whowere very narrowly educated, deficient in their spiritual formation and

An active and influential school depended for its vitality on the resourcesnecessary to copy books or to obtain them from other centres. Hameln, atiny dependency of Fulda with its twelve monks and eleven students, musthave had minimal library resources in contrast to St Gall and Lorsch, whoseninth-century library catalogues list 264 and more than 450 codicesrespectively, or Corbie, from whose library almost 250 manuscripts andmanuscript fragments survive.65 Books were at the heart of Carolingianeducation. Cathedrals and monasteries that were endowed by their patrons

Approximately seventy schools were sufficiently active in the ninthcentury to leave some record of their activity.59 On a map these centres forma rough triangle with the broad base extending from Hamburg in the northto Tours in the southwest and with the apex in Italy, roughly near MonteCassino (see Map 20). Most of the schools in this area were concentrated infour regions: north of the Loire, in northern Italy in the former Lombardkingdom, in the parts of Germany that had seen the most intense missionaryactivity in the eighth century, and in the Spanish March and Septimania.Among major regions, only Aquitaine stands virtually devoid of knownschools although some forty saints' Lives were composed in CarolingianAquitaine and the historian Ermoldus Nigellus and the author of theWaltharius poem - if they are not the same person - came from that region.60The influence of the Carolingian courts along with epistolary and personalexchanges combined to establish loose networks among the schools. At thediocesan level, bishops such as Theodulf of Orleans and Hincmar of Rheimscould exert some measure of control and organisation over the schools intheir jurisdictions. Lothar I authorised an ambitious scheme in 825 whichwould have funnelled masters and students from throughout northern Italyinto nine centres (Pavia, Ivrea, Turin, Cremona, Florence, Firmo, Verona,Vicenzo and Cividale) for further schooling. He even offered to provide forthe schools to encourage participation.61 Four years later the bishopsgathered at the Council of Paris tried to enlist the support of Lothar's father,Louis the Pious, to establish three schools under royal patronage.62 But boththese efforts appear to have been stillborn. Dungal, the Irish master placed incharge of the school at Pavia, retired to Bobbio within two years. Lothar's

For most young boys, the path to that vocation began approximately atage seven in the monasteries and two or three years later in the cathedralschools. School was but one part of their daily routine. Children engaged inappropriate forms of work, assisted in prayer services, and, in general, wereacclimatised to the ecclesiastical regime. Ninth-century monasteries andcathedral cloisters were generally solicitous of their young charges. Theyconstituted the next generation of clergy raised up from a tender age withina religious atmosphere and therefore uncontaminated by the evils of theworld.53 Precise information on the number of students at any one time isdifficult to come by. If the lists of monks and scholastici from Fulda'sdependencies are representative, students could account for between 26 and49 percent of the adult population.54 In small monasteries and cathedrals,young pupils studied with a magister or grammaticus who did everything,even tripling as the director of the scriptorium and as the librarian. Larger,better endowed schools with more pupils might have several masters whowould teach a speciality: chant, copying, grammar, explication of Scripture.The relationship between students and masters was an intensely spiritualand personal as well as academic one. Alcuin remembered fondly all hisformer pupils, called them by pet names, and lamented that his studentscame and went, but that the 'old man' remained.55 Heiric of Auxerrepublished part of the lectures of both his teachers, Lupus of Ferrieres andHaimo of Auxerre, and prefaced them with a poem in praise of his masters.56Ercanbert of Fulda wrote down the lectures of his master, Rudolf, becausehe was concerned that the master's instruction would fade from memory ifnot committed to writing.5

Women active as scribes supplied the needs of their own communities aswell as those of other houses. The list of female writing centres, Chelles,Jouarre, Remiremont, Sackingen, Poitiers, Herford, Soissons, Essen, Bres-cia, continues to grow. Nuns at Vreden and Neuenheerse copied outreliquary tags and probably also books. The important 'Corbie a-b' scriptwhich survives in thirty-nine manuscripts and fragments has been attributedto the skill of female scribes.49 Copying, like embroidery and working withcloth, could at one level be viewed as manual labour which in the monasticroutine along with reading and prayer was both pleasing to God and a usefulantidote to idleness.50 But books were meant to be read and to be ponderedand occasionally to beget new books. Women read intelligently, as theexamples of Gisela and Rotrud, Dhuoda and the nuns of Soissons demon-strate. But with the exception of Dhuoda, did they contribute to Carolingianliterary culture? From the time of Peter of Pisa in the last quarter of theeighth century to the time of Remigius of Auxerre (died 908), approximatelysixty male authors are known by name. These authors did not writeeverything that has survived from the period and many works remainunattributed. Of these, histories such as the Liber Historiae Francorum, theAnnales Mettenses Priores, the Annales Quedlinburgenses, as well as Vitae offounders of female monastic houses may well have been written bywomen.51 And just as recent palaeographical research suggests that it is nolonger possible to assume that unattributed manuscripts were copied inmale scriptoria, so too the large collection of excerpts, glosses, andpedagogical and devotional extracts that survive cannot be attributed solelyto the activity of male compilers and readers.

The most visible female teacher of the Carolingian period is also its onlyknown female author. Dhuoda's case is made even more remarkable by herlay status. She wrote her Liber Manualis in the early 840s to instruct her son,William, in a way of life that blended Christian principles with thearistocratic virtues of family loyalty and respect for paternal authority. Thatshe expected her sixteen-year-old son to read her book and went on to urgehim to read many of the books of the learned masters to learn more aboutGod suggests that William was himself well educated and that Dhuoda mayhave been his first teacher.46 The roster of lay women who participated inliterary culture extends beyond Dhuoda. Judith, the second wife of Louis

A capitulary intended to establish empire-wide policy in 802 barred allmen from female monasteries except for priests who, accompanied by awitness, could visit the sick; they could also say Mass for the nuns, but were

The dearth of known women writers during the Carolingian perioddoes not necessarily mean that only boys went on for further schooling andthat opportunities for creative intellectual activity were non-existent for

Fulco knew that parish priests bore primary responsibility for thereligious education of the Christian people. They had to make sure thatgodparents knew both the Lord's Prayer and the Creed before they couldsponsor a child in baptism and that they could teach these basic Christianformulae to their godchildren.26 The few parish inventories surviving fromthe ninth century suggest that rural priests stocked modest libraries and,therefore, could read and presumably teach reading and writing to child-ren.27 It is most unfortunate that so little is known of these elementaryschools for through them some measure of literacy, enough to assureminimal participation in the liturgy, reached the broadest level of Carol-ingian society — the children of free parents as well as the children of peasantsif the Admonitio Generalis was obeyed.28 Angilbert assumed that the men,women and children of the seven communities near St Riquier wouldparticipate actively, singing and following banners which probably borelegends, in intricate processions during the liturgy for Rogations; Pascha-sius Radbertus envisaged peasants lamenting the death of Adalhard ofCorbie in their own tongues and also joining in with clergy in an antiphonal

These new directivesflewin the face of tradition.

The legislation of the closing decades of the eighth century clearlyassigned to monasteries and cathedrals responsibility for educating theclergy who would lead God's people. Neither the Admonitio Generalis northe Epistola de Litteris Colendis distinguished between the different functionsof monasteries and cathedrals, no doubt because in missionary and frontierregions monks had long functioned as secular clergy. One of the provisionsof Bishop Theodulf of Orleans' capitula for his parish priests suggests howblurred the lines were between clerical and monastic education. Priests inTheodulf s diocese who wished to send any of their relatives to school hadthe bishop's permission to send them to the cathedral school or to any one ofthe monastic schools under the bishop's jurisdiction.17 There is no hint herethat those who chose to learn their letters in a monastic school intended tobecome monks. Theodulf was not an innovator nor was his invitation forstudents to use the resources of local monasteries unique. When Louis thePious and Benedict of Aniane set out to reform monastic life, the presence ofso many non-monks within the monastic precincts and within the routine ofmonastic life was, from the perspective of the reformers, alarming. In 816,the Council of Aachen established guidelines to reform cathedral canonsalong monastic lines and to isolate young men training to become canons

The Carolingian court provided the setting for one kind of school, thepalace school. All the Carolingian monarchs, but especially Charlemagne,Louis the Pious, Lothar I, Lothar II and Charles the Bald drew scholars totheir courts. These masters acted as advisers on all sorts of matters for theirpatrons, wrote poetry for them, and dedicated their books to them. Theyalso had a hand in teaching royal and aristocratic children such as the youngCharles the Bald who was tutored by Walahfrid Strabo. The court school, asfar as it can be detected, was a loosely organised institution. Its memberswere transient, and after its heyday in the 780s and 790s its leading membersdispersed to the monasteries and cathedrals of the realms where theyestablished more formal educational institutions.1

The prismatic effects of the Carolingian programme were furtherrefracted by the stances thoughtful writers took about the shape of theirsociety. Enthusiastic court, cathedral and monastic scholars responded tothe challenges of the Carolingian reform programme, but they were no merepuppets. While some 'Christianised' the arts, others worried about therevival of pagan models.14 In the heady days of the great king and emperor,Charlemagne, court scholars used their talents to legitimise and praiseauthority; a later generation of intellectuals would criticise and instructerrant political leaders.15 And in what is perhaps the surest sign of thevitality of learning in the Carolingian realms, scholars engaged in polemicon critical political and theological issues and even proved themselves adeptat forgery.

These accomplishments give substance to the notion of a renaissance in

Seen from the top down in this fashion, the Carolingian renaissanceappears as a well-organised programme which, given the authority behindit, could not help but achieve its goals. Observers from the time of NotkerBalbulus and Heiric of Auxerre, who celebrated the intellectual brilliance ofCharles the Bald's court, to the present day have been impressed by theCarolingian achievement.9 Monastic, cathedral and court scriptoria toiledindustriously to provide Carolingian Europe with the texts needed forserious study, worship and government — perhaps as many as 50,000 in theninth century alone by one estimate.10 Books counted as treasure andformed an important part of the venerable gift-giving tradition among theFranks. When Emperor Lothar made peace with his brother, Charles, in849, he entered into a spiritual association with one of the foremostmonasteries of his brother's realm, St Martin at Tours, and sealed hiscommitment to the community by underwriting the copying and decorationof a sumptuous Gospel Book, the Lothar Gospels.11 Librarians and teacherscarefully built up the collections of their local libraries and constantlysought copies of books they did not possess. The libraries supportedschools, many of which in the ninth century offered sustained instruction fortwo or three generations of masters.12 Carolingian scholars were asproductive as their colleagues in the scriptoria. They revised the biblical textseveral times, produced compilations of patristic works, wrote their owncommentaries on the Bible and classical texts, and glossed treatises on thearts for their students. They wrote histories and poetry and exchangedinnumerable letters among themselves.

Royal patronage inspired an outpouring ofpanegyric poetry, biblical commentaries, histories, translations of the GreekFathers into Latin, and even the financing of three schools.8 Ecclesiasticalauthorities such as Claudius, bishop of Turin, Theodulf, bishop of Orleans,Hraban Maur, abbot of Fulda and archbishop of Mainz, Lupus, abbot ofFerrieres, and Hincmar, archbishop of Rheims, among others, were alsoformidable scholars who made their cathedrals and monasteries intellectualand educational centres

A programme of such ambitious scope might well have remained only aset of ideals had it not been pursued with the kind of determination thatCharlemagne directed towards the Saxons. It is worth noting, too, that hislongevity meant that he was able to give continued impetus to theprogramme. His sons and grandsons in their legislation and the clergy inepiscopal councils and statutes throughout the ninth century continued toemphasise the importance of book learning and study.5

The mission, defined in the Admonitio Genera/is, in the Epistola de LitterisColendis and in other programmatic documents, was simple enough but hadprofound implications.2 What the leaders of Carolingian society wanted todo was to prepare the clergy, 'the soldiers of the Church', to lead 'the peopleof God to the pasture of eternal life'. The evidence of poorly written lettersthat had come to the court signalled that the clergy needed schooling, forignorance of language went hand in hand with a more serious lack of

I am well aware of the dangers of maximalist interpretations, but itseems to me that the evidence for the education of the laity, though hardlysubstantial or overwhelming, does point to a far stronger possibility of thewidespread provision of basic instruction, at least among the nobiles andpotentes, if not the mediocres and fortunate pauperes or infimi, than hashitherto been acknowledged. There are too many 'exceptions' for the'rules' to which they were once related to hold any longer. The 'rules'themselves must be changed to allow for the acceptance of the existenceof the wider spread of basic literacy which provided the necessaryfoundation for administrative careers, for the conduct of judicial businessdependent on the use of the written word and for the practice of theChristian faith.

ow exceptional was Dhuoda ? Is she the model of a noblewoman, or arare instance of a Carolingian bluestocking, bound in marriage for the sakeof her family and beholden to her husband for support thereafter? Hername is thought by most scholars to suggest a north Frankish originrather than membership of a family in the region of Uzes where herhusband's property lay, and to which he bid her retire at some stage after826. Dhuoda had been married to Bernard in the palace chapel at Aachenon 29 June 824, but this is more likely to reflect her husband's position inrelation to Louis the Pious than that of her own family.49 She was,therefore, an educated woman of a Frankish aristocratic family, well read,and with a high sense of duty and loyalty. If she is a random example ofthe educational attainments of a Frankish noblewoman (in the sense thatchance survival has preserved her treatise for posterity) is it justifiable tothink in terms of her being representative of her class? Of course oneshould like to think so. Fragments of other evidence extant, in the form ofworks dedicated to or commissioned by laywomen, books owned by them,borrowed or given to them, indeed, make an assessment of Dhuoda as onewell-documented instance of a general phenomenon more than merewishful thinking.

What of Dhuoda herself? She wrote her Liber Manualis between 840 and843 in her home in Septimania. It was a moral treatise, full of both thearistocratic ideals of honour and fidelity to one's lord and Christian moralteaching. Dhuoda does not usually cite her sources, except for Donatus,Isidore's Synonyma and Gregory the Great's Cura Pastoralis, but she doessay she wrote from books in her own possession. Some of her sources havebeen identified. They include the Liber Cathemerinon of Prudentius,Augustine's Enchiridion, the Tractatus in lohannem, various homilies andthe Moralia in Job of Gregory the Great. She had also read some saints' livesand books on virtues and vices, including those by Alcuin and AmbrosiusAutpert. She knew her Bible well, especially the Gospel of St Matthew,

On his departure, at sixteen years old, to serve Charles the Bald, Dhuodapresented her son William with a manual of good conduct, from thereading of which he was clearly intended to derive benefit.43 He was,furthermore, going to serve at a court where he can hardly have been theodd man out, or eccentric, in being possessed of the ability to read. It maywell have been Dhuoda who taught her son. Whatever the case, sheherself was adequately educated and believed with all her heart that it wa

Although most of the evidence about schools relates to the training ofthe regular and secular clergy, it would be a mistake to assume thateducation and learning were confined to clerics. Some laymen clearly didgo to school, but how much they learnt, and whether their education wasseparate from those intending to enter, or who had already entered, thechurch cannot at present be established. Notker Balbulus' story from hisGesta Karoli is relevant in this respect. Notker describes Charlemagne'sanger at finding the sons of his lay magnates slothful at their studies. Theking had entrusted boys of noble parents, mediocres (boys of the middleclass), and poor boys to Clement the Irishman for instruction. Thecompositions and poems of the noble boys were weak and full of errors.They were scolded for their neglect of the pursuit of learning. The goodboys, that is, the mediocres and the infimi, were to be rewarded withbishoprics and monasteries.39 An infimus who had especially distinguishedhimself was transferred to the palace chapel. The castigation of the sons ofthe nobility can be understood as a warning that they would never receivehonours and offices from Charlemagne unless they acquired a respect forand knowledge of those things Charlemagne wished to promote. Whetherit was Charlemagne or his successors who emphasized the value ofeducation in quite this way cannot be established. The story, nevertheless,is important in so far as it indicates the opportunities that were envisagedas emanating from the patronage of the royal court, and that it wasaccepted that even those not necessarily destined for an ecclesiasticalcareer would be expected to acquire some Latin learning. They needed itto qualify for administrative office.

There were clearly some schools in the Carolingian realm open tolaymen. The chronicle of St Riquier by Hariulf, furthermore, preserves thefamous description by Abbot Angilbert of a liturgical procession in whichboys of the lay school and boys from the abbey school took part.34Angilbert did not make it clear whether one had to belong to a particularsocial group or intend to join the church to go to one or the other school,but it is possible that the lay school was the place for elementaryinstruction before one decided, one's parents decided, or particularaptitude made it appropriate, to pursue a clerical career. In other words,a local monastery could provide basic education for lay boys. TheAdmonitio Generalis of 789 had decreed that schools were to be set up forboys in every station of life, but in 817 Louis the Pious' conciliar decisionsat Aachen included the statement that only those intending to be monksshould be taught in monastic schools.35 He may here have been making adistinction between monastic schools for aspiring clergy and lay schoolsfor others, though he does not mention the latter. As the decrees wereconcerned wholly with monastic reform this is perhaps not surprising. Itremains a possibility, however, that Louis' measure was the first step

We can add to this hint from the Vita Liutbergae as to the existence ofschools. Another is recorded in the Vita Liudgeri. This describes how thesaint, as soon as he could walk and talk, began to collect bits of animal skinand bark and sew them together to make little books for himself withwhich he and the other children would play. He then began to write booksand read them for himself, and swore that God had been his teacher.Thereafter he asked his parents to entrust him to some man of God forinstruction. He was sent to Gregory of Utrecht in whose school wereeducated other 'noble and wise fellow pupils' some of whom becamebishops and others 'in lower orders' became teachers in churches.31 Thestructure of the passages concerning Liudger gives the distinct impressionthat the education came first and that only subsequently did he put aside

Benedict of Aniane, too, before he turned to the monastic life, was, as alayman, learned in unusual studies, but he gave up all those things inwhich human weakness is accustomed to take pride.26 The Old Latin Lifeof Gerard of Vienne records, for what it is worth, that he was well versedin law as well as being a good warrior, well educated and active in theroyal administration.27 As Riche reminds us, the Lives of Landeric, Clodulf,Paul, Gerald and Poppo tell of the saints when young being entrusted toclerics, not to be made into clerics in their turn, but to be given the literaryinstruction suitable for a noble.28 Other lives, such as those of Gerald ofAurillac, John of Gorze, Bernard of Menthon and Theodoric of Andage,describe the children of the nobles being instructed at home.29 Girls werenot necessarily excluded from this. Liutberga of Halberstadt is said to havetaught girls the psalmody and to let them go home after their lessons orelsewhere as they wished.30 That is, she was clearly not simply instructingnovices destined for an enclosed life in a religious community, but teachingwithin a region in a 'public' school

That other nobles shared the experience of basic instruction but wereable to attain rather more is evident from such accounts as John ofSalerno's description, derived from Odo himself, of Odo of Cluny's father,Abbo, reporting Odo's words: 'My father knew by heart the ancienthistories and the novella of Justinian. In his conversation there was alwaysto be found something of the Gospel. '24 Odo himself, when old enough, washanded over to a certain priest who dwelt in a remote place to be educatedand introduced to the study of letters. Again there is the hint that furtherlearning is only permitted if a military life seemed impracticable: ' As timewent on my father began to withdraw me from the ecclesiastical life andsent me to military exercises, and with this purpose he sent me to serve aspage in the household of Duke William. '25 Thus a basic education wasreceived by boys at a tender age before being removed from a tutor's careto continue their education in the physical skills required of the noblewarrior class. It may be that the phrase vita ecclesiastica is here used as a

On the face of it, this account would appear to support the prejudices ofthose who believe the lay nobility received no instruction in reading, foronly when some affliction, perhaps dreadful acne, made his parents fearthat he would have to be diverted towards an ecclesiastical career, was theboy Gerald permitted to pursue his studies. This is to miss the significanceof the account of Gerald's earliest instruction. He was only permitted to gothrough his psalter. This could simply be a reference to Gerald's ability torecite the psalter by heart. It may be the outcome of purely oral instructionand he may not even have been able to read. But the text does say hestudied letters and that he subsequently learnt grammar, which in aCarolingian context appears to have meant learning to analyse a languagealready known.22 The Latin psalter is in any case a demanding compilationof figurative and evocative poetry. It may have served as a primer, but itwas a primer that provided a key to and a means of expression of faith inGod23 in an idiom that was the foundation of much of Carolingian Latinprose. It seems clear that Gerald received a standard basic education, andonly extended it to more advanced learning when it appeared that hemight be unfit for a martial career. This accords with the impressionprovided in the Merovingian sources of the content of the basic instructionof freemen, that is, a grounding in the fundamental Christian texts.

It is becomingincreasingly clear that the essential foundation for Carolingian literacywas Merovingian literacy, and that there was no dramatic break betweenthe Merovingian and Carolingian periods. Nor was there decline in layculture by the eighth century. Yet here, interpretation of the evidence forthe seventh and eighth centuries would appear to be largely a matter ofperspective. Pirenne, Riche and others thereafter interpreted sympa-thetically for the Merovingian period, both in quality and quantity,precisely the same kind of evidence they dismissed as inconclusive,exceptional or too sparse to be significant for the Carolingian period.Riche's own work has contradicted his own conclusions implicitly, and, inhis brief survey of the education and culture of aristocrats, even explicitly,but it does not seem yet to have changed current perceptions ofMerovingian and Carolingian literacy as a whole.7 The Carolingianmaterial, far greater and less ambiguous than that of the Merovingianperiod, is ripe for more sympathetic and positive assessment. A beginningtherefore is made in this chapter; it builds on the foundations laid by Riche,though much of it is necessarily speculative.

For my conclusion is that the limits of the Carolingianrenaissance hardly exceeded, even in the ninth century, the dimensionsof a religious culture that was largely confined to the clerical andmonastic orders—what contemporaries in fact so often meant by theterm ecclesia. At the same time, in lay society, pre-Carolingian legaland religious ideas and practices—and, I suspect, (though this hasbeen beyond the scope of this paper) political ones too—persisted andevolved with a momentum of their own, affected but not determinedby ecclesiastical novitas.

Did the Carolingian renaissance then have no effect on secularlegal practice or—which is something different—on ideas aboutsecular law? Its direct effect was significant in practical terms inonly one sense: Carolingian scholars copied and preserved thetexts of the laws, and ecclesiastical institutions in using these texts(as, for instance, St Gall in a whole series of land cases appealed to theLex Alamannomm)iS contributed to their continuing vitality. Theformal characteristics of later Carolingian capitularies owed much tothe improved latinity achieved through the renaissance of scholarship.On the other hand, a tendency to rely on the written word in legalprocedures, though it probably increased simply through the readieravailability of scribes, was already common in the vulgar private lawof the fifth century and had shown continuous if patchy developmentin the practice of the barbarian kingdoms.4

What UUmann argues, ifI read himcorrectly, is that in the 'totalitarian' world of this medievalChristian society, personal renewal implies institutional renewal:those same Christian norms which applied to the baptised individualwere, he suggests, applied quite naturally to society as a whole. 'Theeffect which this Carolingian renaissance in the social sense was toproduce in the public field was a "baptism" on the largest conceivablescale'. To the question of how this was achieved, Ullmann answers:by the absorption of 'ecclesiology . . . into the governmental systemitself. For, 'just as the individual, through the juristic effects of baptismwas incorporated in, or absorbed by, the Church, in the same way thecomponent groups of Frankish society were absorbed within thecorporative union of the Church'. Ullmann goes on to argue that themajor instrument of his social and governmental renaissance was 'thelaw applicable to the whole of Frankish society'; and in subsequentparts of the book, he attempts to show how Carolingian legislationaimed at the suppression of 'Frankish or Germanic or any othernaturally grown habits and usages' by 'the laws of God'. Ullmann'snovel approach seems to me to focus upon a vital question, and hisconcern with law points to a fruitful source of answers. In whatfollows, however, viewing the problem from a different stand-pointand Carolingian law in a different perspective, I reach some differentconclusions.

Like Marx's Aufhebung, Carolingian ideas of rebirthtranscended any crude polarisation of 'conservative' and 'revolution-ary'. Carolingian scholars perceived their present as fully continuouswith the Roman, and especially the Christian-Roman, past. This senseof continuity through renewal presents, I think, a noteworthy contrastto the renaissance ideologies of the fifteenth/sixteenth and, in somedegree, even the twelfth centuries. Historians prone to emphasising thealleged novelties of the Carolingian age should at least consider theimplications of re-viewing it in Carolingian perspective. Third, andperhaps especially in this period of Christian expansion, the rebirthmetaphor could have reference to baptism, the sacramentum regenera-tions through which a person is reborn into the church.14 In this case,the rebirth, being a personal matter, could be interpreted more dir-ectly in terms of prevailing notions of community. I shall return tothis point below.

Or was he, as Ladner suggests,particularly attracted by 'the idea of a society embracing earth andheaven, a society which a man could join through personal renewal'?If Ladner is right, then, he tells us, we should talk not of a Carolingianrenaissance—'secondary classicising features notwithstanding'—butof a Carolingian reform 'asjust one phase in the unfolding history of therealisation of the Reform idea in Christian history' and specifically 'anattempt to recreate the religious culture of the fourth and fifth cen-turies'.4 But is Lander right about Charlemagne? I have my doubts:perhaps what he really enjoyed most was book 22's meaty chapter onthe resurrection of the flesh or its rattling good miracle-story

Muslim civilization acted as teacher to medieval Europe of otherbranches of knowledge as well as philosophy and medicine, these beingmathematics, astronomy and astrology. Here too the legacy of Classicaland Hellenistic Antiquity was presented to the West enriched with thefurther studies, comments and experience of Islamic science, one proofof this being the number of technical words that passed from Arabic intoLatin and the other languages of Western Europe, e.g., algebra, al-gorithm, zenith, nadir, azimuth and cipher. The work of the great Arabmathematicians, astronomers and astrologers (it is not always easy todistinguish the three activities) was among the features of Islamicscience that appealed to the translators of Toledo or at the court ofAlfonso X the Wise of Castile and Leon, and in general everywhereduring the twelfth and thirteenth centuries. Here we can only give abrief list, in chronological order, of the authors who were most widelyknown and studied in the West. We begin with the great al-Khuwarizmi,whose name, as is well known, as a result of medieval Latin distortions,gave rise to the term algorithm; his little treatise on algebra, the earliestof its kind in Arabic, was translated into Latin twice during the twelfthcentury, by Gerard, who retained the Arabic title, Dejebra et alnmcabala,and by Robert of Chester, who gave an exact Latin rendering of it,Uber restaurationis et oppositionis numeri, while al-Khuwarizmi's astronomi-cal tables, as rearranged about the year 1000 by Maslama al-Majriti, weretranslated by Adelard of Bath. With a contemporary of al-Khuwarizmi,Abu Ma'shar (the Albumasar of the Latins, d. 272/886) we pass frompure mathematics to astronomy and astrology; his great introduction toastrology, al-Madkhal al-kabir, was translated by Johannes Hispalensisunder the title of lntroductorium maius, and in abridged form by Hermanthe Dalmatian; his Daldldt al-ashkhds al-ulwiyya was also translated byJohn of Seville under the title De magnis coniunctionibus et annorumrevolutionibus. Both these works had a great influence on Westernastrology, one reflection of them being the representation of the tendegrees of the zodiac as described by Abu Ma'shar on the frieze

The nexus between philosophy and the sciences, which dates from theorigins of Greek thought and can be followed throughout Antiquity,was bequeathed to the Arabs as part of the ancient heritage and was bythem transmitted to the Western world. Just as Hunayn b. Ishaq and hissuccessors turned their attention to Greek science, in particular tomedicine and mathematics, so did the Latin translators devote themselvesto Arab and Greek works of pure theoretical speculation (Aristotle andpseudo-Aristotelian logic and metaphysics, Plato, and their greatMuslim commentators) and at the same time to the patrimony of antiquescience or pseudo-science that Muslim culture had greeted so eagerly andwhich it had so much enriched. Our own differential specialization tendsto make us break down this nexus, and deal separately with each singlebranch of thought and knowledge, but in the sphere of medievalcivilization, whether Eastern or Western, this unity of conceptionmust never be overlooked. The particularly close connexion betweenphilosophy and medicine (of which there is a reflection in the ambiguityof the Arabic word hakim, often used indifferently to denote either

In the field of philosophy it is generally maintained that what theWest knew of Greek thought, and in particular of Aristotle, wastransmitted to it by the Arabs. Such a statement needs qualification anda more precise formulation, but on the whole it remains valid. Inreality, the direct channel of transmission through Byzantium was nevercompletely closed to the West, and during the twelfth and thirteenthcenturies works of Plato and Aristotle were translated directly fromGreek into Latin (the Meno and the Phaedo in Sicily by Enrico Aristippo(d. 1162); the Metaphysics, the Nicomachean ethics, the Physics and the Deanima during the first half of the thirteenth century). Of some works themedieval Latins received two translations almost at the same time, onefrom the original Greek and the other from Arabic. Yet, at the end of thethirteenth century, in one of his most famous passages, Roger Baconcould affirm that the knowledge of Aristotelian philosophy had re-mained hidden from the West since the days of Boethius and had beenrevived in his own time thanks to Arab mediation, and above all to IbnSina. And it is a fact that during the late Middle Ages and the Renais-sance, Greek philosophy was studied in the West on the basis of Arab

What impressed the West in the intellectualachievements of the Arabs was the role of mediators of Greek philosophyand science which they had assumed, and the impulse they had impartedto the various branches of learning. The attitude of the Latin Westtowards the ancient heritage, and in particular to Greece, was much thesame as that of the Islamic East—indifference to the artistic element, butkeen interest and admiration for the philosophic and scientific aspects,direct contact with which, however, was generally precluded byignorance of the language. Now it was discovered that these barbarianinfidels had translated into their own tongue the wisdom of the ancients,

My concern here is to argue that it is now Islam’sturn to be seen as Antiquity’s “natural end,” albeit thanks to the drift of the

Latterly, Aristotelianism’s role in the indispensable intellectual underpin-nings of the period we are concerned with has been underlined by the An-cient commentators on Aristotle project guided by Richard Sorabji, whichhas liberated this whole thought world from the dignified obscurity im-posed by the Berlin edition’s twenty-three stout, austere volumes (them-selves originally intended, and used, mainly as a mine for fragments of thePresocratics, Peripatetics, and Stoics35). In tandem with Sorabji’s project,research has intensified on the Syriac and Arabic Aristotle translations andcommentaries.36 What is emerging is a picture of a coherent and profoundly

But what truly holds the First Millennium together is the question of Is-lam’s relation to the deeply rooted cultural traditions that dominated theworld it was born into. It was a precondition of Islam that rabbinic Judaismand patristic Christianity should have matured (though in Christianity’s casethat also meant fully ventilating its doctrinal inconsistencies and improbabil-ities) before it appeared; but this does not mean that the First Millenniumfalls into two separate, independent halves. It is only the dialogue of continu-ing Judaism and Christianity—especially continuing Syriac Christianity17—with Qurʾanic, then maturing Islam that generates the potent synthesis wefind in tenth-century Baghdad.What exactly do I mean by the “maturation” of a tradition?18 In any tradi-tion there will be conservatives and progressives. The former will believeenough has already been said, and maturity attained, even if the founder orthe founding event is still a fresh memory and a potential stimulus to furtherdevelopment. If we give heavy weighting to the perspective of contempo-raries, we will find the concept of maturation too controversial to be helpful.From the historian’s viewpoint, though, a tradition may reasonably be calledmature if it has—first and most fundamentally—acquired a clear sense (orsenses) of what it is and what it is not.19 Notions of “orthodoxy” and “heresy”may already be invoked; but that stage may come quite early on. A maturetradition needs also to have built up enough of its institutions and doctrinesthat it seems to correspond to what we perceive, now, to be its broadly char-acteristic articulation, however self-contradictory. I do not intend any neces-sary rigidification in the “mature” tradition, but rather its arrival at a pointviewed by a widely influential sector of posterity as paradigmatic, worthy ofimitation (or “classical”).20 After all, “decline” does not necessarily follow im-mediately upon maturation. Often we see diversification into channels lessobviously “classical,” but no less vigorous.

In 1999, the same year Andrea Giardina denounced late Antiquity’s elephan-tiasis, was also published what still stands as the most recent and authorita-tive statement of the maximalist position, namely Harvard’s Late Antiquity:A guide to the postclassical world, edited by Glen Bowersock, Peter Brown andOleg Grabar.1 By taking as its cutoff point approximately the year 800, thisweighty tome espouses—and up to a point exemplifies—the view that theearly Islamic world shows significant continuities with late Antiquity. But at

European civilization makes little sense without the religious and philo-sophical developments that occurred during the long late Antiquity. The re-fusal of many in the Islamic world to acknowledge the late antique pluralismto which the Qurʾān responds undermines their grasp on history and theiraccess to the context and contacts which are Islam’s birthright. Clearly, then,the attempt to understand the pre-Islamic world of ideas offers a profoundlyserious, nonantiquarian reason—and one relevant to developments today—for taking the period into consideration. The question remains, though,whether even the long late Antiquity is a sufficiently broad stage for the in-vestigation proposed. Is something “beyond late Antiquity” required? As weshall see, Peter Brown’s apparently generous terminus at c. 800 ends up treat-ing Islam as merely an extension of late Antiquity, rather than allowing it toreach a stage of intellectual and institutional maturation comparable withfully developed patristic Christianity, or capable of being used as an approachto the Islamic world we know today. If, then, we are to have the full benefit ofstudying early Christianity and Islam comparatively but not ahistorically, inother words within a firm sociohistorical framework, we need an alternativeto the late antique paradigm. This is what I shall provide in the next chapter

Current scholarship maintains, in the light of a great deal of archaeologydone since Pirenne’s day, that there was less commercial continuity than he sup-posed, and that things had been coming adrift ever since the third century.Trade gradually got more and more localized; Pirenne concentrated too muchon literary sources that focused on exotic long-distance exchanges and the lux-ury goods favored by the elites they wrote about. On the other hand, it contin-ues to be held that there was indeed a further relaxation of Mediterranean-wide communications and commerce, reaching its nadir c. 700.84 Our concernhere, though, is not so much the economic realities of late Antiquity as thehistoriographical trends of the twentieth. Not only was Pirenne’s thesis widelydebated, and indeed adopted by many; it was also perceived to have moraljustice on its side. Who better than the Muslim Arabs, doubly alien, to assumethe role of Antiquity’s executioner? Hence the still widespread conviction thatlate Antiquity lasts until c. 600 rather than ending in 410, or 476 with the lastWestern emperor, Romulus Augustulus, or in 529. After Pirenne, a student oflate Antiquity might elect to include the Germanic successor states on formerlyRoman soil, while a student of late Roman history might exclude them, as didA. H. M. Jones in what remains the standard reference work,85 The later RomanEmpire 284–602: A social, economic and administrative survey (1964). But allwere agreed that the Arabs and Islam were not relevant. Their coming markedan entirely new epoch not just in religion but in society, economy, and admin-istration—not to mention the language of the sources.

For Pirenne, a low-level Romanity survived parallel to theestablishment of the Germanic kingdoms, mainly because the basic patternsof commercial exchange also persisted (Pirenne’s famous Syrian merchants inGaul). It was only with the Arab invasions in the seventh century, especiallywhen they spread across North Africa, that Mediterranean unity broke.Thereafter the Latin world turned in on itself, and substituted Frankishroughness for the old Mediterranean sheen. Just as for Becker Islamic civili-zation could not have come into being without Alexander, so for PirenneCharlemagne was inconceivable without Muhammad.82 The difference wasthat for Becker (reading the Arabic sources) the Arabs’ role was laudable,while for Pirenne (dependent on the Byzantinist Vasiliev) it brought about apermanent rupture.83 But in the end both scholars were interested in onething: the genealogy of Europe, not the development of a complex culturaltradition up to and through Islam.

The destruction of the German and Ottoman Empires, not to mention theAustro-Hungarian and Russian, along with deep disillusionment over thepeace settlement throughout the Muslim world, created a blasted intellectualand political landscape without prospects for rapprochement between theIslamic world and Europe, despite Becker’s continuing pleas. Historians ofthe period in which Islam first emerged found nothing in their new milieu tostimulate them to an inclusive or universalistic approach. As for students ofthe fifth-century West, they could hardly fail to notice that then, too, Romancivilization—in Gaul, for example—had faced Germanic barbarism or, inmore modern terms, Teutonist ideology and rejection of Romanocentricity,in the manner of Strzygowski.75 After 1918, French and other scholars

The eastward pull of the vast mass of Pcrsia in the Islamic empirewas the salvation of Europe. It was not the Grcek fìre of the Byzan-tine navy outside Constantinople in 7r7, nor the Frankish cavalryof Clrarles Marcel ac Tours in73z, that brought the Arab war-machincto a halt. It was the foundation of Baghdad. 'With the establish-ment of the Abbasid califate, the slow-moving ideals of an organizedand expensive imperial administration replaced the fearful mobilityof the Beduin armies. In the new civilian world, the soldier wasas much out of place as he had been among the otiose aristocratsof the fourth-century 'West. The bloodsucking relationships of theHoly War, by which the early Arabs had first irnpinged on thc out-side world, gave way to a meticulous diplornacy modelled on theprotocol of the Persian ancien rêgime. At the court of the califs, the

The division between East and West, which had been blurredthroughout the Late ,\ntique period by the confrontation of Byzan-tium and Persia along the Fertile Crescent, came to rest along the shoresof the Mediterranean itself. Thc Muslim world turned its back onits poor Christian neighbours across the sea. The cultivated man drewhis language from the desert, and the style of his culture from easternMesopotamia. In the more stable worlcl created by this vast shift ofthe balance of culture, western Europe could create an idcntity of itsown. But the stuclent oflate Antiquity, who realizes how rnuch Ettro-pean culture owes to the fruitful interchange betweeir the populationsof the Fertile Crescent, open ât one end to an empire basecl on thc seaand, at the other, to the Iranian plateau, can estimate the cost of thechasrn that 1rx1¡¡¡s¿ across the Mediterranean throughout the Middle,\ges.

For the new commercial opportunities were in Persian hrnds.And, in Persian hands, the eternal lure of Further Asia rersscrteditself, as in the early Sassanian period. The rnosque ancl the fire templecould be seen beside the market-places of Lohang and Canton.Chinese prisoners of war from central Asia brought the art of paper-making to Baghdad in 75r. Sinbad the Sailor woulcl not have consi-dered the Mediterraneân worth his trouble: for the wealth ancl in-terests of the,\bbasid empire poured eastwards, down the Tigris andEuphrates, to the sea route that linked llasra directly with Canton.

Mesopotamia regained a central position that it had lost since theclays of ,\lexander the Great. llaghdad, with its circnlar city wall,owed nothing to the great cities of the Rornan enrpirc': it w¿s anâvatar of the round cities of ,\ssyria and central '\sia. The Mediter-ranean cities declined as the great câravans by-passed them, bringingtracle by camel along the oceans of sancl that strctched from theSahara to the Gobi Desert. In North Africa and Syria, the villagcsthat had sent their oil and grain across the sea to Rome and Constanti-nople disappeared into the sand. The Mediterrancan coast, frombeing the heart of the civilized world, irnpcrceptibly diminished insignificance, as the nnmbed extremity of a grcat Eurasian ernpire

Thus the late seventh and the early eighth centuries, and not theage of the first ,\rab conquests, are the true turning-point in the his-tory of Europe and the Near East. This happened first in a prolongedconfrontation with Byzantium. In the last decades of the seventhcentury, the boundaries between the Christian and the Muslimworlds hardened notably. In ó8o/r, the sixth Oecumenical Council atConstantinople treated the patriarchates of Antioch, Jerusalem andAlexandria as no longer part of the Byzantine Christian world.In ó95, the first fully Arabic coins were minted. In 699, Greek wasreplaced by ,\rabic in the chancery at Damascus. Between 7oó and7r4, the Great Mosque at Damascus was built, to eclipse the tantali-zing magnificence of the imperial churches of Syria and Palestine.The eastern Mediterranean began to take on its Islamic face.

Thus, the Arab empire of the eighth and ninth centuries had to facea similar problem to that of the Roman empire in the third century:the sudden erosion of a proud traditional oligarchy, largely in theinterests of strong government. Just as the narrow certainties of thetraditional Greco-Roman aristocracy were swamped by the diffusedand energetic patriotism of the Roman provincials in the fourthcentury, so the non-Arab Muslims, in the eighth century, came to therescLle of the ,\rab empire. As a result, the culture of the ruling classwidened its franchise: much as the existence of new avenues to power,in the fourth and fifth centuries, fostered the spread of Latin and Greekculture, so - on a far wider scale - the absorption of Arabic, and of an,\rab style of life, opened the court to the provincials of the Near East.Just as, in the Late '\ntique period, Syrians, Egyptians and Cappado-cians, in reading their Homer, took into their lives ideals of behaviourmodelled on the exotic adventures of Mycenaean chieftains, so, fromCordova to Samarkand, educated men, ofwidely different origins andof resolutely urban tastes, spoke classical Arabic and claimed to beacting as true sons of the tents. But whereas, in the Roman empire ofthe fourth and fifth centuries, the traditional culture of the governingclasses remained dominant, remained the giver into whose benefitsbackward provinces were proud to emerge, in the Arab empire of ther98 eighth century a thousand years of civilization found its voice again,

The Arab aristocracy could not maintain its hold on the govern-ment indefinitely, for the Arab supremacy was undermined by Islamitself. Islam made all its converts equal, whatever their racial origin.It opened the floodgates to the gifted or the ambitious non-Arab.As Muslims, Syrians and Persians became the pillars of Islamic civili-zation: they came to be the administrators, the lawyers, the theolo-gians, even, within only a century, the professors of Arab poetry.Medieval Islam was very largely the creation of Muslim non-Arabs.

By 8oo, the traditions formed in the Late Antique period in thevarious countries of the Mediterranean had diverged sharply. Byzan-tium emerged from the crisis of the Arab conquests to find that itsclassical legacy had shrunk to the walls of Constantinople. The ideaof the Roman empire was still very much alive in the streets of the city,in the majestic ceremonial ofthe imperial processions; and a small circleof clerics and courtiers maintained, at Constantinople, standards ofculture that had once been available to the inhabitants of any con-siderable Greek town in the later Roman empire.

This was an exect description of the Near East in the seventhand eighth centuries. Populations formed by the developments ofthe late ancient world found their life continuing, in uacuo, with, ifanything, increased comfort and selÊconfidence. The grain-leviesry6 from Egypt to Constantinople were abolished. A vast Common

To the Arab supremacy, the populations that lay behind theiradvancing armies were not even conquered territories in the strictsense. For they were hardly occupied. They were treated as the richneighbours of the Arabs who paid protection-money to the'Umma,to the Muslims, in return for military defence and as a sort of standingfine for not having embraced Islam. Hence the almost total laissez-Jaire of the seventh-century Arabs. The tax-machines of Syria, Egyptand Persia were encouraged to work smoothly in order to providethe Muslims with pensions. Thus maintained in unparalleled affiuence,the Arab governing class fought out its bitter battles for poweraccording to the laws of Beduin behaviour, in the hermeticallysealed environments of the great garrison-cities - Kufa and Basra,on the edge of the desert facing Persian Mesopotamia, and Al-Fostãt, in Egypt. For them, the conquered provinces were 'a gardenprotected by our spears'.

In the first place, it saved the Arab conquerors from losing theiridentity in the overwhelming mass of their conquered populations.LJnregenerate, utterly selÊconfident and fully articulate, the Beduinstyle ofliving'like a man', lightly Islamized, absorbed and remodelledthe educated classes of the early medieval Near East. The style of life

The victories of the Arab armies created a political vacuum in theNear East. The Byzantines were routed at the battle of the Yarmukín 636 Antioch fell in ó37; Alexandria in 642; Carthage in ó98.The Persian army put up a more stubborn resistance; but after thebattle of Qadesiya in 637, the Sassanian state crumbled. None of thetraditional powers were in a position to win back what they had lostin these lightning campaigns. Only Byzantium survived with itscapital and administration intact. Yet a second Heraclius never came.Hencè an uneasy stillness descended on the eastern Mediterranean.Even under Arab rule, Syria and Egypt remained in close contactwith the rest of the world throughout the seventh century: Italianpilgrims travelled comfortably to Jerusalem, Alexandrian papyrusstill stocked the chancery of the popes. But no Christian army everreturned to these eastern shores until the time of the Crusades

Above all, the Muslim generals came as conquerors, not as tribalraiders. The career of Muhammad, who had created a religiousempire in Arabia almost exclusively through negotiation, providedthe first califs with precedents for acute diplomacy. In the first decadesof their conquests, the Arabs gained quite as much by treaty as by thesword: key cities, such as Damascus and Alexandria, fell becausethe Muslim High Command was instantly prepared to offer generousterms - protection and toleration in return for a fixed tribute

More immediately, the foundation of the Islamic communityplaced an arnazing generation of young men - notably, the firstcalifs, Abu Bekr (ó32-34) and lJmar (61+-++) - in control of theBeduin world. This core of devoted 'true believers' provided theunder-Islamized Beduin raiding-parties with an unequalled HighCommand. The radicalism of the early Muslims extended to the ârtof war. The Muslim supporters of Muhammad had introduced thetechnology of fortification and siege warfare into the Hijaz. AfterMuhammad's death, the Muslim core of the Beduin armies facedthe Byzantines and Persians as equals in the art of armoured cavalry-warfare. They used the traditional mobility of the Beduin, basedon the camel; the camel carried a nucleus of fully equipped soldiers,

Thus, atjust the moment when (as we have seen) the Arab tribesalong the frontiers of Byzantium and Persia had to face the threat ofostracism, and consequent proleterizetion, the message ofMuhammadfilled up the chasm between the Arabs and their contemptuousneighbours, the civilized populations of the Fertile Crescent. Theethical teachings of Islam made the Arab Muslim the equal of the'God-fearing'Jew and Christian. The Qur'ai provided the illiterateArab tribesmen with the basis of a literary culture that imitated, andwould soon rival, the Bible of the Christian monk and the Torahofthe rabbi.

For the Arabs, Muhammad had brought peace, but for the rest ofthe Near East - a sword. Islam had suspended the traditional feudingamong the Beduin tribes, who were now nominally Muslims. Theirancient aggression had to find another outlet: internal feuds wererapidly transmuted into the standing feucl between the 'Umma andthç unbeliever. A year after the last Bedouin recalcitrants were coercedinto the Islamic confederacy, the Muslim generals declared the Holy-War against Byzantine Syria. 'It was not for love of Heaven thatyou fought there,' wrote a Beduin poet, 'but for love of bread anddates.' The conquest of the Byzantine and Persian empires was theprice which others had to pay for the success of the pax Islamicaamong the Arabs.

It was a stroke of genius on the part of Muhammad to tum thisessentially foreign messege into a principle on which the conflict-ridden society of the Hijaz could reorganize itself. He was called uponto cure the malaise of an 'emergent' society. úr the towns, the tribalstyle of life was losing its hold on the nouuedux riches of the merchant-dynasties. Private and public standards of behaviour were beingtom apart by new wealth, new opportunities and new ideas. Facedby this situation, Muhammad cut the knot of conflicting values.He reduced his followers to the loneliness of atoms in the face ofAllah: but in order to bind them together as a new 'people' - the'Umma. 'Within the fold of the 'Umma, the abrasive tensions of triballife were mercifully suspended. Under Muhammad as a religiousleader, peace came to towns whose magnificent style of life hadplainly outstripped the rough ethics of the desert, with murderousresults. As the inhabitants of Medina said : 'Allah has senc us a prophetwho will make peace between us.' It was as an arbitrator, backed by acore of devoted fighting-men, that Muhammad rose to power inArabia. Driven from Mecca, tn 6zz, by traditionalist opinion, heand his 'Umma imposed peece on the feuding parties of Medina.'When he returned in triumph to Mecca in ó3o, Muhammad set abouttransforming the commercial influence of the city, based on the trade

The immediate sources of the guiding ideas of Muhammad areeasy to see. 'Whatever he may have thought about the ChristianChurch, the Muslim guided his conduct by exactly the same consi-derations as did any Christian orJew throughout the Fertile Crescent.FIe, too, was a 'God-fearer'. He, too, had faced the terrible choiceof the Last Judgment, infallibly revealed to him in a Sacred Book.FIe, too, must think on it day and night. The Syrian hermit who'wept like a father mourning his dead child' ar the thought of the LastJudgment was venerated because he summed up an ideal of behaviourto which the populations of the Near East subscribed without ques-tion - even if the majority prudently avoided exposing themselvesto acting on it. Muhammad imposed this ideal on all his Arab fol-lowers. In so doing, he brought the Arabs into civilization as it wasknown in the seventh-century Near East.

In direct contrast to this communal ideal, the Muslim was en atom.Every tie of human society, so Muhammad believed, would vanishlike dust at the Last Judgment. Then, men would stand in awesomeloneliness, without fellow tribesmen, without protectors, even with-out relatives. In this life, the Muslim was to rule himself, not bymaintaining a brittle 'face' against the outside world of his fellowtribesmen, but by a personal, intimate 'fear', driven into his heart bythe thought of theJudgment of Allah. 'Shame' is no longer the bitterwound inflicted on a man by tribal opinion; it is the intimate anxietyof exposure at the Last Day. Even the Muslim taboo on wine wasconnected less with the wish to avoid drunkenness, than with ashrewd concern to remove a traditional aid to motivation. For it waswidely believed that, in his cups, the Arab gentleman could 'feelhis blood speak'. Through wine, he became mindful of the deeds ofhis ancestors; he felt able to live up to an ancient style of life - lavish

The Arab tribal ideal had been wholeheartedly extrovert. '\ manwas held rigidly to the obligations of his tribe. His conduct wasguided by the fear of incurring shame through lapses in publicbehaviour, by the desire to win praise from his fellows, by the needto uphold the nobility of his ancestry by spectacular deeds of genero-sity, of courage, by exacting swift revenge, and by tenacious obser-vance of a network of obligations. To follow this way of life was to'be a man'.

Muhammad cut the inhabitants of the Hijaz loose from the ties oftribal custom and threw them into the Fertile Crescent. His messagedeveloped as a protest against the Beduin way of life. Seldom hasa religion made so explicit the sanctions by which a man should rulehis life as did Islam; and seldom has it come into such immediate andlasting conflict with a fully articulated alternative rule of life, as didIslam with the tribal values of the Arab world.

The Lombard settlement did not , therefore , produce a completelyradical change in social struclUre. Doubtless many Roman landownerswere dispossessed through greed, as Paul says, but enough of them musthave survived to ensure The predominance of the Roman ideology ofproperty ownership in later cenTuries, as well as the Roman elements inroyal governmenl discussed in the preceding chapter. The assumedequality between Lombard and Roman law in Liutprand shows thatthere was no necessary difference of status between Lombards andRomans, even though we need not do ubt that masl Romans weredependenl peasan ts and a far higher proportion of Lornbards were not .The Lombards may even, in some places, have operated the Iwspitali/(ljsystem Ihat Paul refers to, though this is not something we can check.There were not enough of them, however, 10 destroy the socialhierarchies of Italy, and their swift fusion wilh the Romans must showthat they did not. When the Franks came, Lombards and Romans weremuch more similar to each other than either were to the new northerninvaders

The mixture between Roman and Lombard can be seen clearly inthe field of law . Roman law contin ued to exist. Lombard law seldommentioned it, but the kings were only legislating for their Lombardsubjects; Liutprand's references 10 it show that it survived on terms ofeq uality. The Lombard law ofpersons began in the eighth century to beinfluenced by Roman law, but only marginally (see pp. 43- 4 above).On the other hand, the few Romans who referred to themselves as suchin the eighth century had all adopted customs that only properly belong10 Lombard law, as with Felex of Treviso, who gave property to hisdaughter in 780, accepting back 'a handkerchief, as launigild, according10 Roman law '. lAunigild was a totaily Lombard concept meaning theexchange ofgifts, or the countergift that made a gift valid by Lombardlaw .1I This is not very surprising. Written Roman law for theinhabitants ofLombard Italy must have been fossilised after 568, for thekings only legislated for the Lombards, and Roman legislation wouldhave been inadequate 10 dcal with the radically new situation of theLombard state. The only solution was to borrow, and Lombard lawwould have come most readily to hand. Such borrowing must havebeen common. We cannot, in fact , distinguish between a Lombard anda Roman legal tradition in the actions of Italians at any social level inour sources.

This fu sion is still more apparent in our eighth-century texts.Naming, for example, shows a completely unsystematic mixture ofLombard and R oman forms. One of our very earliest charters, forforlonato, a Lucchese landowner in the 7105, a cleric with an obviousRoman name, also lists his live sons: 8enetato is a Roman name, but&nuald, Roduald, Raduald, and Baronte are o bvious Lombard ones.Other famili es show a similar mixture. In extreme cases, Lombard and

Our other evidence is usually later, but it all points in one direction,tbat ora rapid cultural fusion belween Lombard and Roman. This, aswe shall see, must have involved a complex social mixture, impossible ifthe Lombard occupation was as radical as is often thought. It used to besaid that the Lombards settled in free warrior groups away from theRoman population, and this seemed to be confirmed by archaeologicalfinds. 4 Most 'Lombard' archaeology consists ofthe excavation ofburials

The Romans in Lombard Italyvirtuall y disappear from history, so much so that it could be seriouslyargued in the nineteenth century that every one ofthem was reduced toslavery. Even in the eighth century, when our documentation begins, wefind scarcely a rererence to them: three or rour citations in the Lombardlaws, two or three in charters. We tend to refer to all inhabitants orLombard Italy as 'Lombards'; our evidence certainly allows us to. Butwe know that the great mass of the Italians must have been ethnicallyRoman. Assuming (on weak evidence) that there were far moreLombards than there had been Ostrogoths, say about 200,000, theLombards cannot have made up more than 5- 8 per cent of thepopulation of the parts of Italy they occupied, and the percentage mayha ve been less.

THE Ostrogoths vanished without a trace; Ilot so their successors. Bygoo, Italy was a complex ethnic mixture . The bulk oflhe populationwas Roman in origin, though Romans seldom appear as such in oursources. Alongside them we find the Lombards, at all levels of society,but particularly among the aristocracy; and also on the aristocraticlevel, newer immigrants- Franks and Alemans, and a few Burgundiansand Bavarians. In the parts of Italy the Lombards never conquered wefind newcomers too, this lime from the eastern Mediterranean.Byzantine haly, however, was the western fringe of an Empire that hadunbroken links with its Roman past, and in the seventh - cighth.-ccnlur):Exarchate it is seldom easy to tell indigenous Romans from immigrants.Only in th ~ t~nth - d~venth-century Byzantine provinces of the Southwere the Greeks ethnically distinct, as we shall see in a later chapter(pp. 156-f.)

His first act, in 7'3, was 10 revise the inheritance dispositions ofRothari 's code. The speed of lhis is significant: there must have beenconsiderable popular pressure in favour of the C?xtended inheritancerights of women and the legitimation of gifts to the Church . Liutprandsays as much in a nol;lia of 733, when he laments the greed of royalofficials, and wonders why they are not content with these concessions.The laws seem to ha ve been taken up at once; almost our first survivingChurch donations follow immediately on them, in 7t4.32 This concernfor the explicit legitimisation of acts by kings shows how great royalauthority and the force oflaw was, as we will see in chapter 5. II alsodemonstrates Roman influence, for these were changes in the directionof Roman legal practice. The Romans were the vast majority of theinhabitants of Lombard Italy, but they only begin to reappear in thesources of Liutprand 's reign - though in the seventh century a few canalready be seen, such as Peter son of Paul, Adaloald's aide, and theRoman noblewoman Theodora, whom Cunipert fell in love with at thebaths.33 Their actual relationship with the Lombards will be discussedlater (see pp. 66- 7'1), but it can already be seen that Liutprand wasbeginning to draw from thrir law . He did not take over elementsindiscriminately; Roman features fitted inside a firmly Lombard legalframework. The rest ofinherilance law, for example, remained totallyLombard. Again, in 731 Liutprand made the ordeal by bailie a much

Devotion to the patria of the civitas longsurvived the fall of the Western Empire. The average inhabitant ofMerovingian Gaul, as of late Roman Gaul, thought of himself firstand foremost as an inhabitant ofhis civitas; thus Merovingian saints inthe Vitae are introduced as Turonici, Cenomannici or Arverni (in-habitants ofthe civitates ofTours, Le Mans or Clermont), even thosewho are Frankish rather than Gallo-Roman by birth. This patriotismneed not have been based on any long-surviving tribai feelings, ofcourse. Some civitates had been artificially created out of larger civi-tates, as Angouleme was out ofSaintes, in the fourth century, or Car-cassonne out of Toulouse, in the sixth. A civis was no longer the in-habitant ofa territory; more and more he appears in the sources as theinhabitant ofa town, the civitas in its narrowest sense. Thus there is nodirect link, but surely there is a similarity, between the fierce loyalty ofthe Gauls of Caesar's day to their tribe, and the loyalty which led tothe inter-city rivalry and even warfare ofthe sixth century, and whichno doubt lay behind the growing independence of the civitates undertheir leaders in the seventhand early eighth centuries.

Throughout our period the principle of'personality' was dominantin legal thinking - 'personality' as opposed to 'territoriality'. Theprinciple is summed up in Lex Ribuaria 3, 3 and 4: 'And thus wedecree that ifFranks, Burgundians, Alamans or men from any othernation are staying in Ripuarian territory and are called to law, theyshall be judged according to the law of the pIace where they wereborn. And ifthey are condemned, they shall be punished according totheir own law, not according to Ripuarian law,.29 Thus the professioiuris, the dec1aration by which law one lived, became apart of legalprocedure. Theoretically descendants ofGallo-Romans (and c1ericstoo, according to Lex Ribuaria) would live by Roman law, whiledescendants of Salian Franks, Ripuarian Franks, and, presumably,Bretons and Basques as weIl, would live by their own law. In ninth-century Septimania there is evidence of Gothic, Roman and Salianlaw all being used in the courts.

Five hundred years after the disappearance of the Western Empire,there were thus in Gaul a number of distinct peoples, owing theirsense of community more to a combination of cuItural and politicaltraditions and misunderstood history than to any real ethnic distinc-tions. All ofthese peoples, at one time or another, had their own lead-ers, called duces, principes, or, in special cases, reges, and aIthough theirfeelings of community did not determine political developments (andindeed could be.formed and re-formed by them), such feelings form anever-present background to the political changes of the time. Nodoubt, as we shall see in the following chapters, there were other,smaller, communities which could form stronger 10yaIties. But in onearea oflife at least the inhabitant ofGaul might weIl have to take his'nationality' into account: when he went to law.

The extent ofScandinavian settlement in Normandy is as controv-ersial a problem as that oftheir contemporary settlement in England.There is almost no archaeological evidence. Place-names of Scan-dinavian origin are found in some parts ofNormandy, although thebulk of place-names are of Gallo-Roman or Frankish type. TheNorman aristocracy seems poised between two worlds. It was prob-ably largely Scandinavian in origin, and was still apparently capableof enjoying the visit ofthe skald Sigvatr Thorharson, who entertainedRichard II's court in 1025. And yet, three-quarters of a century ear-lier, Richard I had had to go to Bayeux to learn a Scandinavian lan-guage, for no one in Rouen could teach hirn. The Normans preservedtheir links with Scandinavia, allowing Vikings to use their ports as abase for attacks on England (until a peace cemented by the marriageofRichard II's daughter Emma to Ethelred the Unready) and hiringViking warriors for their own expeditions, and yet, in the eleventhcentury, often referred to themselves as 'Franci'. The duality emergesin habits ofname-giving: a number ofNorman aristocrats are knownwho had two names, one Norman and one Frankish, such asThorsteinn and Richard, or Stigand and Odo. It is perhaps signifi-cant that, on the evidence ofnames at least, there were few women ofScandinavian origin in Normandy. It seems likely that the settlementof Vikings was fairly small-scale; that the aristocracy was rapidlyFrankicised; and that the preservation oflinks with their Viking pastwas part romantic nostalgia, part aristocratic fashion, and part com-mercial and military convenience.

'Francia' can be a political concept, the area controlled byFrankish kings; but in the ninth century we also find it used ofnorth-ern Gaul or of a restricted portion of it. 'Francus' seems to changefrom an ethnic to a territorial sense in the course of the seventhcentury. By the eighth century all those living in northern Gaul(except the Bretons and Alamans) are Franks: we even find somedescendants of Roman aristocratic families in Trier in the eighthcentury, bearing clearly Roman names, being called 'senators oftheFranks,.19 In a marginal note in a ninth-century manuscript of theLiber Historiae Franeorum the historical process was explained: 'Clovisexterminated all the Romans who then lived in Gaul, so that scarcelyone could be found. And the Franks at this time are seen to have learntthe Roman language, which they still use, from those Romans wholived there. What their nationallanguage was before this is unknownin these parts.,20 The national language was, of course, Frankish(West Germanic, unlike the East Germanic Gothic), and in fact seemsto have been spoken by some people even in areas now weIl within theRomance (French) areas of northern Gaul.

But old triballoyalties within the Frankish people may have survivedthe elimination of their leaders. The Frankish tribes who advancedsouth down the Rhine quite independently ofChilderic's Franks, andwho led the advance against the Alamans, arguably preserved theirsense of identity after the death of Sigibert the Lame. In the seventhcentury they were given their own law-code, the Lex Ribuaria, nor-mally a sign of ethnic identity. The rapid advance of Childeric andClovis into Roman Gaul mayaiso have helped to bring about divi-sions within the Franks. There was to be a long rivalry between Neus-tria in the west, where the Franks were only a dominant minority, andAustrasia in the east, which was much more Germanic in character.The fairly narrow strip of territory west of the Rhine which was theheartland of Austrasia, and from which the Carolingian dynasty wasto emerge, is now German-speaking. Apart from Alamannic Switzer-land and Alsace it was the only part ofGaul to submit to such heavy bar-barian colonisation that the Latin language was replaced by a Germanic

Nor was he the only Frank of whom this was true. The lastknown Roman official in the former imperial capital ofTrier, for in-stance, was a Frank: Count Arbogast. He was a descendant, perhaps,of the Arbogast who, as Master of the Soldiers, had assisted in thepagan rebellion of394 against Theodosius I. Sidonius' letter to CountArbogast leaves us in no doubt that he had embraced the culture andideals ofthe Roman aristocracy.The Roman tongue is long banished from Belgium and the Rhine;but if its splendour has anywhere survived, it is surely with you;our jurisdiction is fallen into decay along the frontier, but while youlive and preserve your eloquence, the Latin language stands un-shaken. As I return your greeting, my heart is glad within me thatour vanishing culture has left such traces with you; continue yourassiduous studies, and you will feel more surely every day that theman of education is as much above the bOOT as the boor in his turnis above the beast. 13The elose relationship between German and Roman on which theFranks were to build their power was already weIl established beforethe accession ofChilderic's son Clovis.

There is one more source for the history of Childeric's reign: hisgrave. Discovered in 1653, it remains probably the most importantarchaeological find of the migration period. Buried across the riverfrom the Roman town ofTournai, the body was identified by a seal-ring inscribed CHILDERICI REGIS: the words are Latin, and the wholeconcept of a seal-ring Roman. In death Childeric wore a goldcruciform brooch with three onion-shaped terminals, precisely thetype of cloak-fastening presented to Roman officials at their investi-ture by the Emperor, and worn as a badge ofoffice. In the grave wereover a hundred coins minted in the Eastern Empire, perhaps part ofatribute paid in return for military assistance. There was little Fran-kish about Childeric. His sword, buckle and other personal adorn-ments were decorated in the gold-and-cloisonne style common to allGermanic royal and aristocratic circles in the fifth century. But hisgrave did contain a rock-crystal ball, an amulet normally found in thegraves of Frankish women, and another talisman in the shape of a

In 879 Lower Burgundy, including Provence, became akingdom under Boso: in 888 Upper Burgundy, including large partsofpresent-day Switzerland, became a kingdom under Rudolfl. In themid-tenth century Upper and Lower Burgundy were united underRudolfs successors. The great variety of titles used by these kings - rexBurgundiae and rex Burgundionum alongside rex Alamannorum vel Provin-ciarum, rexJurensis (oftheJura) and rex Viennensis (ofVienne) -suggestthat they did not rest their power on any particular feeling of com-munity among the Burgundians. In the tenth century there were somewho reckoned themselves as Burgundians, others (descendantsof the great families of the late eighth century) as Franks, and yetothers as Romans. A number of aristocratic families in Provence stillboasted oftheir Roman or Gallic descent, and still gave almost exclus-ively Roman names to their children, though in Proven~al forms, suchas Pons, Honorat, Maieul and Amelius. The feeling ofbelonging to aseparate Burgundian peopie was probably strongest in the north ofthe area, in the Duchy of Burgundy (the area barely settled by fifth-century Burgundians at all) rather than in the much more diffuse anddisorganised Burgundian kingdom. The biographer of Bishop Bettoof Auxerre (d. 918) speaks of 'our nation of Burgundy', and whenDuke Ralph of Burgundy takes the French throne in 923, the FrankFlodoard ofRheims knows that the 'kingdom ofthe Franks has beentransferred to a foreigner (ad extraneum)'.

By the eighth century, therefore, Romans and Germans in one areaof Gaul had come together as a new people. The descendants of theRomans had forgotten their inheritance; they accepted this false his-tory, and they lived according to Burgundian law. Their sense ofcom-munity was probably strengthened by anti-Frankish feelings, al-though this is a quite unmeasurable factor.

The fate ofBurgundy differed in one important respect from that ofthe Visigothic kingdom in Gaul: the Visigoths had been able to flee toSpain, taking with them their independence and their ethnic tradi-tions, and leaving behind an Aquitaine with its Roman traditions vir-tually intact. The Burgundians had no such refuge; they and theirethnic identity continued, indeed flourished, within the Frankishkingdom. The regional name 'Burgundia', first used in a letter fromTheodoric of haly in 507, survived, coming to be used, by the earlyseventh century, of the south-east portion of the Merovingian king-dom. The greatest of the Merovingian kings to have his power-basefirmly in south-east Gaul was Guntram, one ofthe four grandsons ofClovis who partitioned Gaul among themseIves in 561. He clearly feitthe need to have the backing of the ethnic Burgundians in his king-dom; he named his eIdest son Gundobad, the first (and last) occur-rence ofthat name within the Frankish royal family. In Guntram'sday writers such as Gregory ofTours still distinguished between theGermanic and Gallo-Roman sections ofthe population in south-eastGaul; by the end of the seventh century 'Burgundiones' could mean'all the inhabitants ofsouth-east Gaul', and this became general usagein the eighth century. The author ofthe eighth-century text ceIebrat-ing King Sigismund's martyrdom at the hands of the Franksexplained that in the fifth century the Burgundians had killed allRomans who had not fled from the area; 10 a similar legend, as we shallsee, served to explain how all the inhabitants of northern Gaul hadbecome Franks.

When the Arabs invaded Spainin 711, the last Visigothic kings retreated northwards to Septimania;after Septimania, too, was conquered, it served as a base for Arabraids on Gaul for thirty years, before becoming the first part of ArabEurope to be 'liberated' by Christendom. From then on Gothia waspart ofthe Frankish kingdom, or, as it was often expressed, it was oneofthe regna or kingdoms ruled by Frankish kings, and its inhabitants,Goti, were one of the eight peoples of Gaul. The Goti remained dis-tinct from the Hispani, Spanish refugees who flocked into Gothia.Here we meet for the first time the phenomenon of a people taking onthe ethnic identity of its ruling elite. In the sixth or seventh centuryVisigoths themselves had probably formed only a smalI, but power-

By the last century ofthe Empire all inhabitants ofGaulwere Roman citizens, and many by then spoke Latin rather than aCeltic dialect; to them the old distinctions must have been overlaid bythe general designation of'Romans'. But there is a hint in a fifth-cen-tury dialogue written in Gaul that memories could be tenacious. Oneofthe participants, called Gallus, claimed that his Belgic friends scof-fed at the gluttony of the Gauls. He apologised for his own speech:'When I remember that I am a Gaul and am going to hold forth infront of Aquitanians, I am much afraid that my rather rustic speechmay offend your over-civilised ears. >I Perhaps the Latin of norther-ners sounded uncouth because, as linguists tell us, it was alreadyexhibiting features which were to result in the emergence of a Ro-mance language distinct from that spoken in the south: Langue d'Oil(French) as opposed to Langue d'Oc (Occitan). The linguistic fron-tier in 500 is thought to have run due east from the mouth ofthe Loire,roughly following the Roman administrative division for severalhundred miles. But the uncouthness ofthe Gaul's speech and his needto apologise for it mayaiso have reflected a cultural distinction. Thesouth was more heavily Romanised and long after 500 remained thehorne of a more cosmopolitan and consciously Roman population. Inthe sixth century a northern cleric begged the Frankish king not to ap-point hirn to a southern bishopric, 'not to submit hirn, a simple man,to the boredom ofhaving to listen to sophisticated arguments by oldsenatorial families, or to counts who spend all their time discussingphilosophical problems'. I For Gallus 'Aquitanian' almost certainlymeant an inhabitant of south-west Gaul, south of the Loire; it hadnothing to do with the Iberian Aquitanians who lived south of theGaronne,but derived from the late Roman provincial name. TheRoman structure ofGaul, together with the climatic, cultural and his-torie factors which have helped to make southerners and northernersin France seem foreign to each other right down to modern times, was

By the second half of the sixth century, KingLiuvigild of the Visigoths lifted the old Roman ban on marriagesbetween Romans and barbarians (or Goths in his law): this is the firstextant Gothic legal reference to Gotae.126 The original Roman lawseems to have been enacted in the specific circumstances of Firmus’rebellion in North Africa in the 370s127 and it may well have beenthat there was no specific need to repeal it until women began tohold non-Roman ethnic identities on a significant scale.

The ways in which changes in the precise political significance ofmarriage entailed changes in female identity might also be exploredby looking at the gendering of ethnicity. It was mentioned abovethat early Frankish law viewed Franks and Romans as adult males.Women and children are not, in this code, distinguished by ethni-city. This implies that ethnicity was achieved and that it was therebyperformed, and also that an early Merovingian Frank or Roman was

Crucially, fourth-century developments had produced a new formof masculinity, the military model discussed above, which providedthe route from Roman to non-Roman ethnicity. During the periodthat concerns this book, as far as gender is concerned, one of the mostimportant developments was a shift from the civic to the militarymodel as the hegemonic form of masculinity.115 By the later sixthcentury, it was weapon-bearing, and especially an association withthe army (and thus an attendant ethnicity), that defined the dominantmale ideal. Other forms of male identity were, increasingly, comingto be seen as inferior. In northern Gaul, for example, Salic law rated aRoman dining companion of the king as worth half the ‘wergild’(the fine payable by someone who killed or injured him) of a (byimplication Frankish) member of the royal bodyguard (trustis).116Roman freemen had half the ‘wergild’ of Frankish freemen, andwithin two or three generations of the end of this period, in thesame region, Roman men were the legal dependants of Franks.117The greater relative opportunities for marriage and reproduction athigh political levels constituted another reason why non-Romanidentities were increasingly attractive.

The use of systems of measuring time was not restricted to thesetwo Gallic cities. It has been argued that consular dating was espe-cially prominent in the Burgundian kingdom, though this argumenthas its problems.106 The ‘Spanish era’ (a chronology beginning in theyear we think of as 38 bc) became important and has been plausiblylinked to Catholic, Hispano-Roman identity.107 The Mauretanianera, commencing in 39 ad, continued in use in North Africaninscriptions and might well have had an analogous political usage, inopposition to the Vandal era, which began with their conquest ofCarthage exactly 400 years later.

In Gaul, too, the city was important. This was especially so inAquitaine, where the cities survived better as higher order settle-ments. The civitas unit seems to have been fragmenting in the north,though this change was perhaps not effective until the seventhcentury.101 Nevertheless Trier seems to have retained a particularsense of its Roman identity, especially as far as its local senatorialaristocracy was concerned.102 In the south, the civitas was a par-ticularly important socio-political unit. Gregory of Tours habituallydescribes the people of the south as the men of particular civitates. InAquitaine armies were raised from these city-districts, giving theirinhabitants (or populus) a military basis to their identity.103 Gregoryalso tends to identify Gallo-Roman aristocrats by the name oftheir father and their civitas of origin.104 In two cities we can see a

ery clearly a continuation of the late imperial situation, this tooshould have been a factor easing the gradual promotion of non-Roman political and ethnic identity. The way in which it perpetu-ated the later Roman state of affairs, and was not yet a matter of birth,might be illustrated by study of the earliest Frankish law code, Saliclaw. In this text the ethnic terms francus and romanus are only appliedto free adult males, seemingly with legal dependants.81 When thiscode was issued, Frankish (or Roman) identity, seemingly linked tomilitary service, appears to have been restricted to adult males andtherefore, at least as far as Frankish identity was concerned, probablyachieved through the performance of military service and theacceptance of a right to take part in the army’s activities (effectivelythe political assembly). Younger warriors (pueri) do not appear tohave achieved a full ethnic identity. An individual’s life cycle wasimportant in the adoption of new identities.

Second, the multi-layered nature of ethnicity enables us to counteran argument deployed by Peter Heather against the idea thatethnicity was easily changed. Heather has drawn attention to‘reappearing tribes’ in late antiquity and the way in which, duringthe sixth-century Gothic wars, Roman troops who deserted to theGoths re-emerged to rejoin the Romans at a later point. This, saysHeather, demonstrates that people were not fully integrated into theGoths (or Huns, or other groups).

Nevertheless, given the dominance of ideas of correct Romanbehaviour in the late Roman period, the replacement of ‘Roman-ness’ should have been a traumatic process that called into questionall sorts of other aspects of an individual’s identity. There were,however, resources within late Roman society and politics thatenabled people to navigate these important changes. Ethnicity isdynamic, with constant reordering and renegotiation of the layersof identity according to the situation or setting.67 What it meant tobe Roman, and the relative importance of its different levels – tribalor municipal, regional, provincial and so on – had changedthroughout the imperial period. Particularly important was theemergence of new military and Christian forms of Roman self-definition during the fourth century. The renegotiation of ethnicitywould not in itself be something new to the fifth century.

In the third century theimperial state had not disintegrated because, although the mechan-isms that had bound localities into the Empire in the first andsecond centuries had generally ceased to be very effective,60 noalternatives emerged as legitimate bases for local authority beforethe state revived the attractiveness of participation in the Empire.61The very centrality of romanitas to ideas of power, status and genderwas one of the Roman state’s most fundamental sources of strength

Religion could be another unifying factor. The Goths, Sueves,Burgundians and Vandals adhered to Arianism, seen as a marker ofnon-Roman military identity. The Burgundians, Vandals and Suevesare sometimes recorded as Catholics in the earlier part of their his-tories within the Empire, and the Burgundians seem to haveremained divided between Nicaean and Arian Christianity.57 Thechoice of the, by the fifth century, heretical Arian doctrine musttherefore have, again, been more complex than a simple matter of

Other aspects of the archaeological record might have related tothe creation of ethnic identity. Burial rites, for example, might havebeen among the customs to which an individual subscribed to showhis or her group-membership. In Britain the cremation ritualimported from northern Germany might have functioned in thisway.49 Not everyone who cremated their relatives, or who asked tobe cremated, need have been an Anglo-Saxon immigrant, even ifthe ritual itself was undoubtedly a new introduction. As will havebecome abundantly clear, furnished burial cannot be assumed tomark barbarian immigrants into the Empire. It is always much morecommon within the former imperial territories and seems, whetherone is looking at northern Gaul, lowland Britain or the provincesnorth of the Alps, to be associated with the collapse of the oldRoman landholding relationships. Nonetheless, one thing we can-not know from the archaeological record is what was said over thegrave (where furnished burial is common we do not, at this date,have funerary liturgies). Such spoken elements might, especiallywith time, have included invocations of ethnic identity, as with thelaw. This is plausible given that some of the material, such asweaponry, probably relates to the ethnic identities constructedaround the right to fight in the army.