• 1 The first modern notes dedicated to Sijilmasa were written in the community of French military offi (...)
• 2 The MAPS (Moroccan-American Project and Sijilmasa) is an ambitious Moroccan–American archaeological (...)
• 3 Sijilmassa: ville, oasis, carrefour is a Franco-Moroccan archaeological program directed by Françoi (...)

1Sijilmasa is the large medieval oasis city located on the borders of present-day Algeria and Morocco (Tafilalt province, Morocco) that is today unanimously considered by historians as the most important caravan city of northern Sahara for the 9th–14th-century period. The broad outline of its political and economic history is known to this day thanks to a dozen textual references collected in medieval Arabic written sources that have now been extensively analyzed by several generations of historians.1 The archaeological site is a large tell, located in the heart of the Tafilalt palm grove in the Saharan plain of the same name, surrounded by dozens of satellite sites, most of which have yet to be identified. Two major archaeological missions followed one another, first in the 1990s,2 then in the 2010s;3 however, despite these long-term, rigorously conducted, and widely published excavation campaigns, the materiality of this major site of medieval Saharan history remains poorly documented, because only an infinitesimal portion of the site has been explored to date. Of the main archaeological site of nearly 380 ha—of which 130 ha represent the heart of the urban fabric (Figure 1)—barely 700 sq. m have thus far been extensively excavated. As a consequence, if the curtain is beginning to be raised on the architectural patterns, on the typology of artifacts, on the chronology and temporalities of the site’s development, many other fields and topics remain under-documented—such as the overall morphology of the city, the spatial frames and dynamics of urbanization, the demographic and social organization of the settlement, the artistic traditions of the site, the health status of the populations, and the relation of the town to the surrounding oasis and desert environment.

• 4 C. Capel, 2016a; C. Capel, forthcoming.
• 5 J. Obertreis et al., 2016.

2In the absence of long-term and extensive excavations, which require broad technical skills, material resources, and funding commensurate with the exceptional size of this site, it is necessary to implement alternative modes of archaeological investigation in order to circumvent these obstacles and continue to expand our knowledge of this major site in African and Mediterranean history. This strategy was followed in the framework of a PhD project entirely dedicated to the history and archaeology of Sijilmasa4 and where a cross-disciplinary investigation, combining environmental archaeology and hydrosociology, was tested. The first aim of this investigation, based on the observation of the current irrigation networks of Sijilmasa palm grove, was to explain the modes and systems of water exploitation of this oasis, to reconstruct the conditions of their implementation, to ascertain their antiquity, and to evaluate their evolution over time—in a word: to historicize access to water in this region. The second objective was to cross-check these elements of hydraulic history with the contributions of Political Ecology, developed by sociologists and geographers who highlight the close relationship that exists in contemporary societies between political and social stakes and environment-shaping, particularly in water landscapes.5

• 6 “Hydro-social research envisions the circulation of water as a combined physical and social process (...)
• 7 E. Swyngedouw, 2009.
• 8 K. Wittfogel, 1957.
• 9 C. Bichsel, 2016.

3Recently—that is, since the mid-2000s—the term ‘hydrosociology’, or ‘social hydrology’, has been largely adopted in the field of Political Ecology, notably following the clear definition6 given by the British geographer E. Swyngedouw (Manchester University) based on previous thoughts and current paradigmatic dynamics related to contemporary water management studies.7 In fact, this concept of social hydrology has been progressively built up throughout the world over the last fifty years, mainly among anthropologists. One of the pioneers, as early as the 1950s, was the German-American sinologist Karl Wittfogel, who conducted a political—and autocratic—reading of the great hydraulic laying out of the planet, from ancient Mesopotamia to the Stalinist USSR.8 Even though the conclusions of his work have been questioned for a long time, his theory of the existence of a close interdependence between political structures, water resource management, and hydraulic technologies has been widely adopted and, since the 1960s, considerably amended by a multiplication of case studies throughout the world.9

• 10 As emblematic examples, see J. Brunhes, 1902; P. Penet, 1913; D. Moulias, 1927; J. Lethielleux, 194 (...)
• 11 See for example the following selection: P. Roche, 1965; L. Eldblom, 1968; G. Grandguillaume 1973; (...)
• 12 E. Swyngedouw, 2009.
• 13 P. Pascon, 1974.
• 14 P. Pascon, 1977.
• 15 P. Pascon, 1983, I-84-98.
• 16 P. Pascon, 1983, I-84-85.
• 17 P. Pascon, 1979.

4In North Africa, the cultural and environmental framework for our analysis, many geographers and to a lesser extent anthropologists have been studying, for about a century, the procedures, techniques, and social organization of contemporary irrigation in these arid or desert surroundings. Globally, geographers and agronomists have mainly addressed the environmental (hydrological resources), technical (irrigation amenities), agricultural (crop spectra and cycles), and land use planning (plot morphology) issues of hydraulic practices.10 Alongside this abundant literature, anthropologists and specialists in Islamic law have also taken up the topic of irrigation but have been focusing more on the social issues (rules and rights of use, sharing of the resource, water ownership, conflicts and arbitration) of water management.11 At the crossroads of these two approaches, only a few researchers working on contemporary worlds have explicitly considered the impact of social organizations on the terms of water use and on the shaping of hydraulic landscapes, as well as the co-evolution of social patterns and hydraulic patterns—that is, two key concepts of hydrosociolgy.12 For the western Maghreb region, one of these leading theorists of social hydrology was Paul Pascon (1932–1985). A native of Morocco, trained in biology and sociology, he was entrusted by the Moroccan administration with various sociological survey missions, notably at the Office National d’Irrigation (ONI) between 1964 and 1966 (i.e. in the midst of post-colonial restructuring). This position offered him the opportunity to study in depth not only the hydraulic but also the social organizations of the Marrakech plain (the Haouz Plain)—and to multiply there social experiments that today would be qualified as development initiatives. This extensive fieldwork in the Haouz of Marrakech—an arid alluvial plain intensively irrigated for at least a millennium—would give its name to the masterwork that, 15 years later, he drew from his PhD in history.13  His academic thesis is dedicated to the question of the social organization of the region and to its transformations faced with the successive assertion of the sultanian central state, of local chieftaincies, and of capitalist colonization.14 This broad historical survey, which combines sociology, geography, and hydrology, is based on an original reading scale that locates the issues of water management at the heart of the analysis and leads to a global and diachronic perspective on the foundation, antagonisms, and transformations of this Moroccan society, seen through its irrigation uses. In this book, Paul Pascon demonstrates that the diversity of hydraulic uses and techniques of the Haouz Plain reflects the various social and political organizations that followed each other in this area. In other words, water management, whether it is collective or individual, concerted or imposed, fair or unequal, says a lot about the social frame and effective political powers of this region of Morocco, from colonial to contemporary times.15 Pascon has thus done pioneering work in renewing the study of hydraulic systems, which were until then mainly considered as technical objects, only for the domain of engineers. By introducing human sciences here, he defined irrigation systems as elements of human geography.16 Furthermore, he is even among the first to define this field of research, as he named it the “psychosociology of irrigation”.17 Since the 1970s, this social and political approach to hydraulic practices has become a separate part of the social sciences, until it became, in the 21st century, a full field of research under the now widely accepted name of ‘hydrosociology’.

• 18 P. Gauckler, 1897–1901; J. Coignet, 1925; J. Birebent, 1964; B.D. Shaw, 1982; P. Trouset, 1986; A. (...)
• 19 T. Madani, 2006; R. González Villaescusa, P. Cressier, 2011.
• 20 M. Barceló, 1989; H. Kirchner, C. Navarro, 1994; M. Barceló, 1996; P. Ballesteros Arias et al., 201 (...)
• 21 M. Barceló, 1989, XXV.
• 22 P. Cressier, 2006.

5Hydraulics has also been a major topic among historians and archaeologists, especially among specialists in ancient Roman society, which has remained particularly famous in history for its water engineering. Thus, not surprisingly, the main archaeological studies on Saharan hydraulics have concerned settlements from Antiquity, sometimes building bridges with contemporary oasis practices for the purpose of ethno-archaeological restitution.18 On the contrary, in the field of medieval or modern hydraulics, Saharan studies are still rare.19 As a consequence, it is thanks to works and advances led in other geographical areas that hydraulic archaeology came to be a fully fledged research field. Thus, in the course of the 1980s, in Spain, a movement of theorizing the discipline emerged among historians who were involved in the topic of agricultural landscape evolution of the Iberian Peninsula with the arrival of Islam. Miguel Barceló (Universitat Autònoma de Barcelona, Spain), in particular—but also Pierre Guichard (Université Lyon Lumière, France), Helena Kirchner (Universitat Autònoma de Barcelona), and André Bazzana (CNRS, France)—defined methods of investigation, created conceptual tools of analysis, and developed theoretical models, which allowed researchers to considerably rationalize the approach of the archaeologists.20 As an example, and because the present paper is based on this concept, the theorization of what archaeologists now call ‘hydraulic permanence’ represents a fundamental contribution to the understanding of ancient and current hydraulic landscapes. This ‘hydraulic permanence’ is the process according to which an efficient hydraulic system is perpetually maintained in position over decades and even centuries as long as its purpose does not disappear, while being rebuilt on multiple occasions and according to different techniques to extend its existence. This phenomenon leads to and explains the accumulation of hydraulic ruins from different periods in the same place and the frequent appearance of archaeological hydraulic palimpsests in contemporary agricultural landscapes, which allows one to go back in time by reading ancient, even disappeared, hydraulic practices.21 Moreover, some archaeologists of this Iberian generation made great steps forward in the historicization of hydraulic systems by taking over the theories from social hydrology to apply them to archaeological remains. Among them, Patrice Cressier (CNRS, France), who is active on both shores of the Mediterranean—that is to say, between Morocco and Spain—was aware of the Moroccan work of Paul Pascon and exploited it as a reading grid of North African artifacts. In this way, more than just understanding ancient systems from a technical and chronological point of view, he proposed to reveal the social and political systems linked to these ancient modes of water management.22 A window was thus opened on the population organizations that usually constitute a field difficult to grasp, in particular within societies without writing.

• 23 Undoubtedly, there are more hydraulic phases in Tafilalt history and therefore there are certainly (...)

6This method was applied to a Saharan oasis case study: the palm grove of Tafilalt, closely associated with the history of the caravan city of Sijilmasa, which has now disappeared. The analysis of the sub-contemporary hydraulic system of Tafilalt revealed that its implementation was the result of a significant transformation of irrigation methods over time, from the origins of Sijilmasa to the present day. This system consists of three main elements: garden and domestic wells (about a thousand in 1950) that exploit local groundwater; sets of several dozen khettâra-s (also called foggara-s in other Saharan regions)—that is, underground tunnels that mainly tap the aquifers of the surrounding mountainous foothills to drain them to the outskirts of the palm grove; and the surface seguia-s (open channels) connected to the Wadi Ziz (the wadi that runs across the entire palm grove), which exploit surface water, particularly at the time of spring and autumn swellings. The study of these different systems has led to the historicization and dating of their use, which seems to be divided into three major hydraulic phases; moreover, these three phases correspond to three distinct hydraulic practices that themselves most likely reflect three different modes of social organization that have succeeded each other over time.23 The chronological division is as follows: medieval period (9th–14th centuries); post-medieval period (15th–16th centuries); and modern period (17th–19th centuries).

• 24 Likely etymological meaning of the Berber toponym Sijilmasa (L. Mezzine, 1984, p. 24).
• 25 Ibn Hawqal, Surat, p. 99-100; Ibn Hawqal, Configuration, p. 97-98; Al-Bakrî, Masâlik, § 148; Al-Bak (...)
• 26 Al-Ya‘qûbî, Buldân, p. 358-360; Al-Ya‘qûbî, Pays, p. 225-226; Al-Dimashqî, Cosmographie, p. 90, 238 (...)

7Since its first references in medieval Arabic sources, all historians and geographers agree that Sijilmasa developed in a hostile and bleak desert context. However, in the works of Ibn Hawqal and al-Bakrî—the two medieval authors with the best information on the city, the former having himself visited it during the 10th century, and the latter having used first-hand accounts a century later—this negative image of the Tafilalt environment is counterbalanced by the description of the wealthy hydraulic context of the site.25 These two authors agree in locating Sijilmasa on the banks of a particularly generous wadi, which they do not name, however—and nor do most ancient authors, with the exception of al-Ya’qubî (9th century), who was the first to mention the toponym of Wadî Zîz, and then, much later, al-Dimashqî (14th century) and Leo Africanus (16th century).26 Ibn Hawqal also accurately describes the specific rate of flow of this wadi, which bursts its banks annually like the Nile River and then largely floods the entire region of Sijilmasa, creating exceptional conditions for the development of agriculture. Not surprisingly in a desert context, water and water management are thus key parameters for understanding the history of Sijilmasa.

• 27 Although declining and reducing, the palm grove of Tafilalt is still today the largest oasis area o (...)
• 28 C. Capel, 2019, p. 247-253.

8The archaeological study of the entire region bears out and clarifies this initial perception. Indeed, if the city of Sijilmasa—settled in a sedimentary basin located on the northern fringe of the Sahara—was able to develop in the middle of a desert environment, it was primarily due to the rise of its oasis, where dense irrigated agriculture was implemented. The palm grove of Tafilalt, which still exists today, first led to ensure the food autonomy of the region but also, and above all, to make Sijilmasa a supply stopover on Saharan routes that could not be overlooked, thanks to its surplus production and its obvious large size.27 Although this aspect has been underestimated for a long time by historical research, it is indeed probably because of its agricultural value and of its enormous production capacity, both in foodstuffs (dates, cereals) and in fodder for pack animals, that Tafilalt owes its overwhelming position as a caravan crossroads and its longevity as such throughout the medieval period.28 As a result, a large proportion of the inhabitants of Tafilalt, including those living in the urban area of Sijilmasa, must have been involved in agricultural activities, which may have led to Sijilmasa being characterized more as an agro-city than as a caravan city.

• 29 For the geomorphological, archaeological, and historical demonstration of the artificial nature of (...)

9Now, in this desert context, this major agricultural hub only owes its existence to the implementation of a dense and massive irrigation system and to expert water management; these are primarily based on the exploitation of surface resources provided by the Ziz and the Gheris wadis that run cross the Tafilalt plain, where the palm grove of the same name is nestled (Figure 2). But of these two wadis, it is in fact the Ziz—and more specially its arm known as the ‘Filali’ (‘of Tafilalt’) Ziz alongside of Sijilmasa—that has been mainly exploited since medieval times; indeed, its morphological characteristics are quite different from those of the wadi Gheris and make it particularly suitable for agricultural use. And if this wadi is so well adapted to the needs of agriculture, it is in fact because the Filali Ziz is an anthropic construction—in other words, it is a canal.29

• 30 The pluvionival rate of flow of the Ziz and the Gheris is marked by two major annual floods: the sp (...)
• 31 J. Margat, 1954, p. 192; J. Margat, 1958, p. 19.

10In Saharan contexts, oases are anthropic constructions intended to ensure the food subsistence of the desert populations. Thus, the main purpose of a canal is agricultural irrigation—and, in fact, as medieval texts and sub-contemporary practices show, the role of the Filali Ziz is mainly related to this agricultural perspective. The Ziz is thus nothing more than an irrigation seguia, but with the proportions of a river (45 km long and 20–30 m wide). This unusual size of the hydraulic layout has led here to the implementation an original irrigation technique—likely unique in North Africa—which consists in flooding the plots of cultivation by a massive and widespread water submersion up to 1 m high. Thus, twice a year, during the two annual natural floods of Wadi Ziz, the Tafilalt palm grove is completely submerged under nutritive water and converts into a patchwork of countless basins, limited by high earthen embankments.30 This method is close to flood irrigation management used on the banks of the Nile and the Senegal rivers. Then, after evaporation and deposition of nutritious silt on the plots, the latter were cultivated thanks to plant species with a fast maturation cycle (about 3 months) that did not require further irrigation before harvest time.31 Still in use until the end of the 1960s, this widespread flood irrigation system was documented in the 1950s and echoes the descriptions in ancient texts that rightly compare the Ziz Valley and the Nile Valley. This irrigation technique has now disappeared with the construction of the Hassan al-Dhâkil dam, 100 km upstream from Sijilmasa, which came into operation in 1973 and now protects the lower Ziz Valley from its massive (and sometimes destructive) floods and replaces them with a moderate but regular water supply, which allows for year-round irrigation similar to that adopted in all other Moroccan palm groves.

• 32 As Tafilalt resisted French colonization until 1933 and as WWII put an end to most French laying-ou (...)

11Despite its replacement fifty years ago by water supply systems directly connected to the Hassan ad-Dâkhil dam, the ancient irrigation network of the Tafilalt oasis has been largely preserved in the present-day Filalian landscape, mainly because the contemporary system makes use of a large part of this ancient network—which was perfectly well thought out and adapted to the site’s organization—by joining up with it. It is thus possible to map the network of the main seguia-s that were in use in modern times in Tafilalt and thus to retrieve a global irrigation organization scheme for the Alawite period.32 According to this pre-colonial system, the Filali Ziz included 13 diversion dams, spread out along its course (Figure 3).

• 33 P. Pascon, 1983, I-84-98.
• 34 One might prefer the expression ‘clanic’ or ‘kinship’ structuring.
• 35 P. Cressier, 2006, p. 44.

12The first observation is that the majority (11 of 13) of these dams—still preserved as ruins today—gave rise to a single seguia, which alternately irrigated the right and left banks of the Ziz. The system thus adopted an overall herringbone pattern. Moreover, the water intakes were generally spaced out so that the seguia-s did not extend downstream from the latitude of the zones watered by the next dam. This rigorous organization of the agricultural zone, made up of independent irrigated territories, led to the coexistence of non-competitive hydraulic unities. This kind of hydraulic organization is very similar to that of the Haouz of Marrakech, studied by Paul Pascon.33 Despite different local forms, the common morphological characteristics of these two systems must be related to the “tribal”34 structuring of societies. And according to Pascon, they may physically reflect a fair plot allotment between neighbors, associated with equivalent water exploitation rights and based on the number and variety of social groups composing the full population of the oasis.35

• 36 P. Cressier, 2004, p. 281-282; R. González Villaescusa, P. Cressier, 2011, p. 27-29.
• 37 M. Barceló, 1989, XXV.
• 38 The overall organizational pattern of the sub-contemporary hydraulic network of Tafilalt is suppose (...)

13This system could be ancient and date back to medieval times. Indeed, this is how the set of fossilized seguia-s has been interpreted that shaped the agricultural land of the Saharan medieval town of Tâmdult (Morocco), now abandoned. At the foot of the urban archaeological site, whose main occupation seems to date back to the 12th century, Patrice Cressier (CNRS, France) and Ricardo González Villaescusa (Université de Nice, France) have identified a series of four to six large and parallel seguia-s that creates a mesh of narrow but very elongated strips of land. This orthonormal pattern has no apparent technical or topographical reason here and is therefore more likely to be due to social factors. In fact, it defines agrarian units that are hydraulically independent of each other and do not compete with each other. This ensures an equitable supply of water and plots and thus thwarts possible conflicts between the populations forming the urban community of Tâmdult.36 Similarly, in the case of Tafilalt as in that of Haouz, the water intakes follow one another, spaced out so as not to compete with each other. In addition, each dam gives birth to only one seguia, once again in order to limit any supply conflict. As a consequence, the agrarian territories are hydraulically independent so as not to take advantage of undue resources. In the same way, the upstream territories are generally better watered but also smaller—as they are limited in their extension by the existence of their neighbors—than the downstream lands, which can extend and in fact are generally much larger, but which benefit from less dampness and therefore experience a lower yield. In this way, the people implement a collective system of water management that considers the resource—the Wadi Ziz—as a common good and is based on a fair and collective supply in order to make possible the coexistence of several (and possibly antagonistic) social groups in a horizontal organization of forces. Now, because of the general stability (permanence) of hydraulic systems over time that has been demonstrated by archaeological research,37 this balanced and egalitarian pattern of the Filalian network in modern times could date back to medieval times and thus be an indicator of the organization of the medieval territory of Sijilmasa.38

• 39 For a dating of Wadi Ziz, see C. Capel, 2016b, p. 151-155 and C. Capel, 2018, p. 539-541.

14The digging of the Filali Ziz must have taken place from the time of the founding of Sijilmasa, notably because the city has been settled on its banks.39 However, it will probably remain impossible to determine whether this decision—which means the launch of a massive investment in agriculture and thus the planning of a deep change in the local subsistence economy—was made by the political leaders (the Midrarid emirs, 772–976) of that time alone, or whether it was the result of a broader collective initiative involving tribal chiefs, pastoral communities, religious representatives, or even landowners. In any case, this process of agricultural colonization sheds light on the Midrarid chronology and on the origins of the city of Sijilmasa, as reported by al-Bakrî. Indeed, in this text, al-Yasa’, the fourth ruler of Sijilmasa (790–824) is reported to have conducted a tribal allotment of the city after he ordered the town to be walled. The re-reading of these few sentences leads us to understand this decision not only as an allotment of the urban districts of Sijilmasa following the founding of the city, as has been usually proposed by historians, but also as an ‘allotment’ (or a subdivision) of water resources—that is, as the definition of the egalitarian system of the Filali Ziz irrigation network. This fair planning of the territory would also fit quite well with the historical vulgate drawn from the text of al-Bakrî, which claims that the founding of Sijilmasa was driven by the principles of the Kharejite doctrine, supposed to be of an egalitarian social conception.

• 40 Concerning the foundation of the city of Sijilmasa and its different phases of construction, see ou (...)

15The cross-checking of these textual and historical elements leads us to reconsider the origin of Sijilmasa and, in particular, the organization of its settlement during its first centuries of existence. As mentioned previously, the digging of the Ziz—which must be contemporaneous with the founding of the city40—and the concomitant design of the secondary irrigation network convey the notion of a collective mode of water management where the hydraulic resource is considered as a common good and a shared capital. This original consensus about water management indicates a strong social agreement between all the actors involved in this project. It thus means that the Sijilmasa of the early Middle Ages developed as a single political entity, mostly united, in spite of very likely internal conflicts. In other words, Sijilmasa must have considered itself as a societal unit, as it formed a single settlement and agreed that it shared and managed collectively a finite territory (i.e. the palm grove), and as it created the possibilities of an integrating agglomeration of people in the form of an agro-city. This feeling of a collective belonging undoubtedly prevailed over internal dissensions for many centuries, despite the turmoil of history (e.g. refuge for political dissidents, independence splits), despite external struggles for influence over the region (e.g. the successive Fatimid, Ummayyad, Almoravid, Almohad, and Abdelwadid conquests), and despite demographic and social difficulties (e.g. Almohad purges, destructive floods, the Black Death).

• 41 Date of the last restoration of the original Ziz canal diversion dam, obtained by radiocarbon analy (...)

17The study of the area of the head of the Ziz canal, located 20 km upstream from Sijilmasa and near the present city of Erfoud, showed that, later than the 14th century,41 a violent flood utterly diverted the course of the natural wadi. About 1 km upstream from the Filali Ziz diversion dam, the river turned away from its course and bypassed the original hydraulic system to return directly to its natural bed (called Amerbouh in this sector), depriving the irrigation canal—and by extension the entire palm grove—of its water. This event was undoubtedly experienced as a disaster, calling into question the survival of the inhabitants of Sijilmasa and therefore making it necessary to rebuild hydraulic diversion structures to keep the canal viable. These facilities were relocated about 800 m south of the original dam, to a site that was maintained and rebuilt until the 1950s, when the last renovations were carried out before the Hassan ad-Dâkhil dam was put into operation (Figure 4).

• 42 J. Margat et al., 1962, p. 204.
• 43 P.W. English, 1968, p. 177.
• 44 Mentioned by J. Margat et al., 1962, p. 204.
• 45 H. Goblot, 1979, p. 153.
• 46 D. Lightfoot, J. Miller, 1996, p. 98; D. Lightfoot, 1996, p. 266-267.
• 47 With the exception of D. Lightfoot, 1996, p. 267.
• 48 J. Margat et al., 1962, p. 204.

18Now, some evidence indicates that this reconstruction may not have been immediate and that several years, or even several decades, may have preceded the re-watering of the artificial channel. This thesis is suggested by the existence in Tafilalt of a spectacular hydraulic layout: the khettâra-s (underground drainage galleries). Their large number (about 90), their density, and their great extent (some measure about 20 km, for a total of about 300 km of galleries)42 make it a set among the most spectacular in Morocco, comparable to the largest Iranian systems.43 However, to this day, these structures have remained poorly studied and very uncertainly dated. Emile Gauthier believes that they were introduced in the early Middle Ages by the Miknâsa tribe (the founders of Sijilmasa), who brought Roman technology with them on their migration south.44 According to Henri Goblot, the Almoravids (the founders of Marrakech in the 11th century) must be credited with their construction, as they are reputed to have provided their own capital with a vast network of water supply through underground drains;45 and James Miller and Dale Lightfoot associate their implementation with the Merinid period.46 In addition to not being based on any textual or archaeological evidence,47 these hypotheses remain weak because they assume that the entire network was built in one piece, on a single occasion, and there is no evidence to support this notion. Only the geologist Jean Margat, with the help of oral memory, suggests a more careful hypothesis by staggering these constructions between the 16th and 18th centuries.48

• 49 Section located at the level of the ford/bridge of the Alnif road, near the qsar Oulad Saidan, on t (...)

21This archaeological analysis seems to be confirmed by the pedological observations initiated in 1992 by Tony Wilkinson (University of Durham, UK) and completed by ourselves in 2013. A stratigraphic section of nearly 10 m high of the Tafilalt palm grove could be found in the riverbed of the Gheris, in a location where the river bank was deeply cut by hydrological activities.49 Over the natural soil formations of ancient origin, which developed nearly 5 m high (strata D to J in the figure), several recent layers of anthropic origin can be identified that are over about 3 m high (strata A to C and unnumbered strata) (Figure 6).

• 50 Léon l’Africain, Description, I-425-430.
• 51 M. Kably, 2011, p. 300-313; D. Jacques-Meunié, 1982, p. 594-596.
• 52 Léon l’Africain, Description, I-427-431.
• 53 B. Rosenberger, 1977.

25Thus, at the end of the medieval period, a significant evolution in the social organization emerged: the water was no longer considered as a common good that people exploited collectively and in (forced?) dialogue, but as a resource to be individually capitalized, within a logic of privatizing a good in a context that had become competitive. This relationship with water undoubtedly reflects a similar evolution in the Filalian society, turning to social patterns where solidarity was no longer the structuring link between the inhabitants of the palm grove, but where, with a backdrop of rivalry for water supply, competition redefined social and therefore political organizations. However, these events occurred precisely at a time—after the 14th century—when, according to the texts, Sijilmasa was experiencing a period of deep destabilization, marked by a crisis in its urban model (= city exodus),50 by significant political disorganization (= multiplication of tribal, religious, and state arguing over control of Tafilalt),51 and by possible social weakening (= competition between tribes,52 plague epidemics).53 Far from being unique, the cause of these troubles cannot be summed up in the difficulties identified in the water management, which are perhaps only a symptom of this crisis. But these hydraulic observations alone shed light on why Sijilmasa could not maintain itself as a coherent settlement unit, as an integrating agro-city, and as a homogeneous and indivisible urban model. The problem of water supply underlines the fact that at the end of the medieval era, Tafilalt was experiencing an increasing tendency towards internal political division, made up of family, clan, spatial, or even dogmatic rivalries that were undoubtedly reflected in the hydraulic competition. Thus, perhaps Sijilmasa and its oasis were beginning to be experienced no longer as a unified space and as a single social entity, but as a territory to be divided among several human groups with divergent interests. Thus, the rehabilitation of the Ziz, which represents a large-scale collective and planned effort, could have been permanently hindered by a lack of collective will. At the same time, the single agro-city model was breaking down, giving way to a physical spread of the population within the oasis and to a political fragmentation of its inhabitants. The city of Sijilmasa, which had become an unsustainable social model, was depopulated, while the organizational logic of the Filalian settlement was now recomposed into independent and rival units: the era of the qsûr was initiated.

• 54 T. Madani, 2005, p. 70.
• 55 Al-Ifranî, Nuzhat, I-292, II-484; H. Terrasse, 1960; C. Pellat, 1965; D. Jacques-Meunié, 1982, p. 6 (...)
• 56 Since it does not seem to exist in western Maghreb before the 12th century, the khettâra technique (...)
• 57 G. Grandguillaume, 1978, especially p. 129-131.

26While the Ziz canal underwent a period of drying up during the 15th and 16th centuries, which led to a problematic access to water and to the populations of Tafilalt renewing their hydraulic, demographic, and political organizations, one sector of the palm grove stood out from the others. This area is located immediately to the northeast and east of Sijilmasa and is now divided between two modern administrative districts, called Tanijiout (to the north) and Ifli (to the south). As mentioned above, these are precisely the only two sectors of the inner Tafilalt oasis to be irrigated by khettâra-s connected to the Ziz canal waterbed, leading underground water to be piped in from nearly 10 km away. If none of these khettâra-s is still in use today, the toponymy preserves the memory of this ancient hydraulic pattern since iflî is a Berber word equivalent to the khettâra of Moroccan Arabic (pl. iflân). This expression has become uncommon in most parts of northern Sahara, but it is still commonly used today in the oasis of Figuig, 300 km from Tafilalt.54 This toponymy alone highlights the importance of the economic, environmental, and probably political role of this water supply system in the development of this area, while at the same time the rest of Tafilalt was probably to wither. Now, this region of Tafilalt is precisely the one where (as early as the 14th century?) the Sherifian families of Sijilmasa settled,55 which leads us to draw a correlation between the rise of this socially privileged faction of the Filalian population and the considerable work of planning and digging that meant the implementation of khettâra-s. The accurate date of the building of this group of khettâra-s is unfortunately unknown and, according to archaeological observations, may range from the 15th to the 17th century. But even with such vagueness, it remains chronologically very likely that it was indeed for the benefit of these families that the technology of khetârra was implemented.56 This plan may have called up a large workforce to serve the interests of a minority and would then illustrate not only the increase of the rivalries as the consequence of the monopolization process of hydraulic resources in Tafilalt, but also the development of privileges for small groups—thus the heightening of a strong social hierarchy now visible in hydraulic practices. A similar social process has actually been documented by Gilbert Grandguillaume (EHESS, France) thanks to legal documents in the oasis of Tuat, where the Sherfian clan has been recognized as having progressively captured the social prestige of the group as well as a privileged access to water.57

• 58 P. Guichard et al., 2000; D. Schroeter, 2000.

27The recent hydraulic history of Tafilalt bears out this social dynamic over the past three centuries. Although some of the khettâra-s of Tafilalt are still in use today, their usefulness considerably decreased when the head dam of the Ziz canal was rebuilt and when massive irrigation based on flood control was restored. This event has not been accurately dated so far, but it occurred when the population of Tafilalt was able to raise a collective force to carry out these major works. However, the origin of this motivating will, which would help to date the rehabilitation work on the Ziz, has yet to be identified. The first hypothesis is to relate this event to the restoration of a community solidarity similar to the one of the early Middle Ages leading to a joint decision-making. But this hypothesis remains fragile because of the political context of the region in the modern era, which was marked by multiple territorial conflicts that the texts describe quite well. The second hypothesis consists in attributing this responsibility to the powerful networks of zawiya-s (Sufi brotherhoods) that were particularly strong during the 17th century, as used to be the zawiya of Illigh, which controlled Tafilalt between 1620 and 1640.58 However, since Illigh’s interests in the control of Tafilalt were mainly economic in nature (= trans-Saharan trade taxation), it is unlikely that this movement, originated from the far Tazerwalt (Atlantic Morocco), consented to any large-scale local investment in the field of water management, with no direct exportable benefit. A final hypothesis is to ascribe this decision to the Sherifian powers that successively had authority over Tafilalt: first the Saadians, in the second half of the 16th century, and then the Alawites, from the middle of the 17th century.

• 59 J.B. Gramaye, Africae, p. 188.
• 60 R. Messier, 1999, p. 289-290.

28In the absence of absolute dating of the new head dam on the Filali Ziz, it remains difficult to single out one or the other of these hypotheses. But the Saadian option is quite convincing insofar as texts and archaeology have proven that under the reign of Ahmed al-Mansûr (1578–1603), the sixth sultan of the dynasty ruling over Marrakech, the Saadian power had initiated a vast program of rehabilitation of the urban center of Sijilmasa, at a time when the city was largely deserted and ruined.59 The outlines of this ambitious reconstruction campaign are unfortunately unknown, but the building of a new great mosque on the exact location of the medieval one has been demonstrated,60 and it would not be surprising if this major work was linked to an equally ambitious hydraulic policy of rehabilitation of the Ziz. Indeed, for anyone revitalizing the urban center, it would have been logical and necessary to ensure a permanent and reliable water supply of the city. As for the Alawite hypothesis, it is as convincing because of the local roots of this family, which had every interest in restoring satisfactory water conditions in the palm grove that supported its political rise.

• 61 T. Madani, 2003, unpaginated, II-4-B; the same law case at the same moment is described and analyze (...)

29Whatever the accurate date and origin of the restoration of the hydraulic network connected to the Ziz canal, this rehabilitation is synonymous with a significant evolution in the overall design of the system. Indeed, as previously announced, the sub-contemporary irrigation network of Tafilalt includes a number of anomalies that unbalance the horizontal and egalitarian organization of its origins. The two most important of these anomalies precisely deal with the water supply to the Ifli district. The first has already been mentioned above with the Ghorfa dam. Located far upstream of Tafilalt, this dam shows the particularity of irrigating two distinct sectors located on the same left bank of the Ziz (while all the other dams are made of a single main seguia). The first area (Tanijiout and north Ifli) is logically located immediately downstream of the diversion, but the second one—through a second seguia that bypasses a large part of the oasis from the east—is in a perimeter located in the southern extension of Ifli. As a consequence, the Ghorfa dam exploits the Ziz by diverting part of the water far upstream from its final purpose without respecting the usual priority of upstream lands over downstream lands. It thus appears that the southernmost land irrigated by the Ghorfa seguia could be associated with a colonization process for new agricultural land for the benefit of the inhabitants (or owners) of the Tanijiout and Ifli districts, who are the ones to manage the water supply. It is difficult to shed light on the exact reasons for this agrarian extension (increase in the rural population? increase in production needs? settlement of new social groups?), but the very upstream position of the water intake suggests a political decision-making process that was vertically imposed on the population. A medieval example of the extension of an irrigated territory due to an imposed political decision has been identified (thanks to extant texts) by Tariq Madani (Université Mohamed I^er, Oujda, Morocco), in the region of Fez: in order to favor the tribes that had rallied to their cause, some Merinid princes (last quarter of the 13th century) proceeded to an important reorganization of the agricultural landscapes in order to reallocate plots and water rights to their supporters, especially in the form of iqtâ’ (land tax concessions) and at the cost of the peasant system that had previously operated in the area. This decision by force would be the origin of century-and-a-half hydraulic conflicts among the various owners involved in this long-lasting tussle.61 A similar process could have affected the Tafilalt; and even if it is still undated, it reveals the primacy of Ifli/Tanijiout decisions over the rest of the palm grove.

• 62 H. Dastugue, 1867, p. 369-370.

30The Wengaga dam is the second major anomaly that unbalances the renovated hydraulic network of the Ziz canal. Built on the outskirts of the Sijilmasa site—that is, right in the heart of the palm grove—this dam is located almost against another dam, the Safsaf dam. With its upstream position, this dam creates an astonishingly commanding relation that halts the water previously intended for the second diversion. There is therefore strong competition between the two facilities, which do not respect the usual rule of being spaced out—a spacing precisely intended to avoid any conflict interaction and to ensure the fair supply of water between neighbors (Figure 1). The Wengaga dam is the intake of the so-called Wadi Chorfa (a seguia also known as the Wadi Ifli), which feeds the Ifli district and thus the Sherifian district of the palm grove. In this district stand some of the oldest qs̱ûr of Tafilalt, many of which were occupied in modern times by relatives of the princely family. Some of them date back to the 17th century, as does the qs̱ar Rissani, built by Moulay Isma’îl (1672–1727).62 The purpose of Wadi Ifli/Chorfa consequently suggests that its design was contemporary with the territorial development and organization of this district and that it is therefore relatively recent vis-a-vis the old history of Tafilalt. Similarly, since the Chorfia seguia runs across the archaeological site of Sijilmasa—passing along the northern walls of the medieval city on one side and trimming a poorly characterized sector of ruins (forming ancient suburbs?) on the other side—this observation reinforces the hypothesis that the canal was dug later, at a time when Sijilmasa was already partly abandoned. It is thus very likely that the building of the Wadi Ifli/Chorfa dates from the spread of the Sherifian clan outside of Sijilmasa and is therefore later than the 13th century and earlier than the beginning of the 17th century. This cultural, political, and spiritual context could also explain the dominant position of the dam, immediately upstream from Safsaf.