• 1 Singer, 1992 p. 49; Mukahalah, 1992 p. 247.
• 2 Ḏayl, vol. I, p. 358.
• 3 Tārīḫ, vol. IV, p. 151.
• 4 Sulūk, vol. III, p. 1001.

1Migration is a natural component of the cycle of peasant life, as it is one response to labor shortages on large-scale agricultural projects or those that arise during labor-intensive periods. In such cases, peasants move temporarily to other farms and villages or reside seasonally in makeshift housing at the edge of fields until the harvest is done.1 This leaves no trace in the written record and only the slightest one archaeologically. It is transient, cyclical, and undisruptive. Short-term migration is also one of the few outlets peasants in traditional societies have to escape disaster, man-made and natural. Syrian narrative sources are punctuated by accounts of trauma-induced movements of peoples during the Mamluk period. Peasants availed themselves of this option during times of armed conflict, drought and flood (when roads were passable), but it was almost always a short-term move; migrants usually returned home after the danger had passed and rebuilt their lives. To cite specific examples in what is today’s Jordan, during the Mongol advance deep into Syria in 658/1259 Damascenes fled south and found refuge in al-Karak (Kerak); they had been displaced for a period of 35 days and eventually returned to their city.2 Similarly, the villagers of Aḏriʿāt relocated to fortified ʿAǧlūn during Timūr’s invasion of the Galilee and northern Jordan and returned when his forces left.3 Alternatively, the fortified centers could be the locus of political turmoil, causing a regional, however temporary, exodus: the residents of several villages in the Jordan Valley were forced, for a time, to leave their land in 802/1399, when the first of several conflicts in Kerak spilled out into neighboring areas.4

• 5 Singer, 1992 p. 56.
• 6 Labianca 1990.

2It is, however, permanent migration - specifically the abandonment of villages – with which we are concerned here. Peasants left their homes for good for a variety of reasons, but most of them were political: the violence of a state official, who continued to abuse them, unsupportable taxes, lack of security and justice. From the perspective of the state, migration of this sort was a financial and political disaster, and in the eyes of contemporary historians heralded a society in crisis.5 For the migrants, it was an act of political will or one made when there were no other options. The best archaeological evidence for such a large-scale and permanent peasant migration in Transjordan (hereafter simply “Jordan”) is the decline in the numbers and concentrations of sites identified as late Mamluk and Ottoman in date. The phenomenon has often been described in terms of the “nomadization” of the rural hinterland, as former peasants, under the stress of Bedouin incursions against their villages (exacerbated by the withdrawal of Mamluk forces from the region), general economic decline (with the collapse of the Mamluk state), and repeated environmental disasters (earthquakes and droughts – the theme of climate change is often introduced here), returned to a more nomadic way of life, combining small-scale, subsistence agriculture with herding.6 According to this scenario, the 15^thcentury ushered in the beginning of the long decline of Jordanian village life.

• 7 Miller, 1991 Macdonald, 1992 Kareem 2000.

3The settlement decline of the late Mamluk period (15^thand early 16^thcenturies) is one of the most important phenomena in the cultural history of southern Bilād al-Šām and particularly for Jordan, as there is a growing belief that the settlement and ecological patterns that developed then laid the foundations for many aspects of modern Jordanian society. It is an issue, moreover, that has for the last twenty years dominated debates on the archaeology of late medieval Jordan. Archaeological surveys have documented a radical decline in the number and size of settlements, positing general demographic decline and the “nomadization” of residence and land use (fig. 1 and 2). According to these surveys, there is a striking discontinuity of settlement from the Mamluk to Ottoman periods, and this has generally been attributed to the collapse of the Mamluk state and the turmoil that ensued, including civil war and Bedouin attacks on villages. In the southern plains and Jordan River Valley as many as 60-85% of the sites once occupied in the Mamluk period had been abandoned for full-time settlement by the 16^thcentury, and in the central plains the rate of abandonment can be estimated to be around 30-50%.7 There appears to have been, moreover, no demographic recovery, to 14^thcentury levels, until the 20^thcentury.

• 8 The Northern Jordan Project, launched by the author in, 2003 was designed to investigate these very (...)
• 9 Hütteroth and Abdulfattah 1977.

4Assumptions about widespread settlement decline, however, are not supported by surveys in other regions of the country. On-going fieldwork in the northern hill country is demonstrating a more or less steady and continuous history of settlement in this period and, in fact, economic and demographic growth leading up to and during the 16^thcentury.8 Textual sources of the period, as well, suggest that settlement shifts did occur, but that they were regional and did not necessarily correspond to demographic decline as a whole. Moreover, the same archaeological surveys attest that although a large number of villages appear to have been abandoned, at least for full-time occupation, at the end of the Mamluk period, many new settlements subsequently appeared. These latter are notably smaller, architecturally more ephemeral, less intensive in their use of land, and follow a different pattern of distribution. This uneven picture of settlement presented for Jordan contrasts with other regions of Bilād al-Šām. Abatement in settlement for late medieval Palestine, for example, appears to have been less acute, in spite of the similar settlement structures, economic patterns, and relationship to the imperial state.9 While these differences in settlement history have been frequently noted, they have never been explained.

• 10 Brown and Neuberger, 1977 Cohen and Deng 1998.
• 11 See, for example, Vassberg 1996 on 16^thand 17^thcentury Spain.
• 12 On 19^thand early 20^thcentury settlement in Jordan, see Rogan 1999 and Fischbach; for the “Middle”(...)

5Turning to migration theory can help conceptualize ways in which planned economies and labor markets can transform traditional societies. Such scholarship is largely the product of demographers, labor historians, and geographers interested in the decline of village life as one way non-local economic forces affected traditional societies.10 More recently the impact of wars and genocide, and their resulting “internal displacement”, has motivated scholars to turn to “the political” in their search for catalysts behind migration. Many of the actors regularly cited for modern movements of peoples – the governments of industrialized nations, the World Bank, international corporations – have, however, no correlates in pre-modern society. For the study of migration in a deeper historical context, particularly in a time and place where modern census-taking was not practiced, there are no local maps extant, and peasants regularly moved back and forth between villages (during the most labor-heavy seasons), new methods and lines of inquiry need to be explored. Recent scholarship on human migration in pre-industrial Europe has suggested ways in which anthropologically informed readings of contemporary documents can shed light on such demographic processes, considering movements of peoples in the context of larger political, economic, and environmental forces.11 Borrowing from studies of migration in the modern world, such scholarship confronts the myth of the “immobile village” in describing the interconnectedness of villages with the state and with regional systems; it breaks with modern migration theory in its emphasis on the social over the institutional. Although there has been fine research in recent years on settlement along this line in Tanzimat and Mandate-era Jordan and Palestine, little comparable exists for the late medieval and early Ottoman periods.12

• 13  It should be noted that most surveyed sites in southern Bilād al-Šām as a whole are dated to the l (...)
• 14 Walker, 2011 ch. 3.

8In reviewing the published reports, certain regional patterns, however impressionistic, do become apparent.13 Most Mamluk-era villages had a long history of occupation that extends back to the Byzantine period and beyond. They were concentrated on the plateaus and near wadis and their tributaries. In the Ottoman period they were generally replaced with fewer, and smaller and less substantial, farmsteads, located near springs, in the hills, and on the edges of plateaus. It is not merely that many sites, particularly in central and southern Jordan, were abandoned by the 16^thcentury, but many new ones (however smaller in size and “lighter” in construction) were established. Regional comparisons are revealing in this regard. Moving from south to north: 62% of the medieval sites identified in the Wādī Ḥasā continued to be occupied/used by Ottoman times, although 68 “new” sites of this period have been recorded; on the Kerak Plateau, 87.5% of the medieval sites were abandoned, but 23% of the Ottoman-era sites were new ones; on the Madaba Plains of central Jordan one notes a decline of 28% in the number of sites from Mamluk to Ottoman times; some 78-86% of the medieval sites in the Jordan River Valley were abandoned by the 16^thcentury.14 One should keep in mind, of course, that a “site” is not necessarily a “village”, and the site type is not always clear from the published reports. Nonetheless, the discontinuity of site numbers, distribution, and quality from the late medieval period is undeniable.

• 15 Brown 1992.
• 16 See review in Johns, 1992 p. 365.
• 17 Brown, 1992 p. 440-441.
• 18 Ibid.
• 19 Johns 1994, p. 22; Walmsley 2001, p. 528.

9Data from two of these regional surveys has been recently reinterpreted, positing a dispersal of settlement from the 15^thcentury. In the case of Kerak Plateau, it is possible that villagers of the lowlands simply moved to the highlands, to avoid the political entanglements of Kerak Town and the waning security for its villages. Pulling on data from the Central Moab and Kerak Plateau surveys, Brown argued for a dispersal of settlement, rather than merely disappearance of villages.¹⁵ The surveys indicated that sites dating between 1400 and 1600 C.E. (dated on the basis of ceramics) fell into two categories: those continuously occupied from the early Mamluk period and new settlements, the latter largely distinguished by rudimentary architecture (or none at all) and fewer ceramic remains. She interpreted these patterns as evidence of population dispersal, or a move towards pastoralism, at the end of the Mamluk and beginning of the Ottoman eras.¹⁶ She cites one reason for the marked shift in settlement in the 15^thcentury from the central Kerak Plateau to the hills of its southwest rim as a concern for security, as this region was less vulnerable to attacks by Bedouin tribes.¹⁷ Situating such a scenario in the larger history of the Mamluk state, one could argue today that with the breakdown of the state politically and militarily, and the withdrawal of important resources from the provinces, rural security and prosperity could not be maintained on the open plateaus and in large villages. Dispersal of larger villages, with migration to the more isolated and broken hill country of the plateau margins, was one response of local peoples to imperial collapse. Although Brown cites primarily security reasons for the relocation, she also notes the presence in the hilly territories of springs suitable for the irrigation of orchards and gardens.¹⁸ Echoing this theme, Johns also posits a dispersal of settlement, but one that may date later, to the 17^thand 18^thcenturies, when conflicts between peasants and nomadic communities were more likely to have disrupted agricultural production.¹⁹

• 20 Occasionally,contemporary historians refer to the number of villages under the command of a town or (...)
• 21 While archaeologists tend to think of dispersal as the natural outcome of armed conflict and econom (...)

10What is not clear, either from the Kerak Plateau or country-wide, is whether the settlement patterns documented by surveys reflect demographic decline per se or purely shifts in settlement distribution through abandonment (and subsequent relocation outside the region), dispersal (relocation from large villages to numerous small settlements), or clustering of sites (abandonment of smaller villages for a larger, central one or, alternatively, the merger of several villages), the population remaining stable. There are no extant Mamluk-era censuses for this region to compare with the early Ottoman tax registers (that record tax-paying populations).²⁰ Moreover, as noted above, scale of sites is not systematically recorded in surveys in the region. The relative size of a site is critical to know in this regard, as dispersal and clustering are responses to different kinds of pressure.²¹ Equally relevant is site location (within a particular ecological zone) and land use (intensive or extensive subsistence or market agriculture). These latter are, fortunately, generally noted in published survey reports and give an indication whether archaeological sites are located in optimal zones for certain kinds of agriculture or for defense. Surveys in the Jordanian plains (Kerak Plateau, Madaba Plains, Ḏibān Plateau) indicate a general abandonment of the open plains. Whether this was for purely defensive reasons or to exploit different environmental zones (indicating new agricultural regimes) is not clear from the survey data alone.

• 22 Comprehensive lists of publications from both projects can be found in the relevant chapters of Lev (...)
• 23 The 16^thcentury registers (tāpū defterleri) relevant to Jordan (Liwā’ ʿAǧlūn, Banū Kināna, Banū Ka (...)

11While most of what archaeologists know about late Mamluk Jordan comes from surveys, recent excavations of villages of the period, although few, are documenting the process of very gradual decline in the size of the sites and intensity of land use. There is nowhere in the country any indication to date of a cataclysmic event – a war or earthquake, for instance – that led to the immediate and full-scale abandonment of a village or town in this period. On-going excavations at Tall Ḥisbān and Ḏibān in central Jordan, for example, are consistently providing evidence for a long and piecemeal abandonment of the villages for full-time settlement.22 Built structures at both sites were reused multiple times over the course of the 15^thand early 16^thcenturies, but in squatter fashion with the construction of hearths and storage bins in secondary context and new construction limited to single-row and hastily constructed partition walls and the closing of doorways. These are the kinds of occupation defined by seasonal settlement and non-intensive land use and may be associated with the mazraʿa-s (isolated farmsteads) of the early Ottoman tax registers.23

• 24 Palmer, 1998 p. 132.
• 25 Ġawānmeh, 1985, Shehadeh, 1985, Walker, 2008, Van Zeist, 1985, Heim et al. 1997, Bar-Matthews et al (...)
• 26 Lucke et al. 2008, p. 183.
• 27 Ibid.

13Such are the settlement patterns; accounting for them, however, is another matter. Archaeologists have tended to emphasize climatic conditions, such as rainfall, in explaining settlement shifts. Jordan is notoriously vulnerable to drought; even today the wheat harvest fails once in five years as a result of insufficient rainfall.24 Many regions of the country today, as in the past, receive just enough rainfall to support the cropping of grains without irrigation. Historical, archaeological, palynological, sedimentological, and dendrochronological analyses have all indicated that the late 14^thand 15^thcenturies witnessed several cycles of drought in the region and a general trend towards desertification, with higher temperatures, reduced rainfall, and abandonment, to some degree, of once cultivated fields.25 Annual rainfall in the late Mamluk period fell by an average of 50 mm, which in marginal zones close to the 300 mm isotope would have made it relatively impossible to grow grains26; many of the open plains of the Jordanian interior fall into this category. The climatic changes, then, are confidently demonstrated for the period; what is not clear is the relationship between this and the decline of agriculture and of sedentary settlement. Extreme environmental events do have at least a short-term impact on settlements: floods (which can occur with particular severity during otherwise dry phases) destroy terrace walls and lead to erosion, removing valuable soils and filling cisterns with silt27; several years of drought can destroy the economy of local communities. They do not necessarily, on their own, lead to the permanent abandonment of villages. While the relationship between these climatic trends and settlement are not clear, it is possible that the drier conditions exacerbated problems in the region already created by war and rebellions and general economic decline. It is, moreover, likely that certain regions of the country fared better than others; it is this regionalism in settlement history that is suggested by the survey data. The well watered North and those communities that already depended on irrigation were less affected by fluctuations in rainfall. On the other hand, villages that relied on rain-fed agriculture, such as those located on the open plains, would have suffered.

• 28 The following textual analysis pulls from Walker 2011, ch. 4.
• 29 Ġawānmeh 1985.

14The information about settlement change culled from textual sources is equally problematic to decipher.28 The chronicles for the post-plague era document numerous incidents of epidemic, drought, and famine that struck Jordan with an increasing intensity and frequency in the 15^thcentury.29 The annals for these years note evacuations of villages, as people left in search of food and shelter. They describe the initial crisis (an attack on a village, the rise in prices resulting from insufficient wheat supplies), but seldom cover the outcome of events. One cannot automatically conclude that these crises, as traumatic as they were, directly impacted settlement on the long-run, at least not on the basis of the chronicles alone. Unfortunately, historical references to Jordanian place names after the late 16^thcentury are rare, until the 19^th, when travelers to the region (pilgrims, adventurers, and scientists) wrote descriptions of the countryside and the people that inhabited it. The mantra of an empty land with low population and few villages that is repeated in many of these accounts should be read with caution, as visitors did not recognize many settlement forms and land usages that are traditional to the region. Furthermore, there is no overt connection between the factors behind the socio-political conditions of 19^thcentury Jordan and those of the 15^th. Nonetheless, it is important to note their comparisons of Transjordan with Palestine in this latter period, the latter of which was more densely occupied, in more numerous and larger villages.

• 30 al-Bahīt 2008, p. 149.

15Collectively, the narrative sources describe a Jordanian countryside that is caught up in the political turmoil of the day and the economy of which suffers, to some degree, from lack of investment by the state in security and infrastructure and from repeated natural disasters. Migration from villages was a reality of the fifteenth century, although the degree to which these were long-term moves is not always specified in the chronicles. Some degree of depopulation, or shifts in settlement, however, is suggested by the 16^thcentury Ottoman tax/survey registers for the region, which refer to many villages as ḫālī (empty). These villages, nonetheless, still paid taxes, either on flocks or small plots of land, which indicates that their former residents had either adopted a semi-nomadic lifestyle (an idea with much currency in archaeological circles) or had relocated to other villages, while continuing to cultivate their old fields. “Empty” villages are concentrated in the plains and southern districts; none of the villages of northern Jordan are “empty” or “in ruins” (ḫarāb), according to the tax registers. In fact, for the North the registers document only demographic and economic growth over the course of the 16^thcentury. In some cases, there appears to have been too few imams to service the growing Muslim population there.30

• 31 Walker 2008.
• 32 Johns 1994, p. 25 – pulling on Hütteroth and Abdulfattah 1977.

16Thus, using agricultural production as a proxy for measuring demographic change, the registers document the following. While some villages were no longer inhabited full-time after Ottoman annexation, their fields remained productive enough to tax. Moreover, the large estates in the Jordan Valley created earlier by Mamluk sultans were maintained as taxable units, as they continued to be financially viable and productive enterprises.31 It is not possible, given the nature of survey data described above, to compare numbers of villages from the Mamluk (based on archaeological surveys) and early Ottoman periods (based on tax registers). Nonetheless, it seems likely at this point that the overall population did not so much decrease in number as disperse into smaller settlements, with a shift to more subsistence-based and local market agriculture, family-run agricultural enterprises (honey production, for example), and animal husbandry. Again, there are significant regional differences in productivity from one region to another. According to the last tax register of the 16^thcentury for Liwā’ ʿAǧlūn, all the revenues from Salṭ and two-thirds from Šawbak came from taxes levied on nomadic tribes (their flocks), while full-time agriculture provided all the revenues of Kerak and Ǧibal.32

• 33 Ġaswānmeh 2002, p. 229.
• 34 For the waqfiyya of the village of Ādar (Kerak region) of 777/1375, see Ġaswānmeh 1982 and Al-Qaḥtā (...)

17The decline of certain agricultural sectors is also suggested by Mamluk-era texts. The 15^th-century taqlīd of Ibn Nabāṭa, transmitted by al-Qalqašāndī, charges the leader of the ʿušrān with the task of returning the wheat-producing Balqā’ to the prosperity of former years.33 Contemporary waqfiyyāt describe mills closed, presses no longer functioning, and houses standing empty in villages, the northern hill country included.34 One should avoid the temptation, however, to over-interpret such references, as the same sources describe fully functioning villages with a productive agricultural base: architectural ruins could be the result of leaving an old house for a new one or the calculated decision to move from one industry to a more lucrative one, rather than overall decline in either sector. The abandonment of fields should also be interpreted in this vein, as part of larger changes in the way the land was used and its products marketed. Narrative sources, however, are clear about the collapse of the state-run sugar industry in the 15^thcentury. That industry never recovered; while villages in the valley paid taxes on a variety of agricultural products in the 16^thcentury, sugar was not among them.

• 35 On these government initiatives, see Walker 2011, ch. 3, based in part on Ibn Qāḍi Šuhba’s Tārīḫ.
• 36 Heyd 1960, p. 101, 114-115; Al-Bakhīt 1982, p. 220-221.

18Contemporary textual sources suggest that a combination of political and economic factors, rather than a single catalyst, were behind these trends, launching movements of peoples and leading to changes in agricultural production in the late Mamluk period. One factor may be state-initiated population transfers, a characteristic of Mamluk governance that has not attracted scholarly attention until now. It does not appear to have been a regular practice, but was an action of last resort for largely financial reasons. In 757/1356, civilians, officials, and many public institutions were transferred from the town of Ḥisbān to ʿAmmān by Amīr Sarġatmiš. The rational for this move, the decision by a Mamluk officer that was a failure on the long run, was to revive ʿAmmān’s economy and prepare it to be a functional, and self-sufficient, administrative center. A similar action was taken in ʿAǧlūn by Amīr Tankiz, Governor of Syria, after the great flood of 728/1328, in order to repopulate it and help it recover the extensive losses to its marketplace and residential quarters.35 Population transfers are better known from the Ottoman period, when new villages and markets were established under state initiative for a range of defensive and economic reasons.36 It is not clear at this point what role forced migration by the state played in the settlement transformations of the 15^thand 16^thcenturies, but it is one possibility to consider.

• 37 The data on which this scenario is based can be found in Walker 2009, 2008, 2007, 2004a, and 2011, (...)
• 38 For possible archaeological correlates for this pattern of land use and tenure, see Walker 2008, p. (...)

19The most significant innovation by the state with implications for settlement, however, lies in changes in land tenure resulting in wide-scale transformation of rural lands in the region to awqāf. While the mechanisms of this complicated process, reconstructed from documentary sources (such as court records and tax registers), is beyond the scope of this paper, it is appropriate here to emphasize the local impact.37 In the late 14^thcentury, as part of his plans for financial recovery, Mamluk sultan Barqūq purchased large tracts of agricultural land in Jordan from the Bayt al-Māl. He subsequently endowed this land ostensibly for charitable purposes, but the bulk of the revenues were shifted to his own fisc to cover government expenses. The post-plague era witnessed an expansion in the number, size and frequency of such awqāf, which increasingly included private (family) endowments by local Jordanians, Muslim and Christian, during the late 14^thand 15^thcenturies. The “civilian” endowments were made possible by the ready sale of farmland to local villagers and townsmen from the Bayt al-Māl and reinforced a shift to small-scale agriculture: irrigated orchards and gardens and small plots of land devoted to grains. The trend towards small-scale, intensive agriculture in the hands of private entrepreneurs was followed throughout Syria; 16^thcentury siğillāt (court registers) in Damascus document the sale, lease, and inheritance of fractions of land, much of it constituting family awqāf.38