CITY STATES IN THE LATER MEDIEVAL MEDITERRANEAN WORLD *

This article offers a comparative study of city states in the Christian and Islamic spheres of the later medieval Mediterranean world, with a particular focus on Italy, Syria and al-Andalus. Medieval city states are not usually associated with the Islamic world, but rather with a narrative that has foregrounded the exceptional nature of European cities in world history, especially the famous city republics in Northern and Central Italy, and the role that city states played in the formation of European states. Yet city states were a phenomenon that could be observed across urbanized regions of the Mediterranean world where cities turned into important political arenas in the context of sustained political fragmentation. City states are best approached as political systems that were characterized by brittle regimes and experienced high levels of political volatility: they often lacked a clear boundary between the ‘inside’ and ‘outside’ of city states and were characterized by the multiple political organizations that crystallized in, and fought hard to control, urban political space. The most commonly shared type of political organization in city states was the urban lordship, but city-based lords usually found themselves in intense competition with elite-based collective associations, families and factions, and popular political organizations.

When senility befalls a dynasty and its shadow recedes from the remote regions of the realm, the inhabitants of the cities of that dynasty have to take care of their own affairs and to look after the protection of their own place. For Ibn Khald un, city states were necessarily brittle political constructs that were fundamentally different from the states that they replaced. As these political formations emerged, the city's elders, but also sometimes men from the lowest social classes, would vie for supporters to gain power in the city and oust their opponents. Such regimes tended to be so weak that they would eventually be overrun by tribes, which, in Ibn Khald un's vision of the world, were the real driving forces of history. For Ibn Khald un, city states were in many ways by-products of a swift succession in the crystallization, disintegration and collapse of larger states. 1 Ibn Khald un's historical theory applied first and foremost to his native North Africa, a region that profoundly shaped his historical imagination, but his emphasis on the volatile nature of city states is echoed by many contemporaries elsewhere in the Mediterranean world. Famously, Italian thinkers, from Dante to Machiavelli, lamented the strife that plagued Italian cities and deplored how internal divisions were frequently entangled in external conflicts that threatened the fortunes of city states. 2 City states are not usually associated with the Islamic world, but with a narrative of European cities as distinct constructs. In a powerful argument that ultimately goes back to Max Weber, European cities alone were seen as urban communities that ruled themselves in relative autonomy or even independence. 3 The later Middle Ages have always played a particularly important role in this grand narrative. This is especially true for the city states of later medieval Northern and Central Italy, which have often been seen as preparing the ground for uniquely 'Western' achievements such as the development of democracy or the rise of the modern state. 4 Some aspects of this narrative have been challenged in recent decades. Historians have been especially critical about the notion that Italian communes were proto-democracies, when they were at best oligarchies dominated by elite groups that often coexisted with, or were supplanted by, various forms of lordship. 5 At the same time, assumptions about the 'stateness' of city states, and their inscription in narratives about the development of European statehood, have had more staying power. A large amount of work on Italian city states is concerned with classic themes of statehood, such as the nature of urban governments and the degree to which they were independent. As Andrea Zorzi has pointed out, historians have tended to concentrate on the study of communal governments, but have focused to a much lesser degree on what he calls 'informal' and 'infrapolitical' practices and institutions in Italy. 6 In an important book that challenged the uniqueness of Italian city states in a European context, Tom Scott also predominantly focused on the subject of territorial control and considered a variety of different channels through which city states were able to acquire control of their hinterlands. 7 Attempts to deploy the concept of the city state outside Europe have also often been overshadowed by ideas about the state. At the turn of the millennium, a large research project 4 Anthony Molho, 'The Italian Renaissance, Made in the USA', in Anthony Molho and Gordon S. Wood (eds.), Imagined Histories: American Historians Interpret the Past (Princeton, 1998). 5 One of the classic statements of this position is P. J. Jones, 'Communes and Despots: The City State in Late-Medieval Italy', Transactions of the Royal Historical Society, 5th ser., xv (1965), but there is now a vast historiography arguing along these lines; see n. 12 below. A relatively recent monograph that reasserted the exceptional nature of city-state republicanism in Italy is Mario Ascheri, Le città-Stato: le radici del municipalismo e del repubblicanesimo italiani (Bologna, 2006). 6  headed by Mogens Herman Hansen identified thirty so-called 'city-state cultures' in world history in a way that questioned the exceptionalism of European city states, but at the same time also worked with a definition of city states that recalled fundamentally modern European ideas about statehood and sovereignty. Hansen defined the city state as 'a highly institutionalized and highly centralized micro-state': governments of city states did not always enjoy independence, but they exercised 'internal sovereignty' and enforced 'a legal order within a territory over a population'. From this perspective, city states were not multicentric, but power radiated out from one recognized centre. 8 Even though this project succeeded in broadening the field of inquiry beyond European cities, the baggage that comes with concepts like the state has frequently made comparisons difficult. Urban historians of the Islamic world have often felt uneasy about what they regarded as fundamentally European conceptual categories and it is also for this reason that city states were never an especially important concern in the extensive historiography on the 'Islamic city'. There has, in fact, been an animated debate among historians about whether it even makes sense to speak of 'urban autonomy' in the context of Islamic cities, when these lacked formal legal privileges and did not develop commune-like institutions that were the supposed hallmarks of 'Western' city states. 9 In my research I considered the governments of city states alongside the range of other political organizations with which they competed in cities. I have found two oft-made assumptions about city states to be particularly problematic. First, there is frequently an expectation that city states were independent and that their governments were largely dissociated from other structures of political authority. It has often been suggested in the aforementioned literature that Islamic cities rarely achieved this kind of independence, but it is important to stress that most Italian city states did not do so either. The city states of the later medieval Mediterranean world were generally entangled in complex political and legal arrangements within the fragmented landscapes that typically surrounded them, even as city states constituted distinct spheres within this context. City walls did not demarcate urban political arenas neatly and there was often a blurred line between political organizations inside and outside the city. City-based lords -the most common form of city-state government, as described later -were often closely associated, or even found themselves in relationships of dependence, with various rulers and political formations. Furthermore, divisions between rival political groupings in many city states were kept alive, and were even deepened, by the close links between such groups and external political players to which they were often more closely linked than they were to their immediate neighbours.
Second, it is also not helpful to approach city-based governments as quasi-sovereign entities within the political order of their city states. This is true for city-based lords, and also for Italian communal governments, which have often been discussed as if they were the only political system in city states, when usually they existed in a state of permanent tension with other players. The very high levels of violent conflict in city states have often been seen as something dysfunctional, precisely because there is an expectation that city-state governments should have presided over an integrated political order within cities. In reality, these conflicts were an expression of the limited degree of control that governments were able to exercise over the multiple political organizations that crystallized in city states.
City states are most helpfully thought of as arenas that became highly politicized spaces in the fragmented political landscapes that surrounded them. The term 'city state' still serves a useful purpose to capture the degree to which cities constituted distinct and separate political spheres, but the term 'state' should not be made to imply a government-centred political order with all the assumptions about external independence or internal control that come with it. Many historians would hesitate to apply such a narrow definition to most medieval states, so it would seem problematic to apply it to, or demand such characteristics of, city states in this period and region. 10 In what follows I proceed by analysing, from a comparative perspective, the different competing forms of political organization that can typically be found in later medieval Mediterranean city states. I do not limit myself to formalized legal institutions, but include all regularized forms of social interaction that had repercussions on the governance of cities. 11 In section II, I focus on what is arguably the most common form of political organization in later medieval Mediterranean city states -lordships. City-based lords acted as the rulers of their cities, but we shall see that they were often in relationships of dependence with other governments, frequently pursued their own dynastic interests and generally presided over regimes that could be short-lived or highly volatile. Section III focuses on competing political organizations: collective organizations associated with the city's elite, families and factions, as well as popular organizations. As we shall see, institutions of communal government, for which Italian city states are so well known, should be treated as only one of the various political organizations that could appear in city states.
The focus of this article is on three regions of the later medieval Mediterranean world: Northern and Central Italy, Syria, and al-Andalus. These three regions shared two important 10  characteristics. First, they were heavily urbanized and had medium-to large-sized cities whose history in most cases reached back to the Roman period or earlier. Second, during the period under consideration in this article, all three regions experienced prolonged periods of political fragmentation during which city states crystallized. We shall see that there were important differences between the various city states, but the divide between Christendom and Islam was only one factor; differences between and within particular regions were just as, if not sometimes more, important. Other Christian and Islamic regions also at times witnessed the formation of city states under similar conditions, but my focus here is on comparing Northern and Central Italy to the two Islamic regions of Syria and al-Andalus. For each of these regions, I have chosen a number of case studies within a specific time frame.
The cities of Northern and Central Italy (frequently abbreviated as Italy hereafter) are, by far, the best-studied city states. 12 As is well known, the withering away of the kingdom of Italy, ruled by the Western emperors who were also kings of Germany, ushered in a prolonged period of extreme political fragmentation during the twelfth and thirteenth centuries. Dozens of city-based political regimes sprang up, which had their own governments, military organization and judicial administration and which ruled the city and its wider hinterland (known as contado). This fragmentation lasted until the fourteenth and fifteenth centuries when the vast majority of city states were absorbed into larger regional states. 13 As has already been emphasized, it is important to bear in mind that city communes were not the only type of political organization to crystallize in city states. Nor were city states the only beneficiaries of this fragmented political landscape -so were rural lordships, ecclesiastical jurisdictions and other political entities. In order to capture the sheer diversity of experiences within the confines of an already rather long article I have chosen to foreground the cases of Verona, which saw the development of different forms of lordship, and Piacenza, which experienced particularly pronounced levels of volatility. I have largely focused on the thirteenth century for which we know a great deal about the range of political organizations crystallizing in city states.
The political landscape of Syria (by which I mean geographical Syria or Bil ad al-Sh am) was similarly fluid after the collapse of the iAbb asid Empire. City states emerged at different moments between the tenth and twelfth centuries, as larger states broke up and intense geopolitical competition opened up new opportunities for a range of protagonists. 14 My focus here is on the late eleventh and twelfth centuries when the Syrian political landscape was fractured between a number of political players. The Egypt-based F at : imid Empire had mostly faded away in Syria, but vestiges of F at : imid government were still present in coastal areas. The Iraq-and Iran-based Salj uq Empire conquered most inland areas of Syria in the later eleventh century, but after the death of the Salj uq prince Tutush (d. 1095) the Salj uq territories in the region rapidly disintegrated into numerous smaller states. I have chosen to focus on three cities that illustrate different pathways for Syrian cities to become city states: Tyre and Tripoli where local judges became rulers of the city; and Damascus where foreign military commanders established themselves as lords who permanently based themselves in the city. Only in the later twelfth century was the whole of Syria durably absorbed into the framework of the Zangid and Ayy ubid political formations. It can be argued that certain features of city states were preserved in this altogether different context, but this lies outside the scope of this article. 16 City states can be observed over long stretches of time in the Muslim region of the Iberian Peninsula (al-Andalus). The phenomenon is best known during the eleventh century when, in the aftermath of the collapse of the Córdoba-based Umayyad state, Andalusī cities became the bases for around three dozen small political formations known as taifas. The North Africabased Almoravid and Almohad states reconquered most of these territories, but after their collapse (in the second quarter of the twelfth century and the first quarter of the thirteenth century, respectively) taifas crystallized again, often in the same places. 17 In this article I focus on the particular circumstances of the thirteenth century when al-Andalus became a mosaic of political players -not just city states, but also regional military leaders, the rulers of Almohad splinter states in North Africa and Iberian kings who conquered most of al-Andalus by the mid thirteenth century. All of these players benefited from centrifugal tendencies in the Almohad state itself and the defeat of the Almohad army by Christian troops at Las Navas de Tolosa (1212). 18 Particularly important for our comparative perspective is the city state of Granada, which developed in the 1230s and, though beset by many problems, survived until it was conquered by the Catholic kings in 1492. Because of the particular nature of its urban government, I also discuss the case of Ceuta, which was a North African city, but, located just across the water from Gibraltar, was also part of the Andalusī environment.

II
The most commonly shared phenomenon to emerge from these fragmented political landscapes was city-based lordships. This is not only true for Syria and al-Andalus, but also for Italy where the vast majority of city states came under the rule of city-based lords (signori ). An older historiography has thought of city-based lordships as a phenomenon that followed, and marked the failure of, 'communal' or 'republican' experiments in Italian city states. The idea that lordships (signorie) were somehow distinct from communes has, however, been challenged, most recently in a large research project led by Jean-Claude Maire Vigueur. This project unearthed the cases of some 371 urban signori between the thirteenth and mid fifteenth centuries that overwhelmingly show that the hegemony of individuals or families over the political process in cities was an intrinsic part of the city-state experience, often long before the power of signori was formalized in legal terms. In fact, signorie covered a wide spectrum from informal forms of domination to fully fledged princely governments. 19 Across the three regions analysed here, city-based lords primarily regarded their cities as their main political arenas and turned away from courts, armies or bureaucracies as their principal points of political orientation. At the same time, citybased lords remained entangled in wider political and legal spheres and they frequently occupied a fuzzy position at the intersection between city states and other political formations. In what follows we shall see, first, that even lords who emerged from within urban society were often tied to external rulers in a variety of ways. Second, many city-based lords were actually warlords and regional political players who established themselves in cities as autonomous rulers, but often defied a clear distinction between the 'outside' and 'inside' of city states. Finally, the desire to establish dynastic regimes was itself often an attempt by many lords to consolidate their control of city states in this fluid geopolitical context. In fact, the status of citybased lords was often uncertain and they were frequently prone to challenges from within and without.
Signori call into question the oft-made assumption that Italian city states were uniquely independent political entities, while Islamic city-based rulers were necessarily part of a wider umbrella, such as the caliphate or particular empires. A considerable amount of recent research has shown that Italian city states were, to varying extents, always tied to wider political and legal spheres. Most contemporary Italian jurists agreed that de jure Italian city states remained subject to the legal authority of the Western Empire and debated the extent to which urban governments could claim de facto sovereignty. 20 Imperial government was revived seriously in 1226-50 and 1310-13 by Emperors Frederick II and Henry VII, albeit with different rates of success in different parts of the region. In the wake of Charles of Anjou's invasion of Southern Italy in 1266, many city states put themselves under the formal protection of Charles and became part of temporary intra-and inter-regional political configurations. Many city states in Central Italy accepted the temporal dominion of popes long before there was a Papal State. 21 Signori were particularly likely to be associated, or even found themselves in a relationship of dependence, with other rulers. They were often not only allied with external powers, but also acted as their military captains. Alessandro Barbero has found on the basis of the Maire Vigueur database that nearly a fifth of all signori were employed as mercenary captains (condottieri ) by other rulers or governments. Many signori saw their, sometimes dubious, position in urban politics formalized and even legalized, when they accepted appointment as imperial or papal vicars. As Andrea Zorzi has argued, this external validation often constituted a crucial stepping stone in the consolidation of signorial power in a given city. 22 An interesting example is the della Scala family who became highly effective signori of Verona, but were also closely interwoven with the political and legal framework of the Western Empire. 23 The della Scala were a city-based family, among whose members had been a number of iudices (judges), that rose to power in the 1260s when they put themselves at the head of a new political coalition. Like many other city-based rulers in the Mediterranean world, the della Scala were able to create a dynastic regime and they remained in power as Verona's signori for a comparatively long time until Verona was conquered by Milan in 1387. However, in order to understand the della Scala's entanglement in external frameworks of power it is important to know that their rise was slow and their position was only formalized on a piecemeal basis. Although Mastino della Scala and his brothers had already begun to monopolize a number of key positions in the 1260s, only his brother Alberto (r. 1277-1301) was given a degree of formal recognition by communal institutions after Mastino had been assassinated. Subsequent della Scala signori continued to coexist with the institutions of the commune whose formal recognition every new signore needed to secure. The dynastic entitlement of the 22 Alessandro Barbero, 'I signori condottieri', in Maire Vigueur (ed.), Signorie cittadine nell'Italia comunale, 229; Andrea Zorzi, 'Ripensando i vicariati imperiali e apostolici', in Grillo (ed.), Signorie italiane e modelli monarchici. For imperial and papal vicars, see also Marie-Luise Favreau-Lilie, 'Reichsherrschaft im spätmittelalterlichen Italien: Zur Handhabung des Reichvikariats im 14./15. Jahrhundert', Quellen und Forschungen aus italienischen Archiven und Bibliotheken, lxxx (2000); and P. J. Jones, 'The Vicariate of the Malatesta of Rimini', English Historical Review, lxvii, no. cclxiv (1952). 23 The key study on the della Scala is Gian Maria Varanini (ed.), Gli Scaligeri, 1277-1387 (Verona, 1988); for individual lords the most detailed studies available are the entries in the Dizionario Biografico degli Italiani.
della Scala to the signoria of Verona was, in fact, not formally confirmed in statutes until 1359. 24 Given this degree of uncertainty, it is not surprising that the della Scala were also keen on securing their position in the wider geopolitical theatre of Northern Italy. One well-known strategy, tried by most signori, was the conquest of neighbouring cities, which the della Scala pursued especially vigorously between 1312 and 1341. As we shall see, the subjection of other, often smaller, cities was also a feature of many city states in the Islamic world. Another, much less-studied, strategy was for signori to develop a close relationship with, and sometimes even accept a degree of subjection to, other rulers. As one of the leading Ghibelline families in Italy, the della Scala are a good example of a signorial family who always sought a close association with the Empire. Contemporary chroniclers were keen on stressing that Mastino hosted Emperor Frederick II's grandson Conradin on his mission to reassert imperial control over Italy as early as 1267, while his nephew Bartolomeo (r. 1301-4) married into the imperial family in 1291. 25 In 1311, Alboino della Scala (r. 1304-11) and Cangrande della Scala (r. 1311-29) paid the large sum of 17,300 florins to receive an imperial vicariate for Verona from Emperor Henry VII. This, in theory, meant not only acceptance of the emperor's authority, but also fiscal and military obligations, though the della Scala, of course, also had their own agenda. Sometime after Henry VII's death, Cangrande della Scala contemplated, but in the end decided against, accepting a papal vicariate from Pope John XXII and, in 1328, he conquered Padua against the wishes of Emperor Ludwig IV of Bavaria. Cangrande's successors nevertheless continued to seek vicariates from Ludwig and his successor Charles IV, not least because imperial confirmation remained an important tool of legitimation in city politics. When Antonio (r. 1375-87) and Bartolomeo II (r. 1375-81), two illegitimate children of Cansignorio (r. 1351-75), came to power in highly controversial circumstances after the assassination of their uncle and rival contender for control of the lordship, they were not only keen on obtaining consent from the communal assembly at Verona, but also sought and received an imperial vicariate. In 1381, Antonio killed his brother Bartolomeo, but this time Wenceslas IV allegedly refused to confirm Antonio's position. Even when the Empire became a more distant reality as a political and legal framework in the late fourteenth century, the della Scala still imagined themselves to be occupying a place within the wider sphere of imperial politics. 27 The lords of the Syrian city states obviously acted in a different political context, which, for the period under investigation here, was characterized by the receding empires of the F at : imids and Salj uqs. Even so, city-based lords in Syria occupied a position that was, in some ways, comparable to that of their Italian counterparts: on the one hand, they were largely autonomous rulers, but, on the other hand, they were still tied to larger political formations that they could not ignore. Two cases that illustrate this well are the Syrian coastal cities of Tripoli and Tyre, which shook off control by the F at : imid Empire and, for a brief period, became city states in their own right. In 1070, Ibn iAmm ar, the judge (q a : dī ) of Tripoli, led a revolt against F at : imid control and was installed as ruler of Tripoli. In the same year, another q a : dī, Ibn Abī iAqīl, became the ruler of Tyre in a similar way. Both men were able to engineer their succession along dynastic lines, like the della Scala and so many other city-based rulers. Ibn iAmm ar was followed in his position as ruler of Tripoli by his nephew Jal al al-Mulk who, in turn, was succeeded by his brother Fakhr al-Mulk, while Ibn Abī iAqīl was succeeded by his sons. 28 As far as is possible to tell from contemporary accounts, Ibn iAmm ar, Ibn Abī iAqīl and their descendants behaved like rulers in their own right. Both conquered neighbouring territories, in the case of Ibn Abī iAqīl of Tyre acquiring land stretching as far south as Acre, conducted diplomatic relations with neighbouring powers and led military operations. Fakhr al-Mulk of Tripoli was able to resist the crusader army of Raymond of Saint-Gilles and a Byzantine fleet for two years. 29 That these men played such an important role in Tyre or Tripoli is not entirely surprising. Both their families were very wealthy, as they were mentioned as the owners of ships in the documents of the Cairo Geniza; Ibn Abī iAqīl's ships, in particular, appeared frequently in the Geniza records and operated across the Mediterranean world. 30 As judges, both men had already played an important role in the city. Islamic law regulated a broad spectrum of urban public life, including criminal law, family law and the law of contracts, and the takeover of judges in situations of political crisis is itself a wellknown phenomenon of taifa governments in al-Andalus. 31 To some extent this is paralleled by those Italian signori who were prominent citizens prior to turning themselves into lords. As we have already seen, the della Scala themselves originated from a family that had in the past included judges. 32 28  Both Tripoli and Tyre successfully kept the F at : imid and Salj uq rulers at bay. A surviving inscription named Tripoli's Jal al al-Mulk without any mention of either the F at : imid or Salj uq rulers. 33 It is nevertheless difficult to imagine that the Salj uq or F at : imid empires did not reach into the politics of these city states and it is perhaps more useful to think of Tripoli and Tyre as shifting between the nominal frameworks of these larger territorial formations. We know that the rulers of Tripoli at least turned to the Salj uq sultan for support in moments of crisis. In 1092, when Salj uq troops threatened Tripoli, Jal al al-Mulk allegedly produced documents issued by the Salj uq sultan that confirmed him as the ruler of his territories and forbade all Salj uq military commanders to wage war against him. Sixteen years later, when Tripoli was under Frankish siege, Jal al's brother Fakhr al-Mulk led a diplomatic mission to Baghdad where he was received by the Salj uq sultan and the iAbb asid caliph, neither of whom seem to have held a grudge against him. According to the thirteenth-century historian Ibn al-Athīr, Fakhr al-Mulk was treated very well at court and was even offered a military contingent. The sultan, however, also made sure to signal that this was not a relationship between equals and gave Fakhr al-Mulk a robe of honour as a symbol of sultanic suzerainty. 34 It is unlikely that the sultan made much practical difference to the rulers of Tripoli, but it is nevertheless interesting that these rulers were sometimes prepared to see themselves as part of the wider orbit of Salj uq politics. At the same time, there is also a reasonable amount of evidence to suggest that the F at : imid Empire had its own adherents in Tripoli and Tyre, quite possibly among Shīiites who, according to the eleventh-century Persian traveller N as : er-e Khosraw, constituted the majority of their populations and may have looked anxiously at the repressive measures taken in the Salj uq Empire against Shīiites at around this time. Even before Fakhr al-Mulk was able to return from Baghdad in 1108, the population of Tripoli had rebelled against him and returned the city to the rule of the F at : imid Empire. In Tyre, too, there were sympathies for the F at : imids: the descendants of Ibn Abī iAqīl proved unpopular and by 1089 the city was again in F at : imid hands. Four years later, the F at : imid governor Munīr al-Dawla sought to declare his independence from the F at : imids, but Tyre's population was unwilling to support his rebellion and called in the F at : imid army to re-establish control over the city. As will be seen later, such tight connections between particular urban groups and outside powers often played a key part in the high levels of conflict that city states experienced. 35 The della Scala, iAmm ar and iAqīl dynasties originated from within the urban societies of Verona, Tripoli, and Tyre, respectively, but the blurred line between city states and the world around them is even clearer for another kind of city-based ruler that featured in all the three regions under investigation here -regional political players and warlords who established themselves as city-based lords, often with the same aspiration of creating dynastic regimes. Cities not only offered resources that these players hoped to squeeze, but as existing centres of politics and administration in the highly urbanized regions under analysis here, cities were obvious poles that attracted aspiring petty rulers who nevertheless retained connections with and often even remained nominally subject to other rulers. An especially clear-cut illustration of this kind of ruler is the lords of city states that crystallized in the inland areas of Syria in the late eleventh and early twelfth centuries. Most of Syria had been swept up in Tutush's conquest and therefore came under Salj uq rule, but after Tutush's death centrifugal mechanisms within the Salj uq state allowed essentially foreign military commanders to seize control of important cities and de facto rule these as city states away from the direct control of the sultan. There has been a tendency in the scholarship on the 'autonomy' of Islamic cities to focus on the takeover of autochthonous urban groups, but there is no reason why, in the fragmented context that produced city states, external elites should be excluded, when they fully identified with and permanently settled in the city that they came to rule. This process can be observed clearly for Damascus, but it is also true for Aleppo and, for shorter periods, other cities. 36  passed to his son Duq aq who died prematurely in 1104. After Duq aq's death, his guardian (atabeg) Tughtigin became the city's de facto ruler and was recognized as such by the Salj uq sultan Muh : ammad Tapar in 1116. Like the rulers of other city states, Tughtigin had not severed ties with his supposed superior, although there is little evidence to suggest that he offered much practical assistance to the Salj uqs. Instead, Tughtigin was busy creating his own apparatus of government. He was able to install his son Taj al-Mul uk B urī as his successor after whom the dynasty of this city state, the B urids, were also named. They were able to rule until 1154 when the city was incorporated into the much larger Zangid state, though, as we shall see, not without a considerable degree of conflict within the dynasty and between the dynasty and other urban groups. 37 What is remarkable about this process is that, as Jean-Michel Mouton has shown, the B urids and their associated elites wholly inserted themselves into the city's political arena. Even though they remained distinct from the local Damascene population in terms of their Turkish language, customs and ethnicity, they became political players whose focus was firmly on Damascus and the area ruled by this city state. Work on the city's citadel, located in the north-west corner of the city, began around 1100, and most residences and tombs associated with the regime were also built in that area of the city. Their marriage alliances were not aimed at the wider Salj uq world, but were mostly conducted within the military society of Damascus or neighbouring principalities such as Homs, Shayzar and Aleppo, often with the aim of gaining recognition or securing territories. 38  What was interesting about the B urids from a Mediterranean perspective was not so much that they were warlords who had turned themselves into city-based lords, but that, as Turkish invaders, they had come from such a distance and were, initially at least, foreign political players in Syria. In thirteenth-century al-Andalus and Italy, by contrast, city-based rulers who had only relatively recently arrived from outside the region were less common. 39 In al-Andalus, the Almohad governors of Baeza and Valencia, members of the Marrakech-based caliphal dynasty, tried to establish themselves as autonomous rulers in the 1220s. However, their efforts were unsuccessful, even by the standards of city-based rulers. iAbd All ah al-Bayy asī of Baeza was assassinated in an urban uprising after only three years, while Ab u Zayd of Valencia fled another uprising and ended up in Aragon where he took the name of Vicente Belvis. 40 In Italy, lords from outside the region were even rarer. A well-known, but unusual, case was that of Walter of Brienne, signore of Florence between 1342 and 1343, whose family of French descent was based in Southern Italy and who was the nominal lord of the Angevin Duchy of Athens. 41 Warlords and regional players who established themselves as the rulers of cities were nevertheless a prominent feature of city states in both Italy and al-Andalus, but they tended to come 39 In al-Andalus, outsiders were more common in the eleventh-century taifas; see Guichard and Soravia, Los reinos de taifas, 42-7. 40  from within, rather than from outside, the region. In many ways, such lords epitomized the blurred external boundaries of both Andalusī and Italian city states. Even while they were busy building their dynastic city-state regimes, they were and often continued to be entangled in wider political and legal spheres.
Many Andalusī and Italian warlords already had a degree of association with, or had even been implicated in the governance of, the city states that they came to rule. An interesting Andalusī example is Zayy an ibn Mardanīsh, the commander of the Almohad cavalry in al-Andalus who seized control of Valencia in late 1228 or early 1229. Zayy an was no stranger to the city, but was the great-great-nephew of Muh : ammad ibn Said ibn Mardanīsh who had ruled Valencia between 1147 and 1172. Even though Valencia had thereafter been absorbed into the Almohad state, many aspects of the Mardanīsh regime were not actually disrupted, as Pierre Guichard has shown: Muh : ammad's brother Y usuf had remained Valencia's governor until his death in 1186, one of his military commanders had continued as the governor of Chinchilla and even religious elites had kept their positions. Zayy an may have tried to resurrect the regime of his ancestors by capitalizing on his existing ties to the city of Valencia. In the hope of receiving support from abroad, he also pledged allegiance to the iAbb asid caliph at Baghdad and later the H : afs : id ruler of Tunis. However, his project ended prematurely when Valencia was conquered by Aragon in 1238. 42 Many Italian signori matched the profile of city-based rulers who had already enjoyed close ties with the city, but also operated on a regional scale as military leaders. The Malatesta became the dominant political players in Rimini during the course of the thirteenth century and, after leading a coup with the help of the local Guelf party in 1295, turned themselves into Rimini's effective signori, a position that was confirmed by statutes in 1334 as well as by the bestowal of a papal vicariate in 1355. At the same time, the Malatesta already had an important profile outside Rimini, as they were lords of Verucchio in Rimini's hinterland and had already acted as military leaders, usually for pro-papal coalitions. Well before establishing himself 42  as Rimini's de facto lord, Malatesta da Verucchio was appointed vicar of Florence by Charles of Anjou in 1267 and acted as Pope Martin IV's military commander in Romagna before falling out with the curia. In the fourteenth century, the Malatesta went on to become one of the foremost families of condottieri in Italy. 43 Not all Andalusī and Italian city-based rulers necessarily had a prior connection with the city, even when they came from the surrounding region. In al-Andalus, this type of lord is best exemplified by the Ban u'l-Ah : mar or Ban u Nas : r (Nas : rids) who became rulers of Granada. 44 We know little about the Nas : rids' origins, but they were almost certainly military commanders at the Andalusī frontier. Exploiting the uncertain geopolitical circumstances in early thirteenth-century al-Andalus, Muh : ammad I initially became ruler of the small city of Arjona after an urban uprising there in 1232, and in 1237 he was able to enter Granada, which Muh : ammad's dynasty would rule as a city state until 1492, paradoxically making it the most long-lived Andalusī (and possibly Islamic) city state. As will be seen, Nas : rid Granada was perpetually prone to dynastic in-fighting and almost continually subject to threats from other political players inside and outside the city, but the Nas : rids' continued grip on power was at least in part owed to their ability to quickly strike an alliance with the city's establishment. According to the fourteenth-century Granadan historian and statesman Ibn al-Khat : īb, Muh : ammad had been invited into the city by two shaykhs and a delegation of the city's notables who swore an oath of loyalty to their new ruler. The Nas : rids entertained close relations with four leading families who dominated appointments to q a : dī-ships and they were soon viewed by Granada's religious elites as the last bulwark of Islam on Iberian soil. 45 Calling themselves sult : an or amīr al-muslimīn, issuing coinage and conducting their own quixotic foreign policy, the  (Rimini, 1998). 44 The key works on Nas : rid Granada are Rachel Arié, L'Espagne musulmane au temps des Nasrides, 1232-1492 (Paris, 1973); Miguel Á ngel Ladero Quesada, Granada: historia de un país islámico, 1232-1571, 3rd edn (Madrid, 1989); and L. P. Harvey, Islamic Spain, 1250-1500 (Chicago, 1990). 45 Historia de los reyes de la Alhambra, ed. and trans. J. M. Casciaro Ramírez and E. Molina Ló pez (Granada, 2010), 137; a similar version is reported in Ibn Nas : rid rulers were to all intents and purposes autonomous political operators, but they also at times acknowledged the suzerainty of various other rulers, such as the iAbb asid caliph (based in Cairo since 1261), the Almohad and H : afs : id Empires, the H udid kingdom of Murcia and even the kings of Castile. 46 In Italy, especially in the Po Plain, there were also signori who did not have a prior base in the city where they established themselves as lords. Arguably, however, these rulers found it much harder to build durable city-state regimes in Italy than the Nas : rids did under exceptional circumstances in al-Andalus. One example is Ezzelino da Romano, a notorious and muchmaligned warlord who had extensive landed holdings in the triangle between Vicenza, Treviso and Belluno, but went on to become Verona's de facto lord between the 1230s and his death in 1259. Never holding a formal title which confirmed his position as the ruler of Verona, Ezzelino owed his success to his ability to connect with Verona's Monticoli faction. According to the chronicler Paride da Cerea, the decisive moment for Ezzelino came in 1226 when he was called to Verona to become the city's chief judicial officer (podestà) in the wake of a realignment in Verona politics when the Monticoli faction were joined by another urban group known as the Twenty-Four (quattuorviginti ). He was thereafter expelled for a short period, but Ezzelino was soon re-called and, after becoming one of Emperor Frederick II's main military leaders in Northern Italy, he established himself as Verona's ruler. Ezzelino then also came to dominate other cities, but his main base was Verona where he also built his own urban castle. As Gian Maria Varanini has shown, Ezzelino did not supplant, but exercised his power alongside the institutions of the commune, while also pursuing an aggressive policy towards the Monticoli's opponents who were exiled from the city in 1239. 47 Many lords tended to gravitate towards one major city, but it is important to point out that some warlords also tried to establish supra-urban political formations. Uberto Pallavicino was a major landowner and warlord in north-western Italy who was appointed imperial vicar in 1249 and during the course of the 1250s obtained dominion over many Lombard cities, including Cremona, Piacenza, Brescia and Milan. In many ways, however, Pallavicino at best presided over a loose federation of city states. Pallavicino only seems to have directly ruled Cremona (alongside the local leader Buoso da Dovara), while the other cities were de facto governed by local allies. As we shall see later, in Piacenza the city's actual ruler was the local faction leader Ubertino Landi who governed more or less independently from Pallavicino. In fact, in the wake of Charles of Anjou's invasion, Pallavicino's political edifice dissolved, as individual city states broke away from his control. 48 Ezzelino da Romano and Uberto Pallavicino not only shared the experience of establishing themselves as signori without a pre-existing urban base, but they also illustrate the difficulties that such rulers faced. Unlike so many other Italian lords neither of them were, in fact, able to build hereditary regimes. In contrast to Syria or al-Andalus, complete outsiders to cities may well have found it particularly hard to establish themselves in Italian city states.
There were, therefore, a variety of paths to urban lordship, almost all of which involved the entanglement of lords in wider political and legal spheres. The shared aspiration of city-based lords across Italy, Syria and al-Andalus to establish dynastic regimes was certainly an attempt to consolidate their control of city states amidst the challenges that they could face from both within and outside. For some, cities were undoubtedly useful arenas within which to pursue dynastic interests that, in a different environment, they might have pursued in the context of other frameworks of power. City-based rulers not only thought of themselves as rulers of cities, but also viewed their projects as dynastic enterprises in which family interests could sometimes trump their interests as rulers of city states. The dynastic nature of many urban lordships in Italy manifested itself not only in the hereditary succession of signori, but in the degree to which their wider family was involved in the political process. There is a distinct sense that the della Scala of Verona, for instance, regarded their rule over Verona as a shared undertaking. The succession to the signoria of the della Scala involved not only sons, but also brothers and nephews of the incumbent and, for long periods of time, particular signori of the della Scalausually brothers -were co-rulers. 49 The city-based lordships of the B urids of Damascus and Nas : rids of Granada were also dynastic enterprises. In both these cases, the first three rulers saw a succession from father to son, but thereafter the succession involved a broader pool of candidates. In Damascus, the third B urid prince, Shams al-Mul uk, was assassinated by his mother Zumurrud Kh at un and real power passed to leaders of different palace factions, including at one point Zumurrud Kh at un herself, although the B urid line remained nominally in charge under the third prince's brother, Shih ab al-Dīn, who was in turn succeeded by his half-brother and then by his nephew. 50 In the case of the much more long-lived Nas : rid dynasty, rulers were drawn from a broader pool of family members through the agnatic line, involving sons, brothers and cousins from the first to the fourth degree once removed. 51 It is remarkable that in the Islamic world attempts to dynasticize city-based lordships were not only undertaken by princes of military families like the Nas : rids or the B urids, but also, as we have seen, by the judges of  (table 1), and passim. The only Nas : rid sultan not to come through the agnatic line was the short-lived Y usuf IV (1431-2) who was only related to the Nas : rids through his mother, while his father came from the Ban u Mawl: Arié, L'Espagne musulmane au temps des Nasrides, 134-7.
Tripoli and Tyre. An interesting case of a city-based dynasty of religious professionals from the Islamic West were the iAzafids of Ceuta. In the wake of the collapse of the Almohad government in Ceuta, the eminent religious scholar Ab u'l-Q asim al-iAzafī staged a successful coup in 1250. Although he initially acted in concert with the sea captain H : ajb un al-Rund ah : ī -who took command of the city's considerable fleet -Ab u'l-Q asim eventually managed to sideline H : ajb un and to install a dynastic regime. After his death, he was succeeded by his elder son Ab u _ T alib who, for a while, seems to have ruled jointly with his younger brother. Between 1307 and 1309, iAzafid control was briefly interrupted by a period of rule by Granada, but thereafter Ab u _ T alib's son Yah : y a returned to Ceuta as the city's ruler, while his brothers were in charge of Ceuta's navy and naval shipyard. Yah : y a was in turn succeeded by his son Ab u'l-Q asim Muh : ammad who ruled under the tutelage of his cousin until the city was brought under the authority of the Fes-based Marīnid rulers of the Maghrib after an urban uprising in 1322. 52 The flip-side of the dynastic nature of so many city-based lordships is that lords were often dangerously exposed to serious threats from intra-dynastic strife, which added to the already precarious position that so many city-based lords found themselves in. Even a relatively long-lived signoria like that of the della Scala experienced its fair share of challenges from within the ruling family's own ranks. Brothers or half-brothers of della Scala signori led at least five major conspiracies in 1354, 1359, 1365, 1375 and 1381, three of which resulted in the assassination of the incumbent lord. It is worth noting that these conspiracies were all concentrated in the latter half of the della Scala's reign when significant territorial losses also reduced the resources available to the family. assassination. 53 As we have already seen, intra-dynastic strife was also an important feature of city-based dynasties in the Islamic world. The most notorious, but by no means exceptional, case was the Nas : rids of Granada who were so plagued by internal divisions within the dynasty that fifteen rulers were deposed or assassinated in the wake of coups, as sons, brothers and cousins competed over control of the city state. Although the Nas : rids were able to survive for a remarkably long period of time, the factiousness of the dynasty offered many opportunities to other political actors to plot against particular rulers and, therefore, played an important role in stimulating the extraordinary volatility of the city state of Granada. 54 The evidence across the three regions suggests that the regimes of city-based lords were often volatile and many were ultimately short-lived. Even in Italy, where there were some famously long-reigning dynasties of signori, the evidence presented in the Maire Vigueur project suggests that many Italian signori, especially in Central Italy, were not able to survive beyond one or two generations. 55 There were many reasons why urban lordships were brittle. First, as we have seen, it was in the very nature of city-based lordships to flourish in the context of political fragmentation and so it was possible for lords, whose external position was often uncertain, to be swept away by changing fortunes in the wider geopolitical landscape. The dynastic nature of city-state regimes was also problematic, since the resources and size of city states were often finite and there was a limit to the extent to which urban lords could provide for their extended family -not only multiple sons, but also brothers and cousins who, as we have seen, usually asked for a share in the enterprise. Finally, urban lordships were very rarely the only form of political organization that crystallized in city states and they had to compete over the same resources with a number of other players, as I now analyse.

III
Lordship was the predominant form of political organization in Mediterranean city states, which, in its various manifestations, can be found across the three regions under analysis here. At the same time, multiple other forms of organization also crystallized in urban political arenas in the wake of the particular geopolitical circumstances that gave rise to the development of city states. Indeed, it was precisely this profusion of political organizations that contributed to the brittleness and high levels of conflict that were such an inherent feature of city states. It is possible to distinguish between a number of such political organizations: collective organizations associated with the city's elite, families and factions, as well as popular political organizations.
Italian city states are best known for the various types of councils and colleges that were associated with their systems of communal government. 56 Claims about the exceptionalism of Italian, and more broadly European, cities have often rested on these types of collective institutions, since their development was closely tied to medieval European ideas about the community as a legitimate, and even legal, basis of political organization. 57 There were indeed no Italian-style councils and colleges in Islamic cities, but it is also problematic to exaggerate this contrast, which lends itself so easily to grand narratives about the rise of democracy in the West. Even when viewed 56  solely from the perspective of Italian cities, it is important to bear in mind a few caveats that somewhat reduce the distance between Italy and the other regions. First, communal institutions were at best political organizations for the city's elite or, very commonly, for a faction within that elite. The propaganda apparatuses of Italian communes never tired of stressing that communes governed their cities for the common good, but the actual institutions of communes were always dominated by a very restricted set of the urban population: women, foreigners and most individuals below a particular level of wealth had no access to communal colleges and councils, very often particular factions in the city were also excluded and particular families anyway dominated the restricted elite who populated these institutions. This was even true for Florence, which had one of the most expansive apparatuses of communal government. In the period between 1282 and 1399, John Najemy found that fifty-one families (only about 4 per cent of the total sample) held twenty or more offices in the urban government, while the vast majority of families (962 families or 71 per cent) only held office between one and three times. 58 Second, the seemingly elevated legal status of communal colleges and councils was not necessarily reflective of the actual political dynamics in Italian city states. The very rich sourcesordinances, statutes and minutes -documenting their operation should not lead us to believe that they were always as important as they claimed to be. Signorie were often much less well documented in archives and, as we have seen, frequently took a long time to acquire a formal legal status, even though de facto they generally superseded the much better documented and frequently more formalized institutions of communes. 59 Some collective organizations, in fact, enjoyed hardly any formalized legal status in Italian cities. A very pertinent case is factions, some of which, like the Parte Guelfa of Florence, had their own councils, statutes and buildings, but the vast majority of factions were informal networks that had no such formalized dimension. Factions nevertheless played an absolutely critical role in urban political life, leading revolts that changed the balance of power in the city, dominating communal institutions or propping up the regimes of signori. 60 Islamic cities did not develop councils or colleges that would be recognizable as such to an observer steeped in the Italian political reality, but it is nevertheless possible to find evidence for organized forms of elite-based collective action. An interesting case to consider is the involvement of sh ur a-councils in the governance of city states in the Islamic West. In early Islam, the term sh ur a designated the advisory body convened to elect caliphs, but in the Islamic West the sh ur a had become a council of legal experts that was composed of the city's most eminent scholars and regularly advised judges. 61 One case for which we have repeated references to the role played by the sh ur a in the context of a city state is Ceuta under the rule of its iAzafid lords. In his Kit ab al-iIbar, Ibn Khald un described how the sh ur a swung into action at critical moments in the city state's history. According to Ibn Khald un, Ceuta's lord Ab u'l-Q asim had already been the head of Ceuta's sh ur a and we can presume that this association was important when he seized power in 1250. 62 After Ceuta had briefly fallen under the rule of Granada, Ceuta's population rebelled and it was a delegation of members of the sh ur a that was dispatched to negotiate with the sultan of the Fes-based Marīnid state. We do not know what they asked of the sultan, but shortly thereafter Ab u'l-Q asim's grandson Yah : y a was installed as Ceuta's ruler. 63 Yah : y a initially acted as governor for the Marīnids, but after a few years he broke away from Marīnid control and established himself as the city's independent ruler alongside the sh ur a. 64 It appears from all this that the iAzafids acted in close co-operation with the sh ur a. Whether the sh ur a were always on the iAzafids' side is difficult to tell and it is likely that, as in other city states, there were divisions in the city. As J. D. Latham has suggested, iAzafid rule probably collapsed relatively quickly after Yah : y a's death in the wake of internal divisions within Ceuta itself. The city's notables submitted to Marīnid rule and the pre-eminent role in Ceutan politics was now played by the H : usaynids, a Ceutan family of descendants of the Prophet who came to control the city's sh ur a. 65 Ceuta was not necessarily unusual. There is also evidence of the involvement of sh ur a councils in other city states in the Islamic West, though generally we know relatively little about them. 66 It is worth mentioning that, even when there was no mention of a sh ur a council, urban elite groups were able to engage in, and also sustain, various forms of collective action. It is hard to imagine that Ibn iAmm ar of Tripoli or Ibn Abī iAqīl of Tyre did not, even on the most informal basis, act in concert with other members of the urban elite in some way. The same is true for those Andalusī city states where religious elites were known to have played a leading role. In the 1240s, Orihuela came under the rule of the scholar Ibn iIs : am who, according to surviving correspondence, governed in consultation with other scholars who had fled to the city. 67 At around the same time, Seville was governed by prominent local families of judges and religious leaders, such as the al-B ajī and al-Jadd families. It is telling that a sh ur a was convened in the dramatic circumstances of 1246 when the q a : dī Ab u iAmr ibn al-Jadd, Seville's de facto leader for much of the 1230s and 1240s, was assassinated. 68 Elite-based collective organizations could take a whole range of different forms, but it is important to bear in mind that they were always just one of several players who had to coexist with, and were often undermined by, other forms of organization. Family structures were particularly important in all city states. This was not only because so many lordships were family enterprises, but because city states also opened up opportunities for other families to carve out their own political spheres and rival, or even actively challenge, governments, whether on their own or through larger factions. This process is well known for Italian cities, but it can also be observed in Islamic cities, especially once again in al-Andalus. 69 One factor that determined a family's success on this front was their ownership of land. This was especially the case for Italy, although even here it was not the only route for families to obtain political influence. A particularly clear-cut case is Piacenza where, as Pierre Racine has argued, major landed families were so powerful that they ultimately overshadowed other forms of political organization including the commune. Intense competition between landed families, loosely organized in factions, led to major changes of regime in Piacenza at ten-to twenty-year intervals between the 1220s and the city's absorption into the Milanese state in 1313. 70 A powerful player in these various upheavals was the Landi family. During the civil wars that divided Piacenza in the 1220s and 1230s, the Landi were temporarily allied with the city's popolo movement. After a coup in 1250, Guglielmo Landi established a signoria over Piacenza when he acted as the lieutenant of Uberto Pallavicino. After a revolt in 1267, the Landi were exiled, but they continued to lead a guerrilla war against Piacenza that lasted until the early 1300s. 71 The Landi were at least in part such powerful players because they owned about half the city's hinterland -especially the Val Taro, Val Trebbia, Val Tidone and Valle del Ceno -where they also had a considerable military infrastructure at their disposal. One of their castles, Bardi, was defended by 315 men and equipped with four catapults and mobile towers. Alongside their landed wealth, the Landi also relied on their external connections, especially to the emperor to whom they firmly clung in the convulsions of thirteenth-century Northern Italian politics. Ubertino Landi was allegedly related to Manfred, Emperor Frederick II's illegitimate son and effective king of Sicily, and was also made count of Venafro in 1257 with landed holdings in Molise. It is not difficult to see how the Landi could have been part of an imperial elite had Frederick II's project to revive the Empire in Northern Italy worked out, but in the city-state context of thirteenth-century Italy their main focus was really on Piacenza, a city in which they nonetheless never managed to permanently gain the upper hand. 72 The much under-researched topic of landownership by urban elites in Andalusī or Syrian cities is too big a topic in its own right to be tackled here. One should not expect the replication of Italian forms of landownership in the very different legal context of both these regions. However, the case of the Ban u Ashqil ula in Granada is a good example to illustrate how families were able to exploit their control of land in the political conflicts that plagued this city state from the start. 73 The Ashqil ula were a family of warriors who had initially been close allies of the Nas : rids, but had been relegated to the governorates of Málaga and Guadix, two subject cities of Granada. It is from these bases that they fought a civil war against their erstwhile allies between 1266 and the 1280s. 74  controlled the city's formidable system of fortifications, the Ashqil ula almost became lords in their own right and used the city's resources to conduct their own foreign policy against the Nas : rids. In 1266, Ab u Muh : ammad iAbd All ah, the head of the Ashqil ula, entered a pact with King Alfonso X of Castile who despatched one thousand Christian knights under the magnate Nuño González de Lara in support of the Ashqil ula, thereby forcing the sultan of Granada into a costly treaty with Alfonso to bind the latter to withdraw his support. 75 The conflict flared up again in the 1270s, when the Ashqil ula offered their submission to the Marīnid rulers of the Maghrib in return for military assistance against the Nas : rids. After the death of Ab u Muh : ammad, the Ashqil ula even surrendered Málaga to the Marīnids altogether in 1278 rather than see it ruled by the Nas : rids. They then continued their fight against the Nas : rids from their other base in Guadix, but were eventually forced into exile in 1288. 76 Given the sources available to us, we will probably never know the extent to which the Ashqil ula held land in Málaga or exploited the city's resources as the Nas : rids' disloyal military commanders, but it was undoubtedly their control of Málaga that made them such powerful, and disruptive, players in the politics of the city state.
As the importance of external connections for both the Landi and the Ashqil ula suggests, land was not the only resource that was available to families who established themselves in city states. Before they were able to consolidate their signoria, the della Scala of Verona were only minor landowners in the contado of Verona. They nevertheless managed to exploit the vacuum of power after the demise of Ezzelino da Romano and they built a new political coalition with the city's popolo that allowed them progressively to seize control of the city state's political affairs. In fact, Verona was almost a counter-image of Piacenza. Whereas in Piacenza there were particularly powerful landed families, the commune of Verona had acquired an unusual degree of control over the contado in the twelfth and thirteenth centuries and this 75 For this, see the Castilian chronicle Crónica del rey Don Alfonso décimo, in Crónicas de los reyes de Castilla desde Don Alfonso el Sabio, hasta los católicos Don Fernando y Doña Isabel, ed. Cayetano Rosell, 3 vols. (Madrid, 1875-1878), i, chs. xv-xvi. 76 Ibn Khald un, Kit ab al-iIbar, ii, 285-7, 291, 309-10; Ibn al-Khat : īb, Kit ab Aim al al-ail am, 331-5; Crónica del rey Don Alfonso décimo, chs. xlii-xlvi. meant that most families' landed holdings were generally much less substantial than in other Northern Italian cities. 77 In Islamic city states, too, a variety of resources underlay the power exercised by influential families. In fifteenth-century Granada, several major families exercised leverage over the fortunes of the city state because of the crucial role they played at the Nas : rid court. The Ban u Sarr aj (mythologized from the sixteenth century as the Abencerrajes), Kumasha and iAbd al-Barr were behind successful or attempted coups against Nas : rid sultans in 1419, 1429, 1432, 1445, 1455, 1464, 1470, 1482 and 1487. Unlike the Ashqil ula, these families rose to prominence during the reign of the Nas : rids and their power may have been closely tied to the fortunes of the Nas : rid state, since they were, for the most part, high-ranking officials in the city state's bureaucracy. In the context of the centrifugal tendencies of the city statethe brittle nature of the Nas : rid dynasty, combined with the complex geopolitics surrounding Granada -these families were able to act within a faction that turned out to be a decisive power-broker. 78 The city-state context also made it easier for families of religious professionals in the Islamic sphere of the Mediterranean world to control, or even monopolize, local resources than was arguably the case in cities that were part of larger states. Konrad Hirschler has argued for Hama in Syria that local families were able to exercise a greater control over q a : dī-ships and other important offices in the city while Hama was ruled semiautonomously by the Mu _ zaffarid dynasty in the late twelfth to mid fourteenth centuries. The al-B arizī family, for instance, were able to dominate the Sh afiiite judgeship, with one interruption, between 1219 and 1363. 79 This was also true to an extent for B urid Damascus where the city's princes initially attempted to install outsiders in the position of chief q a : dī in Damascus, but from 1125 the B urids made all appointments to this office from the al-Qurashī family, a powerful Damascene family that belonged to the locally predominant Sh afiiite school of Islamic law, and which continued to hold offices after the B urids' fall. 80 There is something of a parallel here with Italian city states where families were able to control ecclesiastical resources in their cities to a much greater degree than was the case in other European regions, although it is important to remember that, in many other respects, Islamic and Christian religious professionals were very unlike each other. 81 The political arena of city states also created the space for the development of political organizations that involved wider sections of the urban population, although it should be emphasized that our information on their precise social composition is often scant and that elites were themselves often implicated in such groupings. Again, this phenomenon is especially well known for thirteenth-century Italy where the rise of new social groups collectively known as popolo -especially merchants, artisans and recent migrants -prompted the formation of so-called popular societies (societates populi ) that often overlapped with or morphed into guilds, neighbourhood organizations and urban militia companies. Across Italy, popular organizations led major revolts against the cities' established elites and many signori came to power or consolidated their power in the wake of such uprisings. 82 Piacenza is again an interesting case because it had a much weaker popolo than the larger cities of Florence or Milan, and yet the popolo still played a key role in the city's convulsions. Piacenza's popolo was embroiled in bouts of civil war in the 1220s and 1230s and also led a revolt in 1250. The sources are somewhat unclear about how exactly the popolo was organized in Piacenza, but it is likely that, as elsewhere, the popolo developed around neighbourhood groups that were militarized and repeatedly engaged in violent confrontations. Piacenza's societas populi had its own structure of councils, elected representatives for negotiations with the city's elites and for a short period was even headed by its own podestà. It was also temporarily allied with the Landi family, although this alliance broke apart in 1236 over the Landi's ties with the emperor, which was in conflict with the popolo's traditionally pro-papal stance. 83 In 1250, after a 'great multitude of men armed for war' (magna multitudo virorum armatorum ad bellum) effectively took control of the city, the popolo proceeded to elect its representatives by city gate district. 84 A much more extreme case than Piacenza was neighbouring Cremona where popolobased networks had an almost centrifugal impact on urban political conflict. Confrontations in 1184, 1198, 1200 and 1209 between the societas populi and the city's established elites who controlled the commune escalated to such a degree that they split Cremona into two cities: the old city, which was controlled by the traditional commune, and the new city, which was effectively ruled by Cremona's societas populi. 85 It is notoriously difficult to find documentation on popular political organizations in Islamic cities, but city states are precisely the context in which we do have evidence of them. An interesting case is the city-based bands of predominantly young fighting men, known as ah : d ath, who appear in many Near and Middle Eastern cities during this period. In Damascus, the ah : d ath even became the city's de facto rulers for periods during the 970s and 980s, but they were also important political players under other regimes, including that of the B urids. 86 It would be wrong to draw too much of a direct parallel between the ah : d ath and armed groups in Italian cities, as has been suggested by Claude Cahen, but we are nevertheless faced here with evidence for city-based popular networks that also exercised considerable pressure on the B urid regime -a phenomenon that largely seems to disappear once Damascus became part of larger territorial configurations under the Zangids and Ayy ubids. 87 For the most part, the ah : d ath appear to have been men from lower social strata, although they were connected with the higher echelons of Damascene society in a variety of ways. Throughout the B urid period, they were led by a rajīs (literally 'leader') drawn from the Ban u al-S : ufī, a major family closely connected with the Sh afiiite religious establishment. As has been argued by Axel Havemann, the precise role of rajīs and ah : d ath varied over time and from place to place. Under the B urids, the rajīs was at least for some of the time the head of the B urids' civil administration (wazīr) and may also have acted as their chief urban official. 88 This is, of course, not the type of formalization of urban institutions that we find in Italian cities, but it is as close as we can get to the official recognition of these political players by B urid princes. In fact, many mentions of the ah : d ath in the principal Damascene chronicle of this period, Ibn al-Qal anisī's Dhayl tajrīkh Dimashq, also link them with activities that brought them into close association with the B urid princes, since they are frequently mentioned as forming part of the princes' entourage or as engaging with them in military activities against other Muslim princes or the Franks. 89 At the same time, the ah : d ath were also directly involved in urban conflict on a number of fronts. In 1129, the ah : d ath assisted the B urid prince T aj al-Mul uk in suppressing the B at : inī movement, a Shīiite sect better known as the Assassins, which had gained a foothold among the lower social strata of Damascus. The Assassins had allegedly been supported from within the B urid establishment, most notably by the vizier al-Mazdaq anī who was assassinated by the ah : d ath. Testimony to their power, al-Mazdaq anī was replaced by the rajīs of the ah : d ath. 90 The ah : d ath's actions could also be aimed at the B urid regime and even the prince himself. In 1149, the ah : d ath revolted against Prince Mujīr al-Dīn, opening the city's prisons, besieging the citadel and plundering the houses of B urid officials. The reasons for this revolt are unclear, but it occurred shortly after the death of the prince's influential guardian Anur and the affair was concluded when the rajīs received a robe of honour from the prince and was given greater powers. Five years later, when Damascus fell to the Zangids, the ah : d ath were alleged to have co-operated with the invaders. 91 It is worth noting that Ibn al-Qal anisī mentioned the ah : d ath as acting together with the 'riffraff' (ghawgh aj), 'mob' (awb ash) or simply the 'crowds' (iaw amm). Apart from indicating the author's barely disguised disapproval of these characters, this may suggest that the ah : d ath were operating in close co-operation with other urban groups and stood at the head of a wider urban coalition. 92 Evidence of popular political organizations in the city states of al-Andalus is relatively scarce, but for the final dramatic period of Granada's history there is information about the role of one of 89  Granada's outlying suburbs, the Albaicín, from a narrative account based on first-hand information about the city. In 1486, the Albaicín twice revolted in favour of Sultan Boabdil (Muh : ammad XI) who had been sidelined by his father and uncle. Boabdil, on this occasion also backed by Castile, supplied the Albaicín with weapons and, although both revolts were at first unsuccessful, they weakened the regime so much that it collapsed a year later and Boabdil returned. We have already seen with regard to Italy that organizations of the popolo could be linked to particular neighbourhoods, and it is indeed possible that behind mentions of the Albaicín we see the operation of a network involving men from lower social ranks. The Albaicín was known as a quarter that was predominantly inhabited by artisans, peasants and migrants, but this may not have excluded the participation of elite families -indeed, one of the great families closely associated with Boabdil, the iAbd al-Barr, also resided in this suburb. In the fourteenth century, the Syrian scholar al-iUmarī reported in his encyclopaedia that the Albaicín's inhabitants were militarized, although his claim that the suburb could supply thousands of warriors is almost certainly overblown. 93 A whole range of political organizations, therefore, crystallized in city states and fought over the urban political arena. It is important to emphasize that just like city-based lords, these political organizations were rarely insulated from the outside world. On the contrary, many of the organizations discussed in this section were closely connected with the multiplicity of players that populated the fragmented landscapes surrounding city states. In Piacenza, the Landi, the societas populi and others were connected with regional political coalitions that linked individual city-based players with protagonists outside the city walls -not just the emperor and pope, but also regional players 93  such as Uberto Pallavicino or Charles of Anjou to whose authority a popolo-based coalition in the city submitted in 1267. 94 In Granada, the Nas : rids, the Ban u Ashqil ula and other players entered shifting alliances with Castile and the Marīnids, but also the crown of Aragon and the H : afs : ids. Indeed, a powerful political force in fourteenth-century Granadan politics was Berber garrisons (known as ghuz at, border raiders, in the sources) who were political exiles from the Maghrib, but became major players in internal conflicts in Granada. One of their leaders, Ut : m an ibn Abī'l-iUl a (known as Don Uzmen in the Castilian sources), was so influential that he was rumoured to have been Granada's de facto ruler in the 1320s and marched on the city when Sultan Muh : ammad IV tried to oust him. 95 In this sense, city states were not only internally divided, but divisions within the city were deepened by the entanglement of urban political organizations in the complicated political landscape that surrounded city states.

IV
There was a seemingly endless variety of city states in the later medieval Mediterranean world. Stepping back from any of these highly localized realities is always difficult and any generalization necessarily has the effect of emphasizing certain aspects over others. Approaching Italian city states from a European perspective, for instance, has often led, directly or indirectly, to a particular focus on the constitutional nature of city-state governments or to their role in the development of territorial states. My objective in this article has been to approach this subject from a different angle and to look for certain traits that Italian city states shared with their counterparts in the Islamic regions of the Mediterranean world. A common feature that emerged from all these cases is just how volatile city states were as political constructs that were submerged in fragmented political landscapes and plagued by severe internal divisions. The volatility of city states is a well-known phenomenon, especially for Italy where historians have often used the label 'crisis of the commune' to talk about it. However, the comparative perspective here adopted allows us to recognize that this volatility should not be dismissed as something accidental or indeed as a crisis, but to recognize it as something that was in many ways part and parcel of the political order of city states.
City states in the three regions crystallized in the context of sustained levels of regional political fragmentation, which turned cities into a political space that a variety of political actors, both from within and outside the city, fought hard to control. What is important about this process is that city states did not emerge as unified political spaces, but that each city state was really a multiplicity of political spaces intersecting and often competing with one another. Across the three regions, the most prevalent form of political organization that developed in city states was city-based lordships, created either by important players from within urban society or by warlords who chose to establish themselves in a city. Christian and Islamic lords shared not only the aspiration to develop durable dynastic regimes, but also a high degree of uncertainty, as city-based lordships were often short-lived and beset by high levels of conflict. The political space of city states was also claimed by elite-based urban organizations, such as Italian communes, by families or factions with landed wealth or access to other kinds of resources and by popular political associations that could develop more freely in the particular structural context of city states. City states were also not cut off from the world around them. City-based lords occupied a fuzzy position between the politics of city states and that of other political formations. Furthermore, opposing parties within city states were often connected to divisions within the fragmented political landscape that surrounded cities, while external actors intervened in city-state politics and in turn reinforced divisions within the city. In many ways, city states stood at the intersection of high levels of fragmentation both outside and inside the city and were for that very reason volatile political constructs.
There were, of course, important differences between the various geographical areas under investigation here, but it is important to note that the divide between Islam and Christendom was only one variable, as differences between, as well as within, regions were also important. An obvious peculiarity of Italy vis-à-vis the Islamic world was communes and the particular documentary culture that was attached to them, but we have seen that this aspect should also not be overemphasized or seen as a reason not to compare Italy with other regions. What was also distinct about Italy was the longevity of the city state as a political system. With the exception of Granada, which was extraordinarily long-lived even by Italian standards, most of the Islamic city states here discussed lasted for around two to three generations. In Italy, city states often survived for longer periods, but this contrast should not be pushed too far as specific regimes ruling Italian city states were frequently just as short-lived as their Syrian or Andalusī counterparts. The difference may, in fact, not have as much to do with variations in the internal constitution of city states as with the different external contexts of the three regions. Islamic city states had to contend with major military invasions, as was already recognized by Ibn Khald un, and both Syria and al-Andalus were ultimately heavily exposed to large and expansive states, such as the Zangids/ Ayy ubids and Castile. Italy, too, faced its fair share of imperial military campaigns, and Charles of Anjou's offensive aimed at the south, but these external pressures were altogether more intermittent than was the case in the other regions. It is nevertheless important to emphasize that most Italian city states were exposed to intense geopolitical competition within Northern and Central Italy and were absorbed into larger regional states from the later thirteenth century onwards.
Syria and al-Andalus showed certain similarities, but not necessarily in ways that make it possible to generalize about the Islamic Mediterranean world. The role of religious professionals as city-based lords was an important feature of cities like Tripoli, Tyre and Ceuta, but this was not a universal pattern -in Damascus or Granada the protagonists were warlords. In some respects, al-Andalus and Italy shared important characteristics. Regionally based warlords played a critical role in both Andalusī and Italian city states. Although city states in the Islamic West did not know European-style colleges and councils, we have seen evidence of the activities of Ceuta's sh ur a as a collective institution of sorts acting alongside the city's lords. Granada in particular shared many similarities with Italian city states in terms of its longevity and the role of families, factions and neighbourhoods in the city state's convulsions.
In spite of all these variations, Italy, Syria and al-Andalus also shared important characteristics, which is also why they gave rise to city states that had many features in common. Each region was characterized by dense patterns of urbanization that manifested themselves not only in a clustering of cities, but also in the presence of middling and large cities, which had a number of components that played a critical role in the evolution of citybased forms of political organization: substantial city-based elites, powerful families that were attracted to cities and large urban populations that could be organized in popular, often neighbourhood-based, political associations. Another crucial factor was that in each of these regions city states emerged from the breakdown of large and sophisticated imperial structures of government, which produced complex and multi-layered political landscapes. A clear manifestation of this phenomenon was that city states were often ruled by warlords who had either been associated or sought association with larger states, while at the same time also choosing specific cities as the principal basis of their political operations.
Northern and Central Italy, Syria and al-Andalus were, of course, not the only regions where city states crystallized. The combination of high levels of urbanization and extreme political fragmentation could have similar effects elsewhere in Europe or the Islamic world. In the Mediterranean world itself, Tripoli in Ifrīqiya or Bari in Apulia shared many characteristics of the cases discussed in this article, as did cities in the south of France. 96 Not all Mediterranean regions gave rise to city states, however. Egypt never witnessed a comparable breakdown in its political structures and survived a series of political transitions -from the F at : imid to the Ayy ubid to the Maml uk state -without breaking up into smaller units. Cairo was one of the medieval Mediterranean world's largest cities, but it was also a capital city, where the Egyptian state's powerful fiscal-military apparatus was based, and it is difficult to see how it could ever have developed into the sort of city state that one would encounter in Syria. Although fiercely fought over in periods of civil war, Cairo could not, and did not, really break away from the state apparatus to which it was so closely linked. 97 A Mediterranean region that did see a large-scale disintegration of existing political structures was the Aegean world after the sack of Constantinople in 1204, but during this period this part of the Mediterranean had altogether fewer and less populous cities than the regions analysed here. The predominant forms of political organization that emerged after 1204 were Frankish rural lordships or small regional states that modelled themselves on the Byzantine state whose tenacity as a blueprint for government may also have superseded that of the Western Empire in Italy. However, there were also cities resembling city states, such as Thessaloniki, by far the largest city after Constantinople, which for short spells was de facto ruled by independent-minded governors or, between 1342 and 1348, by an urban rebel coalition known as the Zealots. 98 There are important consequences here for our way of thinking about city states and indeed states more broadly. First, there is no necessary link between city states and particular constitutional arrangements, such as 'democratic' forms of government. The medieval city state has often been seen as a stepping stone on the high road to Western democracy, but we have seen that collective forms of government were only one part of the picture and they may not have been exclusive to the region that we now consider as the 'West' anyway. 99 Second, it is also problematic to treat the city states studied here in any straightforward way as part of grand narratives about state formation. In such narratives, city states were often necessarily seen as 'imperfect' states, precisely because in many respects they did not fit our expectations of what a state should look like. After all, the governments of later medieval city states neither enjoyed the external independence nor the internal sovereignty that students of state formation look for. Of course, it is possible to recognize some of these traits in particular city states, but the risk is that city states are straitjacketed into something they were not. It is easy to forget that later medieval city states emerged in the context of state de-formation and were not miniature replications of larger states, but gave rise to a political system sui generis that was inextricably bound up with, and itself reflected, the fragmented political landscapes that surrounded them. That city states were such brittle constructs does not mean that they were anarchical places devoid of any order -but it was not a government-centred political order that fits neatly into the history of state formation. 100 Though it is beyond the scope of this article, it is also worth asking how far city-based political orders outside this period are best understood through the lens of the territorial state. Some scholars of archaic and classical Greek city states -the other emblematic type of Mediterranean city state -have also expressed a degree of dissatisfaction with the existing paradigm of the city state. Recent scholarship has emphasized how ancient Greek city states were characterized by multiple forms of political organization (including tyrannies), high levels of internal conflict and interdependence and interpenetration. 101 It remains to be asked how useful it is, in light of all this, to transplant the concept to describe the citycentred nature of political formations in other regions of the world. 102 These findings also raise important questions for our way of thinking about the wider Mediterranean world in the later Middle Ages and particularly the confessional divide between Christendom and Islam. Many a narrative about the distinctness of historical trajectories in the Islamic world has rested on the distinct nature of the 'Islamic' vis-à-vis the 'European' city. The supposed absence of urban autonomy in the Islamic world has, for instance, been used to explain why capitalism did not develop there or why large-scale empires, rather than territorial states and eventually nation states, characterized the Islamic world. 103 If we accept that 'Islamic' and 'Christian' cities were less far apart than is often thought, it may be necessary to rethink such grand contrasts. The degree to which structures and processes were shared across the Islamic-Christian divide has sometimes been underappreciated, also because, as far as the Middle Ages are concerned, the scholarship has tended to focus on studying relations between Muslims and Christians rather than on comparisons. 104 The later Middle Ages are sometimes called the 'Age of the Crusades' in the Mediterranean world for that reason, but it is easy to forget that both Christian and Islamic political formations also faced another shared, and much less-studied, challenge during this period -the disaggregation and sometimes collapse of larger political formations that, in the most urbanized regions of the Mediterranean world, manifested itself precisely in the city states that have been the subject of this article. As renegade military commanders, families and various other urban groups busily sought to mould this political system in the highly fractured and urbanized landscapes of Italy, Syria and al-Andalus, they experienced all the dangers, and exploited all the potential, that characterized the world of Ibn Khald un and his Mediterranean contemporaries.

University College London
Patrick Lantschner