Dividing al-Andalus in space and time

The beginning of Muslim rule in Iberia has a clear start date, that of the invasion of the Visigothic kingdom by the forces of Tarīq ibn Ziyād and his superior Mūsā ibn Nusayr in 711. The history of al-Andalus thereafter is conveniently broken up by major episodes:

• a period of conquest, lasting until 718 in the north-east of the Peninsula and arguably ending still later (Collins 1989);
• a period of rule by appointed governors, punctuated by a revolt among troops recruited from Africa in 741 and terminated by the ‘Abbasid coup in the Middle East, the flight of the last Umayyad scion to al-Andalus and his takeover there in 756 (Marín 1998);
• the fluctuating fortunes of the new Umayyad emirate until 929 (Kennedy 1998, 30–81);
• and lastly the glory period of the Andalusi caliphate, already running into difficulties by the year 1000 and by 1010 in an irreversible decline marked in 1031 by the final extinction of the Umayyad ruling line (Kennedy 1998, 82–108; Scales 1994).

The only part of this that receives much textbook attention is the Caliphate (e.g. Moran and Gerberding 2004, 206–208), because of the splendour of its material remains and its self-proclaimed political might, and also because its narrative is dramatic and can be told relatively clearly from our Cordoban sources. Its rise and fall is, nonetheless, only a third of our chosen period. That rise also ended a counterbalancing period of ascendancy and expansion among the Christian principalities of the north, its sometime opponents.

Among opponents of the time, however, we should count not just Christians but also Muslims. Key figures of that period (756–929) included a convert family of alleged Gothic descent known to us as the Banū Qāsī, from their claimed forebear, one Count Cassius (Fierro 2009). In the mid-ninth century this family could count Huesca, Lleida, Tudela, Zaragoza and several other cities of the north-east of the Andalusi state as theirs. They fought, allied with and married among Christian princes, as the same Oviedo chronicle of above tells us; another states that their greatest patriarch, Mūsā ibn Mūsā, called himself “the third king of Hispania” (Gil Fernández et al. 1985, Alfonso III c. XXV). Certainly at times the Banū Qāsī were loyal servants of the emir, while at others they were effectively uncontrolled, and it was possible for both things to be true as long as no conflict opposed them to the emir. Removing them from this position was only achieved by the slow promotion of a rival lineage to a roughly similar position (Lorenzo Jiménez 2010).

Lords closer to Córdoba could be just as difficult. One of the achievements leading Emir ‘Abd al-Rahmān III towards claiming the Caliphate was the conquest in 928 of the citadel of Bobastro, centre of a rebellion by an alleged apostate to Christianity, ‘Umar ibn Hafsūn. The conquest was a notable achievement mostly because this rebellion, 150 km south of Córdoba, had by then been running for forty-six years, during which time Ibn Hafsūn had not just been occasionally recognised as an independent whose army fought alongside the emir’s, but had died and been succeeded in his disobedience by his sons (Acién Almansa 1994). Even more surprisingly, it has been calculated that Toledo, the old Visigothic capital and home to a large Christian population but also not far from Córdoba, cumulatively spent a full half of the duration of Umayyad rule in rebellion (Manzano Moreno 1991, 261–310). Was this a territory of al-Andalus in any meaningful way? And if the Emirate’s supposed frontier was frequently hardly 300 km away from its capital, how on earth to categorise the areas that lay beyond it but remained attached to the Umayyads (cf. Manzano Moreno 1999, 38–40)? Certainly, defining a single polity with a clear edge must be incorrect for this area however it be done. This situation of al-Andalus in the period of 756 to 929, in which it was occasionally effective outside its territories and always a source of wealth but rarely integrated enough to be threatening to those beyond its control, was an opportunity for those who wished to build rival power bases outside.

The northern redoubts

Few of those who thus set up can have aimed to become “kings of Hispania,” but it is, naturally enough, those zones where kingdoms or states emerged whose identities persist today that have dominated scholarship, and any history of Spain (though perhaps not of Portugal) will necessarily focus especially on the northern coastline where Christian rule persisted after 718 (Sánchez-Albornoz 1991; Riu i Riu 1999; cf. Linehan 1993, 95–127). While the autonomy and distinctiveness of some of these areas is still a live political issue now, with consequent effects in historical discourse, over the eighth and ninth centuries the sources certainly justify discussion of independent princedoms in at least Galicia, Asturias, Cantabria, Aragon and Navarre, and the tenth century added Castile and at least ten different counties in the modern Catalonia, ruled by a kaleidoscopic succession of counts, several of whose territories are now in modern France. These powers did not exist both autonomously and simultaneously, however. Although sources from their court are naturally keen to assert this, it seems clear that kings in Asturias ruled Cantabria by the end of the ninth century and could usually aspire also to rule Galicia (Escalona Monge 2004). By the early tenth century, however, this nascent kingdom had relocated its capital to León, and though that proved vulnerable, after 924 Asturias never again had its own king (Collins 2012, 52–82). Castile emerged as a county of this new kingdom of León, and only took its “definitive” form by a coincidence of inheritances under Count Fernán González in 931; its autonomy between then and the succession of Count Ferdinand I to the throne of León in 1037 is a matter of highly politicised debate (Mínguez Fernández 2004, 183–199).

The area of modern Catalonia presents further complications for a nationally organised perspective. The territory owes its distinctive political constitution and language to the fact of its having been ruled intermittently by powers across the Pyrenees, before our period as part of the Visigothic kingdom of Toulouse and thereafter as part of a group of counties assembled as a defensive marca hispanica by the Carolingian kings of the Franks from 785 onwards (Chandler 2013). Modern Catalonia cannot be identified with the marca, however, as that grouping also included Navarre and Aragon, both of which seceded in the 820s. ² The rump remaining was then diminished by a further coup in 826 whose losses were only recouped half a century later. In what remained, Frankish royal appointments to lay office were made as late as 878 and the groupings of counties thus created proved durable, but this still left four families in charge of them, and while that based in Barcelona was the most powerful, they did not control the others until after our period. Resort to the Frankish king for legal privileges continued until at least 987, when the last Carolingian, Louis V, died (Jarrett 2011). This date has since become inarguable in the local scholarship as the beginning of a “Catalan” independence that no one at the time would have recognised. ³ The only things that then unified all the Catalan counties were their Frankish history and the fact that they were not part of any other polity.

The danger of a teleological perspective on this period will by now be evident. The history of medieval Spain has often been told as the survival and revival of Christian kingship in Asturias, its move to León and takeover by a nascent Castile, with a subplot in which Catalonia is removed from this “national” story and begins the pursuit of its own, joining with Aragon in 1137 to allow the eventual completion of “Spain” in 1492 (e.g. Castro 1948; Sánchez-Albornoz 1956; Barbero and Vigil 1974, otherwise sharply differentiated). This would have been completely unforeseeable in 1031. For one thing, future events in al-Andalus would have been deeply obscure; could Umayyad rule really have been assumed finished after so long? Even if so, the Almoravid and Almohad caliphates would have seemed fantastic. More importantly, however, a Castilian-Aragonese narrative would have been implausible because the supreme kingdom in the Iberian Peninsula in 1031 was actually Navarre. Its ruler, Sancho III the Great, self-proclaimed King of the Spains (rex hispaniarum), numbered the Counts of Barcelona, Pallars and indeed Castile among his vassals and ruled León directly (Mínguez Fernández 2004, 123–132). This would not endure, of course: Navarre had no access of its own to expansion into al-Andalus and its later bisection by France has removed it from the main narrative. The point is, however, that this period’s importance is not what it led to: its developments frequently “led” nowhere, but they were as life-changing as any later for those involved. The period is in fact a frontier zone itself, between destruction and restoration as one recent work has it (Deswartes 2003), and, like the political frontier we have delineated, it requires examination on its own terms.

The space between and its definitions

The polities on either side of this zone conceptualised it differently. Muslim thinking envisaged a core dār al-Islam, domain of Islam, outside which lay the dār al-harb, domain of war, although such areas might be placed under tribute without conquest, constituting a dār al-‘ahd, domain of the pact. Between the dār al-Islam and these outer zones lay the thughūr (sing. thaghr). This Arabic word is usually translated as “frontier,” but these frontiers were not barriers but passages, from continuous Muslim rule to uncontrolled spaces (Manzano 1999, 38–42, 45–46). The breach of such a pact would convert al-‘ahd to al-harb, however, and turn the conceptual frontier inside out. This formulation does not fit easily into such categories as “linear,” “zonal,” “enclosing” or “expanding” frontiers (Berend 1999) but it did and does provide a way to think of most of the Peninsula as being within a Muslim polity despite that polity’s faltering central control.

Between the thughūr and territory inarguably Christian was still a considerable space. The strongest idea that Christian documents dealing with it expressed was vacancy. For the Count of Barcelona in 973, the old city of Isona was in “the extreme ultimate ends of the Marches,” and consequently it was unsurprising that it had been “anciently destroyed by the pagans,” although he was garrisoning it anyway and that garrison was apparently being supported by local agriculturalists (Udina Martorell 1951, no. 174). The Barcelona monastery of Sant Cugat del Vallès, when expansion into such areas began to consolidate, was keen to point out that royal concessions of a previous century had allotted them extensive territory in these “places of solitude” (Rius Serra 1946, no. 452). Obviously if the places had not been empty the monastery’s claim would have been disputable, making all such claims of desertion suspect (Jarrett 2010a). In Asturias this rhetoric was taken to a grander level in the so-called Chronicle of Alfonso III, which claimed that most of the lands south of the kingdom were not only long deserted, but so by royal fiat, King Alfonso I having moved their occupants north (Gil Fernández et al. 1985, Alfonso III cc. 13–14). The use of such a claim for Alfonso III, striving to bring those areas back under obedience, is evident, and cannot be taken uncritically (Mínguez Fernández 2002).

This language, nevertheless, was often taken literally in the historiography, because of the Arabic sources’ similar distancing of their opponents but also because the idea of an empty space into which bold pioneers might rightfully move, but where they might need military protection by the burgeoning state, was mirrored in the fashionable work of Frederick Jackson Turner on the frontier in the American West. ⁴ Here as there was an open space, inhabited by no one that mattered to the writers, the occupation of which was the common experience that bound the Iberian kingdoms (except Navarre) together and explained their eventual unification. The power of such a narrative could overwhelm most scruples about exactly how deserted a full third of the Peninsula could have been (Linehan 1993, 95–127).

Inhabiting the frontier

Over the past thirty years, however, historians and archaeologists have become increasingly sure that the frontier zone was not empty. In texts, the hints are evanescent: the evidence is largely transactional documentation that was by definition only generated once a land market existed, unlikely in an autarkic community. As a result, we see areas only as they pass into the normalising structure of government. Nonetheless, differences detectable between groups can expose their deviations from such normalisation (Jarrett 2010b) and, in the later part of the period under discussion, these differences were enshrined in the local law codes known as fueros, or in Catalonia cartes de població (Manzano Moreno 1994, 94–95). Less programmatic but equally telling hints can be got from place- or personal names displaying antiquity or linguistic mixing between Arabic and Romance languages, although the latter phenomenon is highly regionalised and imperfectly understood (Aguilar Sebastián and Rodríguez Mediano 1994; Barrios García 1982; Hitchcock 2008, 53–74). Archaeology’s contribution has been more recent, and struggles with the difficulty of proving continuity as well as inadequate publication of many excavations, but enough has been done to show that settlement did not obviously cease in the frontier areas on the Muslim conquest, and that maintenance of churches continued, suggesting the continuing articulation of communities (Gibert Rebull 2006; González García 1995). With this and work on the survival of land boundaries, on the re-use of ancient sites, and on the endurance of local saints’ cults (e.g. respectively Bolòs i Masclans 1995; Jarrett 2010b, 99–103; Linage Conde 1971), some generalisations about the circumstances of these almost-invisible populations are possible (cf. Martín Viso 2005).

The most obvious generalisation must be variation, of course. We find settlements both nucleated and dispersed, which has been put down to the origins of such communities but which may in fact have had more basic environmental causes (Mínguez Fernández 1998). We cannot say much about these peoples’ origins: while arguments have been made for very deep local continuity from even pre-Roman times in some areas (Barbero and Vigil 1974), population movement cannot be ruled out either, even if sometimes it may have been very short-distance (Jarrett 2010a, 339–341). A charter of 943 recording the occupation of a particular area of the kingdom of León in settlements which included a villa de cordobeses and a villa de asturianos prevents us from doubting that people were in movement, northwards and southwards, and the Cordobans, at least, could hardly have been relocated by Leonese royal command (Mínguez Fernández 1976, no. 87).

As regards matters of religion, refusal to guess is our only safe position: Christian sources talking of christiani perversi in the frontier zones may not have been making doctrinal judgements, but such persons were beyond diocesan structures and Church correction (Jarrett 2010b, 98–99). Similar problems arise with language: later Arabic sources tell us of occasional autochthonous Arabic-speaking populations far beyond Andalusi control (Manzano Moreno 1994, 92, 95), but in certain areas people with Arabic names were frequent occurrences (see Aguilar Sebastián and Rodríguez Mediano 1994; Barrios García 1982; Fernández 2009; Hitchcock 2008, 53–74), the most startling of whom is the scribe of a charter for King Ordoño II of León, a deacon called Muhammad (Sáez 1987, no. 40). On the other hand, Latin documents continued to be issued deep into al-Andalus, although almost none survive (Barrett 2011, citing a charter from Lorvão).

One well-known case in a document from Ribagorza in Aragon brings out the ambiguity of loyalties that we must imagine among the inhabitants of these zones. In it, the villagers of what is now Aguinaliu recorded having settled a boundary dispute with their neighbours in Juseu. These two communities lie about sixty kilometers from the then-new bishopric at Roda d’Isávena, yet rather than resort to either bishop or count of Ribagorza for arbitration they sent much farther afield to Muslim-held Lleida, to a priest called Fortún, “judge of all the Christians in Lleida” (Davies and Fouracre 1986, 255). This was perhaps again the pact at work, but it may also tell us that the villagers wanted to avoid the obligations, like tithe or censum, that might be enforced upon them by engagement with the more local powers-that-were. Arguably, the kingdoms of the north expanded by interesting such populations in taking part in the wider polity (Carvajal Castro 2012); if so, in these cases they failed.

Crossing the frontier

Despite such disconnection, there was travel through these areas. Manuscripts made in the south, annotated in the north in hands we can date, silently demonstrate such traffic (Díaz y Díaz 1969). A small number of ecclesiastical travel narratives show us that different standards of travel existed: when relic-seekers from St-Germain-des-Prés went from Paris to Córdoba, their crossing of the frontier was a complex series of escorted transits between jurisdictions (Christys 1998), albeit largely for political reasons (Nelson 2010, 19–20); the unimportant Cordoban cleric Eulogius had no such difficulty while hunting manuscripts in Navarre (Christys 2002b, 55–62). It was not just churchmen who travelled, either. One assumes that a string of Christian emissaries who visited the court of Caliph ‘Abd al-Rahmān III in the 940s (Christys 2002b, 109–113), or his Jewish doctor Hasdai ibn Shaprut, sent as diplomat to Catalonia in 940 and León in 941–942, travelled in company and under watchful eyes. There were also fugitives to Córdoba such as the frontier warlord ‘Aysūn ibn al-‘Arabī, who supposedly escaped from a Frankish prison by guile (Manzano Moreno 1991, 218–222), or the Frankish convert to Judaism, Bodo-Eleazar (Riess 2005).

Movement within the frontier zone can also be seen. A well-studied case is that of ‘Abd al-Rahmān ibn Marwān Ibn al-Yillīqī, whose very name suggests a family origin in the dar al-harb of Jillīqīya, a locution which covered not just modern-day Galicia but also León and Castile. It was in Mérida that he rebelled in 876–877, however, whence he quickly fled north to King Alfonso III of Asturias, governing a frontier castle for him for eight years before returning south to seek reconciliation with the emir, who placed him in charge at Badajoz (Christys 2002a). Although this man had royal lords, it seems that his real loyalties were to a homeland between them. Other such defectors also never quite lost touch with the other ‘side’ (Collins 1995, 281; Manzano Moreno 1991, 190).

There must also have been trade, but we can hardly see it. In 1018 we have in Barcelona the first documented appearance of a merchant there; he was dealing in cloth, and Andalusi cloths are documented in many a Catalan will, though these may have come by sea rather than across the frontier proper (Wolff 1963). The same might be true of the Caliphal gold mancuses that start appearing as coin of payment in the Catalan counties in the 970s, but it cannot so easily be the case with their occurrences farther west (Balaguer 1988). Coins are not by themselves good evidence of trade, however, especially in a context where tributes were frequently paid (Grierson 1959) and since the historiography is relatively agreed that there was substantial migration, little that is conclusive can be demonstrated yet.

Mozarabism and arabicization

In more traditional historiography, that migration is assumed to have been northwards: the ascendant Christian kingdoms should naturally have called to the Christians of al-Andalus labouring under the Saracen yoke. This is not without support in the evidence. For one thing, the most visible category in the Arabic-named Christian population already mentioned was churchmen, a professional class who could expect employment in the north and might well have felt an increasing pinch in the south as al-Andalus began to tax the Church in its territories more heavily (clear in Samson 1969–1973, II Preface c. 2). ⁵ The voyages of manuscripts northwards already mentioned tend to confirm such a movement, and this has been used by believers in the magnetism of the new kingdoms to explain those areas’ adoption of Andalusi styles of building, decoration and ceramics; all this was supposedly brought by Mozarabs (classically Gómez Moreno 1919; see also Werckmeister 1993).

This term, Mozarabs, is full of problems. It is called upon at once to refer to Christians living in al-Andalus, including those such as the Córdoba martyrs who resisted arabicization, but also Christians outside al-Andalus who were distinctive precisely because of their Andalusi acculturation, thus differing in origins and attitudes almost entirely, as well as those Christians who stuck to the Hispanic liturgy after the Christian takeover of their territories, a different problem that falls beyond our remit here (Hitchcock 2008). This differentiation, however, highlights that aside from the movement of clerics, and hints like the villa de cordobesos already mentioned, our evidence for this phenomenon is primarily cultural, and need not mean that the bearers of such culture were born into it. The possibility of northern visitors to the south bringing ideas and books back, for example, is little considered. The belief in Christianity’s cultural superiority implicit in such scholarship seems to allow only immigrants to have such cultural uses, immigrants who won admiration for their creativity, but whose styles no native would lastingly adopt. Few if any scholars would outwardly profess such a belief, yet as soon as that assumption is discarded it must be clear that the works themselves could be transmitters of such ideas, that anyone of suitable ability who had seen such things, whatever their origin, could have thus appeared “Mozarab-like,” and that we should therefore expect considerable cultural fluidity on both sides of the frontier.

In common and in conflict

A person crossing this space could thus likely not have told from material culture alone when they had left al-Andalus or entered Christian territory. The ceramic record also suggests this: a coarse grey ware is found almost all over the Peninsula between the tenth and twelfth centuries, despite considerable earlier differentiation, and the only finewares usually recovered are of an Islamic, and even Persian, type (Barceló 1993; Riu-Barrera 1999). Much of the grey ware must have been of local manufacture, and it is probable that more detailed studies will locate more differences than have been identified, but a culture more or less shared across the zone and beyond it seems easiest to suppose even so, however scanty the current evidence.

This should not, of course, be taken to imply that all was harmony in this space. While the vituperations of monastic documents against the bandits who lived in such places, failing to respect the monasteries’ unenforceable land claims there, were likely more programmatic than realistic (Jarrett 2010b, 98–99), armies travelled through these zones and the first presence of outside interests often came with fortifications, whose upkeep became the job of the hitherto-ungoverned locals (Benet i Clarà 1991–1992; Escalona Monge 2000–2001, 109–113). At Cardona, on the edges of Catalunya Vella, in 987 the count of Barcelona offered the settlers and citizens many legal privileges to motivate them to stay and defend a fortress that was obviously impossible for him to protect alone: this was the Barcelona family’s third attempt to repopulate the city in a century, which shows the problem, but it is that the count undertook to guarantee the citizens’ possessions of any lands they could take, from either Muslims or Christians, that shows how few guarantees the area really offered (Galera i Pedrosa 1998, no. 7). Other indications of border warfare and general preparation for conflict are not hard to find, even if fewer in some areas than others (Lourie 1966). This then also needs to be part of our picture of these societies: without wider links of membership of polities, they did not necessarily share friendship either, and in the old adage, no man may have peace longer than his neighbour wishes. ⁶

Reconquista and neogothicism

Against this background the idea of a Reconquista, or even as more modern literature would have, a “restoration,” seems somewhat strange (cf. Ríos Saloma 2008). The Christians or Muslims of this environment were, while different by profession from each other, not far away and far from alien. The adoption of Andalusi material culture at an élite level demonstrates the lack of commitment to its extinction those élites yet had. Nonetheless, it cannot be denied that at certain times in certain places we find the idea expressed that the Christians ought to fight the Muslims and even that the Visigothic kingdom was to be restored by so doing (Isla Frez 2011). It is easy enough to understand why King Alfonso VII evoked such ideas as he laid siege to Toledo, the old Visigothic capital and Andalusi problem spot, in 1085, but what was the impact of such ideas before then?

The answer seems to be that this was a way for kings whose power rested primarily on military leadership to add symbolic importance to their plundering, often as part of a general campaign of symbolic uplift. We need not, for example, necessarily doubt the Chronicle of Albelda when it tells us that King Alfonso II restored the Gothic ordo at Oviedo just as it had been in Toledo, even if nobody can have known how correctly he did so (Gil Fernández et al. 1985, Albeldense XVI.9), since we can see his ambitious self-presentation in the building projects he undertook, which included a palace complex in Oviedo (Collins 1995, 280). It is hard to see how anyone who did not already come to the court could have been influenced by such changes, however, and a battle to balance the various factions of the Asturian kingdom probably lies behind this program (Escalona Monge 2004). Alfonso’s palace was, after all, supplanted by his successor Ramiro I’s probable conversion of what is now Santa Maria de Naranco into a new palace, literally looking down on his predecessor’s works from its hill above Oviedo (Collins 1995, 280–281).

It is also easy to see why such Gothicizing efforts were eagerly reported in the reign of Alfonso III, during which the Chronicle of Albelda was concluded. The king’s name is attached to another chronicle that makes sweeping claims of royal action in the frontier zones (Gil Fernández et al. 1985, Alfonso III). Also produced in his reign was the peculiar Prophetic Chronicle, which distorts the Book of Daniel’s prophecy concerning Gog and Magog so that it could be taken to refer to the Goths (Gog) being oppressed by the Muslims (Magog) for a fixed period of 170 years, by the author’s calculation due to end in a few months with Alfonso III’s reconquest (the word does not seem wrong in this context) of the whole of Hispania (Gil Fernández et al. 1985, Albeldense XIV–XIX). Although this text must have been proven wrong almost immediately, as Alfonso’s successes were not of this order, an audience for it is demonstrated by its travel with the later chronicles, albeit perhaps for the polemical Life of Muhammad it also includes (Daniel 1994). Also clear, however, is that Alfonso III’s self-representation was also challenged: his reign ended in deposition by his sons and the transfer of the royal capital to more southerly León (Collins 1995, 284).

It is hard to see how any of this would have mattered farther into the frontier. Claims of ancestral defeats of the Muslims presumably did something to strengthen a claim on power, but in both this case and the few Catalan examples of such claims the combats involved were in the past, and often belied by contemporary sources (Jarrett 2011, 15–19). Meanwhile, along the Duero or the Llobregat, what would claims to have succeeded the Gothic kings have meant? If those kings were remembered, it would presumably have been as rulers of Toledo, which still lay in Muslim territory; the north, whence the new claims, was in fact the zone that had given those kings most difficulty of control (Manzano Moreno 1999, 42–43). Ancient tributes or loyalties could not have been called on here; the best that could be expected is that whatever low-level organisation of labour and renders might accrue to the local élites of these zones could somehow be attached to the new structures of rule northern magnates were now extending into the frontier zones (Escalona Monge 2001, 14–54; cf. Carvajal Castro 2012). The performance of the Gothic inheritance was almost certainly effective only closer to home.

The dangers of the frontier

The picture with which we close, then, is one of detachment and disconnection. The sources’ claims of desertion or banditry seem more rhetorical than factual and descriptions in terms such as no-man’s land (d’Abadal i de Vinyals 1958, 78: “terra de ningú”) refer to political possession, not habitation. This land was not so dangerous as to be empty, but it was not safe, and it was certainly not easily governed. Within it we do not find a single “gente de frontera” but many communities and identities with very local reflexes, whose involvement with others would have ebbed and flowed according to need, interest and circumstances. This makes almost any generalisation fraught but the composite of many different possibilities, not all verifiable, that is proffered here ineluctably undermines efforts to extrapolate from territories closer to the developing centres of power.

Another kind of generalisation that is dangerous is one that sees the frontier zone as unchanging over time. As has been stated previously, that its overall state remained similar over much of the period under discussion should not blind us to what we can see of a great many small changes in different places and of different durations. Towards the end of our period, however, even large-scale changes became more durable: under the Caliphate of ‘Abd al-Rahmān III and al-Hakam II the fractured polity of al-Andalus was sufficiently bonded that its periphery responded to its centre with unprecedented heed, and Christian rulers who had advanced control into the Islamic borderlands were forced to relinquish it (Kennedy 1998, 82–124). With this situation only slightly relaxed, in the 970s, came the new wars of the Amirids. Every polity in the north now found itself a target for repeated campaigning, yet these were in fact the death throes of a centralised Muslim polity in the Peninsula, unbeknownst either to victims or perpetrators. In the year 1006 civil war broke out and by 1010 the situation had deteriorated so far that Catalan and Castilian troops were called in on opposite sides of a succession struggle for the Caliphate; the Catalans sacked Córdoba itself. With this began the period of raiding and tributes known as parias, and in 1031 the last Caliph died and was not replaced, leaving the array of city-states and their rulers known as taifas to compete and contest the new and increasing Christian dominance (Wasserstein 1985).

In this changed world the old frontier zones could not remain unattached and disconnected for long, although new disconnections were now in progress throughout the Caliphate. Rather than disappear, therefore, the frontier can be seen as having moved, into the areas where in the twelfth century we find El Cid taking advantage of such larger-scale detachment to form a new principality around Valencia, among others like him. The ways that the polities of the north had developed to advance their control into such disconnected zones in the centuries before would serve such men well, but this does not erase the long period in which that control was not extended and in which the people we have studied made their own ways.