KINSHIP AND HISTORY: TRIBES, GENEALOGIES, AND SOCIAL CHANGE AMONG THE BEDOUIN OF THE EASTERN ARAB WORLD

Most scholars of tribal organization among the Bedouin of the eastern Arab world utilize a two-dimensional, hierarchical model of Bedouin kinship that represents only relations of descent and af=inity. This model resembles a genealogy and shows how small descent units are enclosed by larger ones. It implies that tribes grow in size only through biological repro-duction. Such a representation of the Bedouin tribe fails to distinguish politically central lineages from politically peripheral lineages and also ignores the processes through which foreign lineages become “attached” as clients to politically powerful, central lineages. To correct and supplement this genealogical model, the author presents a concentric model of Bedouin tribes that adds a “central/peripheral” distinction. This model also includes relations of political “attachment” that can affect the internal morphology and growth of Bedouin tribes in ways that are comparable to the effects of af=inal and suckling kinship relations on internal organization. The proposed concentric model thus allows us to represent historical change more accurately and also brings us closer to Bedouin concepts of tribal organization.

In this paper I will present a model of relations among unilineal descent groups in a particular sub-set of human societies: the rural, Arabic-speaking societies who call themselves -or who are called by others -Bedouin. The model focuses on political and spatial relationships among descent groups in these societies and portrays the assembly of these groups to form tribes. To some extent, the model represents the way in which these kinship relationships are conceptualized by the members of Bedouin societies. However, my model also is designed to facilitate the comparison of how descent groups are related in one society -that of the Rashāyidah Arabs of eastern Sudan (see Figure 1) -with descent group interrelationships in other Bedouin societies. Thus, it incorporates features -such as political hierarchy -that are not present in native models but that must be included to make cross-cultural comparison possible. To this extent my model is also etic. It is primarily a descriptive, rather than explanatory, model. It highlights significant variation in the organization of Bedouin tribes but does not show the causes behind this variation.
Most anthropologists who have researched unilineal descent groups in Arab societiesthat is, tribes, clans, and lineages -have used a branching hierarchy model to describe them. This is particularly common in ethnographies of Bedouin tribes (Lancaster and Lancaster 1981;Peters 1960; for a useful review, see Varisco 1995:139-141). This model stems from segmentary lineage theory in anthropology, as exemplified in the works of E. E. Evans-Pritchard (1949) and many others. I myself have used a branching hierarchy model to describe the Rashāyidah tribe of eastern Sudan (Young 1996:85-88).
I will argue that the branching hierarchy model is not useful for describing and explaining historical change in Bedouin societies. Although this model has some advantages, it tends to equate the internal composition of a Bedouin tribe with its genealogies and obscures some political relationships and distinctions that are just as important for Bedouin kinship as genealogy. As an alternative to the branching hierarchy model, I will present a concentric model of the Rashāyidah tribe that distinguishes "authentic" (Arabic: aṣlīyah; see the Appendix, below, for the Arabic transliteration system used in this paper) lineages from lineages that are "attached" (Arabic: multaħaqah) or are "included among" (Arabic: min ḍimn) the "authentic" lineages of the tribe (see Ħasan 1974:11). When using glosses of Arabic terms to label the relationships and components of this concentric model, I will enclose these glosses in quotation marks (ex. "attached", "authentic"). This serves to make it clear that I am trying to render native terms about intra-tribal relationships. I am not making historical claims about the nature of these relationships, however. I employ these glosses only to convey the ideas expressed by Arabic speakers or found in Arabic documents. Finally, I will apply the same concentric model to many other Bedouin groups in the eastern Arab world and will suggest that their social organization and the historical changes that these groups have experienced can be represented more accurately using this concentric model. Before proceeding with a discussion of the models, let me pause briefly to consider the word Bedouin. I have previously expressed my discomfort with using the word Bedouin as an etic, analytical category (Young 1999). Although the word seems to have an objective meaningi.e., "Arabic-speaking nomadic pastoralists" -it has other connotations. It is derived from the Arabic word baduw, which is etymologically associated with the word bādiyah, "desert" (Hava 1899:24). I inquired about this word while conducting field research in Jordan and discovered that it is sometimes given as the answer to a question about "origin (aṣl)." Those who reply to the question "What is your origin (aṣl)?" by producing a lengthy genealogy linking living people to a tribal ancestor say that they are baduw. Those who reply by identifying an agricultural village as their "origin" call themselves fallāħūn "cultivators/peasants," while those who reply by linking themselves to a politically powerful or wealthy ancestor who owned land and fine houses in either cities or the countryside call themselves ħaḍar, "settled/urban people." Since such terms are mythological/historical, they do not necessarily describe the current economic activities of living people.To quote another researcher who has arrived at the same conclusion: "In modern Jordan, very few of the large Bedouin groups are…pastoralists or nomads. The decisive elements of their Bedouin identity -that which makes them baduw -are rather their memory of a lineage associated with a distant nomadic past, and their adherence to ideologies of equality and autonomy" (Rapoport 2004:5; see also Lancaster 2006:338 andEl Guindi 2012:546-47). Many Jordanians who describe themselves as baduw have no livestock and spend no time in desert pastures, but they do tend to marry other people who call themselves baduw. The same preference for marriage with baduw over marriage with any other category is found among the baduw tribes of northern Saudi Arabia (Lancaster and Lancaster 1981:24, 46-48;Reilly 2013:383-84).
The Rashāyidah do not use the word baduw to identify themselves. They say "We are people/Arabs who take up (their tents) and put (them) down again (in new places)" (in Rashīdī Arabic: ħinna ʿarab nishidd wa-niħuṭṭ) (see Young 1996:8). For them, the key elements of their identity are Arabic speech and nomadic pastoralism. But their Arabic-speaking Sudanese neighbors call them Bedouin and believe -correctly -that the Rashāyidah are culturally and socially similar to the Bedouin of Saudi Arabia. Moreover, like other Bedouin, when the Rashāyidah are asked about their origin they produce a long genealogy that links them to their ancestor, Rashīd al-Zawl (see Ħasan 1975:174-175). Thus, the models that we can use to represent the social organization of the Rashāyidah should also apply to other Bedouin societies, that is, Arabic-speaking communities with long genealogies that link living people to their tribal ancestors who were nomadic pastoralists at some time in the past.
As I said earlier, the model of the Rashāyidah tribe that I have used previously is a branching hierarchy model (see Figure 2). In Figure 2, the names of the three main sections of the tribe appear in bold upper-case print. "Authentic" clan names appear in bold lower-case print, while "authentic" lineage names are not in bold print. The names of "attached" clans appear between parentheses in bold italic print. Each "attached" clan is joined to one of the three main sections with a dotted line. The Rashāyidah say that six of the clans in Figure 2 are only "attached to" or "administratively associated with" the tribe and are not descendants of the Rashāyidah's eponymous ancestor, Rashīd al-Zawl. In contrast, clans such as the al-Shanānīr and lineages such as the Dhuwī Ħayyān are said to be "authentic" (Arabic: aṣīl or aṣlī) descendants of the Rashāyidah's ancestor (see al -Ħasan 1974:11). Table 1 summarizes the same information presented in Figure 2. The names of "attached" clans appear in italic print. Although the branching hierarchy model in Figure 2 and Table 1 provide the same information, Figure 2 does have one advantage over Table 1. The spatial layout of Figure 2 has vacant spaces where the names of the eponymous ancestors of the various sections, clans, and lineages can be inserted. When this additional information is incorporated, the branching hierarchy model brings us closer to native concepts, since the Rashāyidah normally refer to the eponymous ancestor when referring to a section, clan, or lineage. It also makes the composition of the entire tribe resemble more closely the composition of individual families and sub-lineages; in other words, it makes a purely formal hierarchy into something resembling a genealogy. This makes the branching hierarchy model useful for analyzing cases of conflict between descent groups and the application of Bedouin customary law to punish criminals and reconcile parties in cases of homicide. Indeed, the Bedouin themselves apply a branching hierarchy model when they adjudicate a legal case and determine which relatives of the law breaker are obligated to contribute to the monetary compensation owed to the victim (cf. Bailey 2009:12-17, 60-64, 85-90;Behnke 1980;Murray 1935;Oweidi 1982).
However, the branching hierarchy model has the disadvantage of suggesting that the tribal structure really is nothing more than a genealogy and that the names of the eponymous ancestors of all lineages in a tribe are actually known by all members of the tribe. In fact, this is not the case. For example, when I interviewed members of the Barāṭīkh section of the tribe, I found that they did not know who the eponymous ancestor of the al-Saħāmīn lineage was. At best, they knew that the al-Saħāmīn lineage is a branch of the Barāʿiṣah section at some vague level of segmentation. They assumed that the members of the al-Saħāmīn lineage knew the names of their ancestors, just as they themselves could name the ancestors of the Barāṭīkh section. I found the same combination of missing information and genealogical assumptions among the members of other tribal sections. Quite naturally, the members of each particular section could name their own ancestors and merely assumed that the ancestors of the other sections were known by someone and could be recited from memory by other members of the tribe, even if they themselves did not know these names. But there was no general consensus about or knowledge of the names of all of the tribe's lineage ancestors. The only parts of Figure 2 that were universally acknowledged were: (1) the names of the "authentic" sections of the tribe, and (2) the names of the six clans that are "attached" to the Rashāyidah tribe but that do not descend from Rashīd al-Zawl. I should note that informants' statements about the positions of the al-Manāfīr and al-Kurayfāt clans are contradictory. Some say that the al-Manāfīr and al-Kurayfāt are not clans of the Barāṭīkh but belong at a higher level of segmentation, at the same level as the Barāṭīkh, Barāʿiṣah, and Zunaymāt tribal sections (see al -Ħasan 1974:2-20;Young 1996:88-89). Some informants, in other words, claim that the tribe is divided into four major sections (Barāṭīkh, Barāʿiṣah, Zunaymāt, and Manāfīr) while others add yet another section (al-Kurayfāt). These findings made me think that the Rashāyidah's emic model of their tribe was not exclusively genealogical. When they trace relationships inside a clan or lineage, they invoke genealogies. However, when discussing relationships between clans or between units at the highest level of segmentation, they also rely on the distinction between "authentic" and "attached" components. Thus, the relationship between the Dhuwī Ħayyān and the Dhuwī Yumaynī is not purely genealogical. The Dhuwī Ħayyān are said to be genealogically related to the Barāṭīkh via their common ancestor, Rashīd al-Zawl, and the Barāṭīkh are related to the Dhuwī Yumaynī via "attachment." Hence the Dhuwī Ħayyān and the Dhuwī Yumaynī are related via a combination of genealogical descent and "attachment." The inadequacies of the branching hierarchy model were identified as early as 1969 by Joseph Chelhod, who called it misleading and simplistic. He complained that, by insisting on the unity and genealogical homogeneity of the tribe, the traditional Arab genealogists -who, like the ethnologists who developed segmentary lineage theory, adopted branching hierarchy models -"have completely lost sight of heterogeneity" (Chelhod 1969:89, 92). He argued that all Bedouin tribes are genealogically heterogeneous. That is to say, he did not believe that the members of any Bedouin society could all be truly descended from their eponymous ancestor. He felt certain that some members of any tribe must owe their membership to some non-genealogical principle of membership, especially in very large tribes and tribal confederations which he believed are built upon alliances between genealogically unrelated groups as well upon the principle of common descent. After his review of historical evidence in pre-Islamic Arabia, he was convinced that the largest segmentary units, at least, included people who were not agnatically related to each other. "The Arab tribe or qabīlah…resolves itself into…many small, almost autonomous groups that form lineages….If we take into account all the elements from outside that the tribe has gradually assimilated, it becomes obvious that its consanguineal unity is, at least, seriously compro-mised" (Cheldhod 1969:89, 91). As for the clan (Arabic: 'ashīrah), it "admits to its bosom foreign elements (clients, protected elements, confederates) who end up totally integrated there" (Chelhod 1969:90, 92; translation from French by William Young).
Chelhod grounded his objections to the branching hierarchy model in a general review of the ethnographic literature and in his own fieldwork among the Tiyāha Arabs of the Negev. The Tiyāha do not completely fit the branching hierarchy model because they are not genealogically uniform. Although the Tiyāha as a whole are said to descend from ʿAdnān (the ancestor of the northern Arabs), they acknowledge that a major section of the Tiyāha, the Ẓullām, are classified as descendants of a different ancestor, Quḍāʿah. Furthermore, the Banī ʿUqbah section of the Tiyāha traces its origins to Quraysh. The membership of the Banī ʿUqbah in the Tiyāha tribe is a matter of political dependency, not descent from the Tiyāha's founding ancestor (Chelhod 1969:90-91;see also Chelhod 1965:385).
Chelhod did not completely discard the branching hierarchy model, however. He argued, rather convincingly, that it applies fairly well to the lower-level segments of Bedouin tribes, such as the "lineage" (faṣīlah or ħamūlah). At this level, lineage endogamy tends to create an involuted descent group in which relations of affinity between families become almost indistinguishable from patrilineal descent lines (Chelhod 1969: 90-92, 98). This blending of patrilineal and matrilateral/affinal ties among closely-related kin due to endogamous marriage has been widely observed in the Arab world (see El Guindi 2012:550; Holy 1989; Khoury and Massad 1992;Jaouad et. al. 2009;Jurdi and Saxena 2003;Reilly 2013: 374-76).
But if the branching hierarchy model applies mainly to low-level units, how are we to model the high-level units such as the "tribe" (qabīlah) and "tribal section" (ʿimārah or baṭn) (cf. Chelhod 1969:92)? Chelhod proposed a model that is based on a kind of dual division which, he argued, is pervasive at all high levels of segmentation in Bedouin societies. He regarded this as an unusual form of the dual organization which obtains in many small-scale societies and which elsewhere -i.e., in non-Arab societies -is based on exogamy and the exchange of ritual and political services (Chelhod 1969:101, 112;Lévi-Strauss 1963a). To illustrate this model, he presented two historical cases: As can be seen, the Banī Ṣakhr are divided into two large fractions: the Ṭuwaqah and the Kaʿābinah, each fighting for supremacy. For a long time, the chieftainship belonged to the latter, then it passed into the hands of their rivals. The number of fractions of a tribe is variable. But it seems that pluralism tends to be reduced to bipartism: around the two main leaders a whole small world gravitates, like satellites. When the qabīlah is made up of three fractions, the most recent, if not the least important, strives to maintain the balance between the other two. From this point of view, the example of the Ħuwayṭāt is worthy of interest. Divided into three groups, two of which do not feel sympathy for each other (Ibn Jāzī and Abū Tāyih), the third (Ibn Nijād) observes a positive neutrality, trying to make the best use of this ambiguous position. During the First World War, while Ibn Jāzī was with the Turks, ʿAwdah Abū Tāyih fought alongside Lawrence with the English; Ibn Nijād oscillated between the two adversaries (Chelhod 1969:97).
That is to say, at the level of the tribe there are either two segments that are politically opposed to each other or an arrangement of three segments in which the weakest of the three segments allies itself alternately with one of the other two. Such a combination of competition with alliance thus generates a mediated binary structure, with the weakest of three segments playing the role of mediator between the other two. The mediation involved here is not simply a link between two opposed categories. Rather, it modulates the relationships of the two largest segments with respect to an exterior goal: political power. The weakest, third, segment prevents the strongest segment from attaining complete dominance by allying temporarily with its opponent. Yet, by refusing to join forces permanently with its ally, the third segment also denies total victory to the strongest segment's opponent.
The capacity of this model to represent many ethnographic cases makes it attractive. Chelhod applied it to other cases of political antagonism among tribes in two Jordanian cities (Chelhod 1969:106-111). In fact, his model also applies well to relations among the three main sections of the Rashāyidah tribe. During the twentieth century the numerically dominant Zunaymāt section sought to make its leader into the shaykh of the entire tribe. Although a series of governments in Sudan -both during the colonial period and after it -tried to support this bid for hegemony by the Zunaymāt, they failed several times, due to the opposition of the other two sections (see Hasan 1975:196-197;Hasan 1974:49-50). This tripartite structure among the Rashāyidah strongly resembles the Ħuwayṭāt case described by Chelhod. Despite the elegance of Chelhod's model it does have shortcomings. First of all, Chelhod's dualistic model does not represent the historical -as opposed to genealogical -processes that lead to changes in the size of a tribe. Second, some aspects of Chelhod's dualistic model are difficult to visualize. How, exactly, can we represent the relationships of opposition and alliance described in the Ħuwayṭāt case mentioned above?
Perhaps these shortcomings can be remedied by drawing on a second version of Chelhod's dualistic model. He presents it when describing the urban tribes of pre-Islamic Mecca. In this version, the simple political opposition between rivals is supplemented by spatial and economic oppositions: The center of the city is the bottom of the valley where stands the sanctuary whose eastern corner is adorned with the famous black stone. It is there that the ceremonies of worship and the great manifestations of the pilgrimage take place (ṭawāf and saʿy). This is where the waters of the well of Zamzam are stored, where public affairs are conducted, where the main rites of passage -circumcision and marriage -are celebrated. The suburbs form the upper part of the city, the one that receives rain without holding it back. They are far from enjoying the same sacred character as the bottom of the valley, for holiness loses its intensity as one moves away from the center. This is where the second-rate Quraysh -the clients, the needy, and the slaves -reside. This is also where the defense of the city is organized.
In short, Mecca seems divided into two concentric, rival and complementary halves. The opposition is at once between the aristocracy and the plebs, the center and the periphery, the valley and the mountain, water and aridity, the sacred and the profane, life and death. Nevertheless, an exchange of services takes place between the two halves: one provides wealth, the other arms to exploit it; one ensures the respect of the moral principles, the other assures the defense of the patrimony; one governs and administers, the other deals with sustenance (Chelhod 1969:104). Chelhod notes that, when war broke out between two elite families in the center of the city -the ʿAbd Manāf and ʿAbd al-Dār families -the "Quraysh of the periphery refused to be dragged into the conflict and observed a strict neutrality. Finally, the war was narrowly avoided thanks to a compromise between the enemy brothers" (Chelhod 1969:105). All of this suggests a visual representation of the city that incorporates a bifurcated center surrounded by a peripheral element that has a mediating role (see Figure 3).
In what follows, I will build on these efforts by Chelhod and construct a concentric model that represents the non-genealogical features of tribal organization among the Rashāyidah (see Figure 4). As I will show, this model makes it possible to represent processes of historical change  in tribal composition because it incorporates aspects of Bedouin social organization -political hierarchy and processes of attraction and dissociation -that cause tribes to grow and shrink in non-genealogical ways. I will suggest that this concentric model is applicable for describing historical change among the Rashāyidah. I will close by demonstrating that the concentric model is useful for understanding the composition of many other Bedouin tribes in the eastern Arab world and that this model is thus of more general utility.
Before proceeding, I should explain what the historical and political facts are that the concentric model of the Rashāyidah tribe can represent. Most of the Rashāyidah arrived in Sudan during the 1850s and 1860s, when small groups of them left their homelands in western Arabia and crossed the Red Sea. Some of these immigrant groups first settled in southern Egypt and Eritrea and only later joined the people who we now call the Rashāyidah of Sudan (Ħasan 1975:181-183). Ethnographic evidence indicates that some of the "authentic" members of the emigrant group probably came from an Arabian tribe known as the Banī Rashīd while other "authentic" migrants may have come from different but neighboring tribes: the Ħarb, Juhaynah, and ʿAnazah. Although space limitations do not permit a thorough review and evaluation of this evidence here, a comparison of clan and lineage names among the Sudanese Rashāyidah with lineage names among the Arabian tribes of the Banī Rashīd, Ħarb, Juhaynah, and ʿAnazah provides us with preliminary clues about the geographical origins of some of the Rashāyidah (see Table  2).
We should note, however, that some of the Rashāyidah's ancestors may have come from places farther north and east. The ethnographic literature shows that there are seven tribes (or segments of tribes) that call themselves al-Rashāyidah while two others call themselves Banī Rashīd ("children of Rashīd"). Since the word Rashāyidah is the plural form of the name Rashīd, "al-Rashāyidah" is simply a morphological variant of "Banī Rashīd." These groups are found 18 across a huge area stretching between the cities of Medina and Amman and northwards into northern Jordan (see Figure 5). Since some of these groups were nomadic pastoralists (at least until the end of the nineteenth century) they also moved into southern Syria and the Sinai Peninsula when ecological and political conditions were favorable. These group are as follows: A. The Banī Rashīd who were mentioned above. They live due north of the Saudi city of Medina in an inhospitable zone called Ħarrat Khaybar ("the lava field of Khaybar," 25° 35' 27" N, 38° 56' 57" E) which is partly covered by layers of basaltic lava (deposited by ancient volcanoes) (al-Jāsir 1980:30, 44, 76, 81, 88, 93, 146, 148, 165, 175, 187-188, 205, 208, 220, 224, 236-238, 244, 255, 268, 300, 315, 320, 360, 361, 387, 393, 410, 450, 451, 453, 456-457, 487, 492, 509-510, 511, 512, 543, 544, 551, 559, 560, 586, 587, 589, 651, 673, 685, 694, 697, 708, 719, 734, 754, 767, 782, 790 Peninsula, in Egypt (Murray 1935:269) The resemblance between the name of the Sudanese Rashāyidah and the names of these other Rashāyidah groups is certainly not proof of a common historical origin. In some cases, sharing a common name may be mere coincidence. However, the geographic positions of at least some of these northern Rashāyidah groups -for example, groups B, C, D, E, F, and I in Figure 6 -suggest that some members of these groups may have been in contact with some of the Arabian ancestors of the Sudanese Rashāyidah (i.e., the Banī Rashīd, group A) and may have joined them in their migration to Egypt and, eventually, to Sudan. ln eastern Sudan, the Rashāyidah struggled with indigenous Sudanese tribes to find pasture lands and water for their livestock. Armed clashes took place in which leaders of the Barāʿiṣah, the Zunaymāt, and the Barāṭīkh played important roles in coordinating and leading groups of fighters. Members of the "attached" lineages, however, did not take on leadership roles. They have never held prominent positions in the tribe. Ever since the Rashāyidah arrived in Sudan, competition for political leadership in the tribe has always been among the three "authentic" segments: the Barāʿiṣah, the Zunaymāt, and the Barāṭīkh. In other respects -for example, in the context of marriage -the "attached" and "authentic" segments are equals. "Authentic" Rashāyidah have contracted marriages with "attached" lineages and families. Indeed, marriage alliances are one dimension of the relationships between "authentic" segments and "attached" lineages (see Ħasan 1975:188-193; al-Ħasan 1974:2, 10 notes 1, 20). But from a political perspective the "attached" lineages are marginal.
Thus, the migration of the Rashāyidah was not just a matter of movement from one territory to another. It also involved: 1) Processes of association and dissociation. An effort was made by "authentic" lineages inside the tribe to acquire and keep new client lineages. These clients had previously been parts of other Arabian tribes but broke away, in part or in whole, from their erstwhile protectors. To attract these new clients, the core lineages had to provide protection and leadership services to the peripheral lineages, to keep them "attached" to the tribe, augment its numbers, and thus increase its ability to invade and control new pasture lands in Sudan. 2) Competition among the three "authentic" sections to capture positions of leadership.
An effort by the "authentic" sections to compel the "attached" segments to recognize the eponymous ancestor of the "authentic" segments and their genealogies as key parts of the organizing framework for the new tribe. Simply by saying that they were Rashāyidah, the "attached" lineages acknowledged the ancestor of the "authentic" groups, Rashīd al-Zawl, even though they were not descended from him.
Clearly the branching hierarchy model does not visually represent the processes that brought together the many migratory groups to make them into a new social formation. However, the concentric model, which visually represents the "attached" lineages as peripheral, captures these processes more directly and simply. I would go further and argue that the Rashāyidah case is not exceptional. Many other tribal formations have been constructed on the same bases of detachment and attachment. (For comparable processes among the Bedouin of Cyrenaica, see Peters 1990:99-102.) In my view, the distinction between "authentic" and "attached" segments that the concentric model highlights is not merely an incidental aspect of Bedouin tribal structure but is an essential aspect of that structure. For this reason, I would argue, the concentric model is more useful than the branching hierarchy model for understanding the history the Rashāyidah tribe and the histories of Bedouin tribes in general.
To support this argument, I will present a second case of "attached" lineages in some detail: the case of the Jabārāt Bedouin of northeastern Gaza.

Comparable Case of "Attachment": The Jabārāt of Northeast Gaza (Palestine/Israel)
The Jabārāt tribe, which in the eighteenth century was located on the Mediterranean coast of the Sinai Peninsula, east of the Egyptian city of al-ʿArīsh, was radically changed by a series of political events during the nineteenth and early twentieth centuries. In the 1790s, when the Jabārāt moved toward Gaza in search of summer pastures, they collided with the Tarābīn tribe and were split into two segments: one that remained on the coast of Sinai and another that was forced to move some 26 kilometers northeast of Gaza, near Wādī al-Ħasī. They arrived there in about 1799. Shortly thereafter a war broke out between the coastal segment of the Jabārāt near Rafaħ and the Tarābīn tribe that lasted twenty years. Ultimately the Tarābīn, with the help of fellow tribesmen from Egypt, succeeded in expelling all of the Jabārāt from Gaza, pushing them into southern Palestine. In the early 1830s this area was invaded by an army from Egypt. In 1834 the Jabārāt -along with other local people -rebelled against Egyptian rule and suffered significant losses when a punitive expedition was sent against them. At about this time, the members of the al-Rutaymāt tribe -who were living near al-ʿArīsh -were driven eastwards by the Tarābīn tribe into the lands of the Jabārāt. They sought -and obtained -an alliance with the Jabārāt.
The social formation produced by these armed clashes, divisions, migrations, and alliances was a loose confederation of descent groups. The nucleus of the confederation consisted of the four "true" Jabārāt tribes: the ʿUraybāt, the Ħasanāt, the al-Fuqarāʾ, and the al-Daqs. Three of these tribes had lineages "attached" to them. Some of the "attached" lineages were of Egyptian origin, while others were fragments of the Banī ʿAṭiyah tribe that lived south of the Dead Sea. In addition to these "attached" lineages, nine other tribes allied themselves with the Jabārāt and were parts of the broader confederation. The first two of these allied tribes had "attached" lineages of their own, while the remainder of the allied tribes did not. By the early twentieth century, they had established themselves in an area near the town of Bayt Jibrīn, northeast of Gaza (von Oppenheim vol. 2 [1943]:81-90) and were bound together by relationships of co-residence, common use of pastures, and relations of common descent and marriage (see Figure 7).
To describe this complex federation, von Oppenheim presents three pages of text -with footnotes -and twelve tables that list its component tribes, clans, and lineages. He also refers to tribe members who are identified only as "black slaves (ʿabīd sūd)." We do not know exactly what the phrase meant for the members of the tribe. However, it is unlikely, at this place and time, that these people were actually purchased. Among the nearby Banī Ṣakhr of Jordan, "slaves" were clients, not property. They were attached to elite "free" households and provided the domestic services associated with elite status (Alon 2016:85-96). My experience with the Rashāyidah leads me to think that the phrase "black slaves" may have been a label for a descent category that was, at the same time, a marriage class. It could have been analogous to a label that is found in Sudan: muwalladūn. Literally, this term refers to people who are not "pure Arabs" but are of "mixed" ancestry (Qāsim 1972:849). In social contexts it is used to describe the illegitimate offspring of a "free Arab" man and a woman who was a "non-Arab slave." Among the Rashāyidah, muwalladūn (or, in Rashīdī Arabic, mawālīd) belong to their father's patrilineal descent group but seldom marry the "free" members of this group. Most often the marry either muwallad members of their own descent group or marry muwallad members of other Rashīdī patrilineal descent groups (see Young 1996:114-118). Consequently, the members of this category are affiliated to various "free" patrilines and do not form a distinct descent group on their own. My suspicion is that the "black slaves" of the Jabārāt tribes had a similar relationship with the "free" members of these tribes.
In sum, people were incorporated into the Jabārāt confederacy in a variety of ways: through descent, "attachment"/alliance, co-residence, patron-client relationships, and marriage. Much of this information can be represented economically using a variation of the concentric model that was used previously to describe the Rashāyidah case (see Figure 8).
I can represent the alliances and attachments between the descent groups in the Jabārāt tribal confederation more clearly than the branching hierarchy model. What is more, it seems likely that the concentric model can be usefully applied to many other similar cases, especially in the Levant. A social historian of the Levant has noted that, although Bedouin groups in Jordan and Syria often defined themselves in terms of patrilineal descent (nasab), this agnatic ideology was frequently superseded by relations of co-residence and alliance (Leder 2015). Many Syrian tribes -such as the al-Fawāʿirah, the al-Sikin/al-Sichin, and al-ʿUqaydāt -are described as combinations of "authentic" lineages and "attached" lineages (see Zakarīyā 1983:449, 538, 568).
All of this suggests that non-agnatic relationships -i.e., "attachment" and alliance -are important features of Bedouin kinship in the Levant and, possibly, elsewhere. To evaluate this claim, let us examine some other cases.

Determining the Frequency of the "Attached Lineage" Phenomenon among Bedouin tribes
One way to evaluate the utility of the concentric model is to see how many other Bedouin tribes have "attached" lineages. Those that do have "attached" lineages probably have acquired them through similar processes of association, alliance, and competition among core lineages for political prominence that shaped the Rashāyidah and the Jabārāt confederacy. Thus, we should be able to model the historical formation of such tribes using the concentric model presented in the previous pages. If we discover that "attached" lineages are common among Bedouin tribes, we will have all the more reason to prefer the concentric model over the branching hierarchy model for representing historical change.
Unfortunately, it is difficult to determine how many Bedouin tribes have "attached" lineages with complete objectivity because the term Bedouin is not an objective, analytical term. We cannot say whether any particular tribe is a Bedouin tribe without asking its members whether or not they call themselves baduw or whether most other Arabic speakers regard them as such. However, this may be an excessively high standard for social science research, especially since most of our descriptions of Bedouin societies are taken from secondary sources rather than from living informants. At a slightly less demanding level of rigor, we could count the number of tribes that are described as Bedouin in all existing ethnographies and determine what the frequency of the "attached lineage" phenomenon is among these tribes. Even this task is far beyond the scope of this paper. There are references to hundreds of Bedouin tribes in the literature about the Arab world (al-Wāʾilī 2002). It is not practical to search through the descriptions of all of them to see how many of them have both "authentic" and "attached" lineages.
A more modest, but more achievable, goal would be to determine how many cases of "attached" lineages there are among the Bedouin tribes of the four areas most relevant to this study: Sudan; western and northern Arabia; Jordan and Palestine; and southern Syria. If the "attachment" of lineages is relatively frequent -or, at least, present -in these four areas, then the concentric model is applicable there.

"Attachment" in Sudan
Let us begin by examining one of the standard works about Sudan, MacMichael's A History of the Arabs in the Sudan. This book contains numerous references to lineages that have been "attached" to particular tribes. It says, for example, that the Nūrāb, the Sarāqāb, the Barārah, and the ʿAṭāwiyah all joined the Kabābīsh tribe of Kordofan even though they originally belonged to other tribes. According to MacMichael, the growth of the Kabābīsh tribe "is the result of a series of accretions which have been taking place for several centuries." Not only did these segments attach themselves to the tribe, but several other segments -with the names Aħāmidah, Juhaynah, Kawāhilah, Baṭāħīn, Shanābilah, Qarriyāt, and Ghazāyah -have "broken away" from the Kabābīsh in recent times (see MacMichael 1922: 307-311, 312 fn. 1; note that I have changed MacMichael's transliterations of names to make them conform to my own transliteration system for standard Arabic). MacMichael also mentions the Awlād Rāshid lineage of the Umm Qallūl clan of the Awlād Shāyīq branch of the Maħāmīd tribe of Dār Fūr, in western Sudan, as a group of people from areas farther west "who have attached themselves to the Um Gallūl" (Mac-Michael 1922:299 fn. 7). His book is peppered with references to segments of tribes that are not "true" segments of these tribes (e.g., "the conglomeration of various Arabs" who comprise the Ħamar tribe, p. 319; the ʿAbābidah and Aħāmidah segments of the Kawāhilah tribe, pp. 325-328; and the sections of the Kinānah tribe who "attached themselves" to the Kabābīsh tribe, p. 331). So MacMichael is certainly familiar with the processes of association and dissociation found among the Rashāyidah. Unfortunately, he does not make it clear whether the cases of "attachment" that he mentions are explicitly acknowledged by the tribespeople themselves or are products of his own conjectural histories of those tribes. We cannot tell from his descriptions whether the phrase "attached lineage" is a native, emic classification or his own, etic classification. Another difficulty with MacMichael is that even when the native informants recognize a distinction between "authentic" and "attached" lineages, MacMichael does not always seem to be aware of it. His description of the composition of the Rashāyidah tribe, for example, reads as follows: The Rashāyidah are recent immigrants from Arabia…After the reoccupation [of Sudan by the Anglo-Egyptian administration in 1898] they…have…been joined by considerable numbers of… Rashāyidah from the Hijaz…They…number at present between 1000 and 2000 men in the Red Sea and Berber Provinces…. The section in Berber is the Zunaymāt, subdivided into Dhuwī ʿĀyid, Ħilimāt, Dhuwī Barāghīth, Ħuwayjāt, Qazāyizah, ʿAwāzim, and ʿUraynāt. Those in the Red Sea Province are Barāʿiṣah (subdivided into Dhuwī ʿAmrī, Shanānīr and Jalādīn), and Barāṭīkh (subdivided into Manāfīr, ʿUwaymirāt, etc.) (MacMichael 1922:345 fn. 1).
His list and organization of the segments of the Rashāyidah tribe is fairly accurate (compare with Figure 2) but it makes no distinction at all between "authentic" and "attached" lineages. Although MacMichael acknowledges that the first group of Rashāyidah to establish itself in Sudan was "joined" by a later wave of migrants, he does not say that any of these lineages are "attached." This flaw in his account leads me to wonder whether similar information is missing from his descriptions of other Sudanese tribes. Although this may be a minor flaw for most purposes, for me it is an important gap. In light of this, I cannot depend on his work to come up with an accurate count of the cases of "attached" lineages among the Bedouin tribes of Sudan. His research does confirm the idea that such cases are present but it cannot tell us exactly how many such cases there are in Sudan.
Another important reference work -which has become the main source of information on Bedouin tribes in Sudan -is the vast, six-volume encyclopedia published by ʿAwn al-Sharīf Qāsim in 1996. This encyclopedia, which has superseded MacMichael, draws on a huge number of published and unpublished manuscripts in Arabic and English that were not available when MacMichael wrote his study. It is incomparably richer in detail than MacMichael's work and is free of MacMichael's annoying preoccupation with "racial" mixing and efforts to decide the degree to which each Arab tribe has intermarried with non-Arabs -a preoccupation that may have been typical of English writing about tribes in the nineteenth century but which is no longer of interest to us. However, ʿAwn al-Sharīf Qāsim tends to adopt a genealogical view of tribal composition more readily than MacMichael does.
To illustrate: MacMichael describes the Rufāʿah tribe of eastern Sudan as "a composite tribe containing more of the Guhayna element [i.e., sections descended patrilineally from the Juhaynah tribe of the Hijaz] than any other." He reports that some of its families also claim to be descended patrilineally from the Prophet Muhammad's tribe, Quraysh. He points out, quite reasonably, that patrilineal descent from Juhaynah logically excludes patrilineal descent from Quraysh and argues that, if the Rufāʿah have any links to Quraysh, they must be matrilateral links resulting from marriage between Juhaynah families and Quraysh families (MacMichael 1922:239-244). Qāsim, on the other hand, rejects both of these claims. He says nothing about any links to the Quraysh tribe and says that the Rufāʿah tribe is descended from the Sulaym Bin Hawāzin tribe. He argues that originally their only connection to the Juhaynah was based on geographical proximity, since both tribes migrated from the Hijaz to Egypt and then to eastern Sudan and were neighbors in the vicinity of the Sudanese port of Sawākin in the thirteenth century A.D. He reasons that the Rufāʿah tribe may have become closely affiliated with the Juhaynah elements in Sudan through intermarriage and political alliances and that this affiliation, rather than patrilineal descent, is the basis for the claim that the Rufāʿah are Juhaynah (Qāsim 1996:978-981).
Because of his genealogical orientation, Qāsim does not often acknowledge the possibility that a tribe may have composite origins. He is generally less interested than MacMichael in distinguishing "authentic" from "attached" lineages. Nevertheless, his treatment of the Rashāyidah tribe does capture this distinction. Drawing on the same source (al-Ħasan 1974) that I have used, he notes that the Qazāyizah, ʿAwāzim, and ʿUraynāt segments are "attached" administratively to the Zunaymāt section of the Rashāyidah (Qāsim 1996(Qāsim , vol. 2:971, 1029. In general, however, we cannot be sure that Qāsim's descriptions of "attached" lineages are more complete than MacMichael's.
In sum, the present state of research about the "attached" lineages in the Bedouin tribes in Sudan does not permit us to count them or estimate the percentage of the total number of such tribes who say that some of their segments are "attached." All that we can say is that "attached" lineages are known in an undetermined number of Sudanese tribes.

"Attachment" in Western and Northern Arabia
My chief source of information about the tribes of western and northern Arabia is the four-volume work Die Beduinen by Max von Oppenheim. This work is a survey of information from hundreds of articles and books about Bedouin tribes. It also contains his own field data. Von Oppenheim was very interested in tracing the geographical movements, political alliances, internal splits, and dispersals of Bedouin tribes over time. For this reason, he was keen to sift through the multitude of his sources to find statements about tribal composition, processes of tribal fission, and processes of fusion between one tribe and another. Whenever he found evidence that "foreign" lineages had attached themselves to a tribe, he was careful to note it. In some areas, he found many "attached" lineages but in one area -the part of western Arabia called the Hijaz -he found none (see von Oppenheim vol. 2 [1943]:309-355). "Attached" lineages are only slightly more common in north-central Arabia (also known as Najd). For example, the Sinjārah section of the Shammar tribal confederacy -which is spread across northeastern and north-central Arabiahas one "attached" lineage called the al-ʿUmūd (von Oppenheim vol. 3 [1952]:48-50). One of their neighbors to the northwest, in the Syrian desert, is the al-Khiraṣah tribe, which belongs to the al-Fidʿān sub-section of the Bishr section of the ʿAnazah tribal confederacy. The al-Khiraṣah tribe has four "attached" lineages, all of which are said to be fragments of the Wild Sulaymān tribe (see Figure 5) which broke away from their "original" tribe and joined the al-Khiraṣah in the early 1900s (von Oppenheim vol. 1 [1939]:115, 126 fn. 16; see also al-Jāsir 1980:83, 173, 184).
It is odd that the Hijaz should have absolutely no cases of this phenomenon, given that "attached" lineages are documented in most other locations near the Hijaz and Najd. As we will see, they are certainly present in Jordan, which borders the Hijaz and northern Arabia. On the northeastern flank of Arabia, in southern Iraq, there also some cases of "attached" lineages. The al-Muntafiq tribal confederation of southern and central Iraq has "attached" lineages, for example, as does the Albū Muħammad tribe of southeastern Iraq. It must be admitted, however, that most of the other large tribes in southern Iraq -such as the Banī Lām and the Albū Durrāj near the city of ʿAmārah -do not have "attached" lineages (von Oppenheim vol. 3 [1952]:415-495). It seems that, as one moves north from the Saudi border toward Baghdad and beyond, the number of tribes with "attached" lineages increases. In one case -that of the al-Ħaywāt segment of the Zawbaʿ tribe, due west of Baghdad in Abū Ghurayb -there are a total of sixteen lineages "attached" to a single tribal branch (von Oppenheim vol. 3 [1952]:228-230). Von Oppenheim's survey of central and northern Iraq identifies "attached" lineages in twelve Bedouin tribes -the Za-wbaʿ, al-Masʿūd, Banī Ṭuruf, Banī Ħasan, al-Fatlah, Zubayd, Khazʿal, Banī Ħukaym, Rabīʿah, al-Mujammaʿ, Jubūr al-Gharbī, and Jubūr Jabal Ħamrīn -as opposed to twenty tribes in the same general area that do not have "attached" lineages (von Oppenheim vol. 3 [1952]:173-395).
One possible explanation for the dearth of "attached" lineages in Najd and the Hijaz is that von Oppenheim's survey of the ethnographic record is incomplete. Evidence for this can be found in a work by Ħamad al-Jāsir, which lists a number of lineages that have "attached" themselves to the ʿAbdah tribe of the Shammar tribal confederation but that are not listed as "attached" by von Oppenheim. These "attached" segments are: the Saʿd lineage of the al-Baṭnayn clan of the al-Shurayfāt branch of the al-Mugharah tribe; the al-ʿUfr clan of the al-Khasraj branch of the al-Mugharah tribe; and the al-ʿUjāj lineage of the Āl Luhaymiṣ clan of the al-Shurayfāt branch of the al-Mugharah tribe. The tribe to which these three segments are "attached" -the ʿAbdah tribe -is based in northern Saudi Arabia, near the city of Ħāʾil (al-Jāsir 1980:297, 344, 443-444, 476, 700), that is, in northern Najd. There is also the case of the al-Ṭabarah clan, which is "attached" to the al-ʿImrān section of the Ħuwayṭāt al-Tahamah tribe and lives near the town of Ħaql, in the northern Hijaz. This clan originally belonged to the al-Fuḍūl segment of the Ħarb tribe (al-Jāsir 1980:497, 549). Al-Jāsir mentions another lineage that has attached itself to the Ħuwayṭāt: the al-Zalāyibah, which originally was part of the ʿAnazah tribe (al-Jāsir 1980:269). Finally, there are cases of lineages that, having broken away from their original tribes because of conflict, sought "refuge" with other tribes. For example, the al-Mawāhīb lineage lives as "refugees" among the Khizām clan of the Ballī (also called Billī) tribe in the Hijaz, although they are known to be from the ʿAnazah tribe originally (al-Jāsir 1980:726, see also Figure 5, above). If al-Jāsir's work had existed when von Oppenheim surveyed the literature on Najd and the Hijaz, he would certainly have listed these cases of "attachment" in the Hijaz.
I should note that there are social pressures which make the members of a Bedouin tribe in the Arabian Peninsula reluctant to admit that segments of their tribe are of foreign origins. Any suggestion that a tribe may be of composite origins is perceived as an affront to the identities of the tribesmen. The claim that a tribe was constructed through "genealogical mixing" (khalṭ alansāb) is an assault on that tribe's honor and reputation (cf. El Guindi 2012:549;Reilly 2013:383). Furthermore, there are strong religious prohibitions against giving a family's genealogical name to an outsider -for example, an orphan or a foundling -even if the outsider has been raised by that family since childhood. Such a foundling cannot even continue to reside with his caregivers past adolescence unless an incest prohibition has been created between him and his foster siblings through "suckling kinship" (El Guindi 2018:184, 190-194). If the foster mother suckled him during his infancy, he is regarded as a "son through suckling" by this woman and a "sibling through suckling" by the woman's genealogical children. But he does not have the right to take their family name (El Guindi 2012:551-553; al-Najjār and ʿAllām 2016).
These social and religious constraints on "attaching" non-agnates to a tribal genealogy are illustrated by a recent debate in Saudi Arabia about the composition of the Ħarb tribe. Several years ago, a non-Saudi writer, Fuʾād Ħamzah, suggested that the Ħarb tribe "does not descend from one ancestry; it is a combination of alliances comprising many elements, each with a different origin." He cited a well-known Arab genealogist, al-Hamdānī, in support of his position (Ħamzah 1968:147-151). One member of the Ħarb tribe strongly rejected this suggestion, arguing at great length that al-Hamdānī is wrong (see al-Ħarbī 1999:108-128). The tone and vigor of his argument suggest that he was defending his social and tribal identity, not simply trying to determine what the facts are. Another scholar, also a member of the Ħarb tribe, adopted a more moderate position in the debate. He pointed out that "historians always seek the truth, whether their goal is to correct errors or put the facts in their context. There is no point in stirring up emotions, reinforcing rumors, or omitting things here and adding things there if the goal is to diminish other people. Admitting the facts is a virtue." He supported the proposal that his tribe has absorbed populations who were descended from other ancestors: … the tribes…were not fixed in any one place over time. Rather, there were migrations which were characteristics of these tribes, for many reasons. One of them was the occurrence of struggles, quarrels, and wars among the branches of a single tribe that would end with some of the tribal branches leaving the tribe and attaching themselves to a different tribe. Other causes might be famines resulting from lack of rainfall. At other times the causes might be natural factors such as floods or epidemics, as well as other factors too numerous to mention here. (al-Ħarbī 2002:19) This author goes on to argue that … the origin of the Ħarb tribe is Qaħṭān [i.e., one of the two founding ancestors of all Arab tribes] but…it mixed with branches descended from ʿAdnān [i.e., the other founding ancestor of all Arab tribes] by making alliances with them, so that the tribe became linked to Qaħṭān through patrilineal descent and linked to ʿAdnān through residence, since it was living in the land of the Hijaz, between Mecca and Medina, in the area that was previously the homeland of the tribes descended from ʿAdnān as well as some tribes descended from Qaħṭān. (al-Ħarbī 2002:21-22) In making this argument he concurs with the opinion of Ħamad al-Jāsir, who was also a member of the Ħarb tribe (see al-Ħarbī 2002:543;al-Jāsir 1990). Ħamad al-Jāsir was well aware of a similar combination of descent ties with relations of co-residence: the case of the al-Dawāsir tribe. Al-Jāsir reports that a senior member of the tribe acknowledged that two of its segmentsthe al-ʿUmūr and the al-Ħiqbān -are not descended from the tribe's eponymous ancestor, Dawsar, but instead descend from one or more of his brothers (al-Jāsir 1990:206-207). In recording this, he is implicitly recognizing that these two segments are "attached" to the al-Dawāsir tribe by relations of co-residence, even though he does not say so explicitly. Von Oppenheim, who is much less constrained by tribal sensitivities, states openly that these segments are the remnants of a population that previously inhabited the territory of the al-Dawāsir before the al-Dawāsir occupied it (von Oppenheim, vol. 3 [1952]:124, 126).
It is noteworthy that all three of the parties in this debate who published their arguments in academic formats are members of the Ħarb tribe. It seems that scholars belonging to other Saudi tribes have hesitated to join the discussion. Those outsiders who believe, like al-Jāsir, that the Ħarb tribe is composed of multiple elements with different origins hesitate to say so, not wanting to appear biased. (For a detailed discussion of al-Jāsir's part in the debates about tribal genealogies in Saudi Arabia, see Samin 2015.) It should also be noted that truly scurrilous reactions to the debate have also been posted on various Saudi websites. I do not want to lower the level of discussion by including these reactions here. The sensitivities exposed by this recent debate show how difficult it can be to find objective discussions of "attachment" in Saudi Arabia. All of this makes us suspect that some cases of "attachment" in the Hijaz and Najd may have been concealed.
Let me conclude this phase of research by saying that "attached" lineages are known in western and northern Arabia but seem to be comparatively rare. Whatever misgivings one may have about the completeness of von Oppenheim's coverage, I have no strong evidence that "attachment" is common in these areas. I will now look further north and east. 3.1.3 "Attachment" in Jordan, Palestine, and Syria Von Oppenheim says that a great many tribes in Jordan, Palestine, and Syria have "attached" lineages. His information about this is drawn from Arabic sources and European travelers' accounts written in the nineteenth and early twentieth centuries. I summarize this information in Table 3. Note that the ethnographic present for this table is 1943, the year when von Oppenheim published the second volume of his review of the literature on the "Bedouin." The locations of the Bedouin tribes listed -and the names of these locations -have certainly changed since then, but I have made no attempt to update von Oppenheim's information.
In sum: "attached" segments are frequent in Jordan, Palestine, and Syria. My concentric model of tribes clearly applies to many of them.

Determining the Significance of "Attachment" among Bedouin Tribes
My attempt to determine the frequency of "attachment" has revealed a clear difference between Jordan, Palestine, Sinai, and Syria, on the one hand, and the Hijaz and Najd, on the other. The phenomenon of "attachment" is more common in the first four areas than in Najd and the Hijaz. Why? Von Oppenheim's survey suggests an explanation: "attached" lineages are found in Bedouin societies living in areas that have undergone rapid political and economic change. "At  tachment" is thus an adaptive response to rapid change. By detaching themselves from tribes in decline and attaching themselves to tribes whose political power is on the upswing, small descent groups can obtain protection and other social benefits from new patrons. We know that social change in Jordan, Palestine, Sinai, and Syria has always been more rapid than in the Hijaz and Najd because of the impact on local societies there of large imperial states based in Cairo, Damascus, Baghdad, and Istanbul. States having administrative and military centers of control in these cities have always sought to tax the flow of goods through Sinai, Palestine, and Jordan. They have frequently sent armies through these areas. Although the early Islamic empire had its headquarters in the Hijaz for a few decades, it then moved to Damascus and Baghdad early on, making Mecca and Medina into relative backwaters, politically speaking. Although the Islamic pilgrimage has made Mecca and Medina into important ritual centers, they are not centers of political and economic power. Thus, there are fewer external influences in the Hijaz than in Palestine or Jordan that would impact tribal composition. Although endogenous factors may also have led to social change among the Bedouin on the peripheries of Middle Eastern empires as well as near their centers (see Franz 2011), in general the social order in the Hijaz and Najd has been much more stable (if not stagnant) compared with the social order in Jordan, Palestine, and Syria. Von Oppenheim's discussion of particular cases of "attachment" certainly supports this general thesis. (See the case of the Jabārāt tribal confederation, described in Section II, above.) As von Oppenheim remarks (von Oppenheim vol. 2 [1943]:81), the Jabārāt case was typical of the tribes of southern Palestine. I think that rapid social change is one cause behind the phenomenon of "attachment" in this area.
Unfortunately, I cannot prove that there is a statistical correlation between this phenomenon and the pace of social change in all of the areas surveyed (the Hijaz, Jordan, Najd, Palestine, Sudan, Syria) without developing accurate measures of both variables. As I have said, it is not easy to measure the frequency of "attachment" in these areas because of the varying quality of the ethnographic record in these regions. Many cases of "attachment" may have been missed. In addition, accurately measuring the pace of social change in Najd, the Hijaz, and the Levant is also quite difficult. So, for the time being, the evidence of a correlation between "attachment" and rapid social change must remain anecdotal.
What else can we learn from the cases listed by von Oppenheim? If we examine some of the simpler cases we notice another characteristic: "attachment" always involves hierarchy, even if the core of the tribe and the "attached" segment are roughly equal in size. For example, the tiny al-Fuhaydāt tribe of Jericho consisted of only 5 households in 1943, while the segment that was "attached" to it -the al-Nuwayrāt -had 7 households. Nevertheless, the larger al-Nuwayrāt segment was regarded as "peripheral" in comparison with the al-Fuhaydāt tribe. In another case from Jericho, the core al-Ṣaʿāyidah tribe had only 8 households while the "attached" al-Samnah lineage had 15 households. In another region, east of Jaffa, the al-Sanāqirah branch of the al-Jarāminah tribe consisted of only 7 households, while the lineage that was "attached" to it -the al-Waradāt -had 10 households (von Oppenheim vol. 2 [1943]:45-46, 62). Such cases make it clear that population size does not determine which of two partners in a composite tribe is senior. The "attached" lineages in such cases are always politically marginal, regardless of how numerous they are. These examples also show that composite tribes are not assemblies of equals, even when there are only two partners. To put it metaphorically: this is a universe in which binary-star systems are not permitted. Composite tribes consist of one or more core lineages and "attached" lineages that orbit around them -even in cases where there is only one core lineage and one "attached" lineage.
One could go further and argue that the concentric model applies to all Bedouin tribes and not just to those that happen to have, at a particular moment in history, "attached" lineages. If this is true, then the "attached lineage" appears in Jordan, Palestine, and Syria under the social conditions that permit it to come to light but submerges in Najd and the Hijaz without really disappearing. According to this argument, even in the Hijaz the native model of "tribe" contains "empty slots" for "attached" lineages that are filled during times of rapid social change, when the politically marginal "attached" lineages of weakening tribes detach themselves from their patrons and "attach" themselves to new protectors.
If this is so, then the concentric model is actually the model of a "deep structure" of all Bedouin tribes. Some of its components (the "attached" lineages) are instantiated in rapidly changing social environments but are not realized when and where the social order is stable. If we grant that the concentric model is also a "deep structure," then it is also possible to agree with Lévi-Strauss (see Lévi-Strauss 1963b:11-18, 21, 23-25) and argue that this description of a deep structure is actually a matter of discovering an object that is already present at the unconscious level. The Bedouin of the eastern Arab world do not say that their tribes consist of core and peripheral elements, but when they organize their tents into residential groups, they make sure that political leaders of each camp put their tents in the center of the camp (cf. Young 1996:55-56, 77-78; see also Alon 2016:85). Could this be another expression of an underlying structure which literally places the politically powerful in the center of social space and the politically weak at the margins? Using a model of spatial topography to represent kinship relationships is by no means unique to the Bedouin, of course; members of many other societies do this as well (see Hamberger 2018).

What Is the Native Model?
We might also ask how the Bedouin themselves model their tribe. It turns out that they use the metaphor of "genealogical tree" (shajarat nasab) when they describe how a tribe's component lineages are connected (cf. El Guindi 2012:548-49). Their visual representation of their genealogical relations is literally a tree, with a prominent thick trunk in the center and leaves with the names of the smallest lineages written on them at the ends of the tree's "branches" (see Brandt 2016:122;Samin 2015:3). Many such "trees" have been posted on the Internet. For example, members of the Ballī (or Billī) tribe of northwestern Arabia have posted their "genealogical tree" (see http://wadod.org/uber/uploads/425_01259081667.gif), as have the ʿAnazah tribes of central and northern Arabia (see http://www.3nzh.com/vb/showthread.php?t=62767). The al-Marāshīd section of the al-Barāzāt branch of the al-Suhūl tribe of the United Arab Emirates and Qatar have designed an especially elaborate tree that has branches twisting in many directions (see https://twitter.com/onaizah1424/status/907118077741019136). The largest collection of these "genealogical trees" that I have been able to locate at present (July 2018) is posted on a Twitter account at https://twitter.com/onaizah1424. This account appears to be maintained by members of the ʿAnazah tribes of Saudi Arabia, Jordan, and the Gulf. It shows some 240 "ge-nealogical trees," each one with a thick central trunk, several branches, and leaves labeled with the names of individuals or families. The genealogical relationships portrayed belong to a wide range of tribes -including, for example, the Ħarb and Banī Tamīm -and do not pertain exclusively to the ʿAnazah tribes. It appears that most of these images are purely digital, which means that they cannot be especially old. Some of them, however (for example, the "tree" for the Āl Fāyiz family of the Āl Ghazzī clan of the al-Fuḍūl division of the Banī Lām section of the Ṭayy tribe and the "tree" for Āl Khuwayṭir section of the Banī Khālid tribe of the ʿAnazah tribal confederation), are photographs of "trees" that were printed in color on paper, which makes it possible that they were been made before the widespread use of the Internet. Many other examples can be found outside of this collection. At least one such "tree" was printed before 1948 and portrays a prominent family in Lebanon (al-Khūrī 1948:120). The same kind of image was painted on a house in southern Lebanon (see the front cover of Eickelman and Piscatori 2004). Clearly this kind of "genealogical tree" is part of a widespread and well-established folk tradition in the eastern Arab world.
Some of the Rashāyidah have also constructed a "genealogical tree" and have posted it on the Internet (see http://www.bani-3abs.net/aa/showthread.php?t=91572). As I will explain, however, there are reasons why it might be imprudent to present this particular "genealogical tree" as a native model of the Rashāyidah tribe.
First of all, the "genealogical tree" that is posted at the above address is both visual (since it incorporates the image of a tree) and written (since the names of clans and lineages are written on the tree's branches). In 1980, when I was conducting my fieldwork among the Rashāyidah in Sudan, almost all of their genealogical knowledge was oral and was passed from one generation to the next by word of mouth. This meant that the genealogies they shared covered only lineages and clans in detail, not the entire tribe. Since genealogies were not written down, they were largely the products of informal consensus by small groups. Now that more and more Rashāyidah are literate and can write down their genealogies, one suspects that disputes about some of the details have increased. This has tended to occur in other Middle Eastern areas after literacy rates increase (see Aswad 1971: 51). In short, the "genealogical tree" model on the Internet is more subject to debate than the oral models of clans and lineages that circulated through speech. Second, this visual "genealogical tree" was constructed very recently. It is dated 1431 hijri, that is, some time in 2010 on the Gregorian calendar, many years after my fieldwork among the Rashāyidah. Third, it does not represent only the nine main divisions of the Rashāyidah tribe that appear in Figure 4, above. Rather, it purports to show "all people who belong to the Banī Rashīd tribe, including the main branches in all of the Arab countries such as Saudi Arabia, Sudan, Kuwayt, Egypt, the Sultanate of Oman, and Jordan." To accommodate all of these tribal segments in one tree, it represents them as fourteen branches protruding in all directions from a central trunk. Clearly it contains many more clans and lineages than any model of the Sudanese Rashāyidah tribe by itself would contain. Fourth, the "tree" is still in the process of construction, as the designer and author -one Saʿd ibn Saʿūd al-Shuwayliʿī -makes clear in his dedication. He says "I dedicate this…to all descendants of the Banī Rashīd tribe…and ask them to excuse me if there are any errors or omissions. God willing, the genealogical tree will be brought up to date from time to time. To make comments, corrections, and additions, please contact me via the Internet of the Banī ʿAbs, which is a fortress of glory and history. Please accept the greetings of your brothers." The final sentence of this dedication makes reference to the pre-Islamic Banī ʿAbs tribe, which some of the Rashāyidah have begun to claim as their tribe of origin.
As we can see, the model presented on the Internet is the fruit of a collective effort made by members of the many groups who call themselves Rashāyidah or Banī Rashīd. These groups have only recently started to communicate with each other and explore the possibility that they are historically and genealogically related (see Young 2006). The new, virtual community that they are forming is much larger than the Rashāyidah tribe of Sudan and depends on electronic communication -rather than face-to-face discourse -to create its self-image. This is not exactly the kind of image that Bedouin societies created in earlier times. For all of these reasons, I am hesitant to reproduce this particular image here or make it into an authoritative and final representation of the Rashāyidah's native model.
For my purposes it might be better to strip this recent Banī Rashīd model of much of its content -that is, the names and number of clans and lineages represented -and focus on the structure of the image. Although this image is a two-dimensional painting, it is shaded and colored to represent a three-dimensional tree. In this regard it is very similar to the "genealogical trees" constructed by other tribes in the Arabian Peninsula. Despite much artistic variation in coloring, size, and shape, all of these trees have the same structural elements. By including both a thick, central trunk and tapering branches, these images capture both the hierarchy in the branching hierarchy model of tribes that was critiqued at the beginning of this paper and the center/periphery distinction in the concentric model. Of course, one crucial element in the concentric model -the relationship of "attachment" -is not captured in these images of trees. To include this element, the trees would have to portray branches that have been grafted onto the trunk, not only those branches that grow naturally from the base of the tree.
Lest we have any doubt that Bedouin models of tribal genealogies incorporate all three dimensions -i.e., depth and breadth in space as well as genealogical length -let us take note of the kind of tree portrayed in these images. It is not just any tree (Arabic: shajarah). It is a dawħah, that is, a grand, lofty tree with a rich profusion of leaves, twigs, stems, branches, and limbs that project in every direction from a thick trunk (Qāsim 1972:260). The word dawħah is etymologically related to the verb dāħa, "to be big (belly); to be lofty (tree)" and the verb tadawwaħa "to be distended (belly)" (Hava 1899:212). Clearly the meaning of the term dawħah entails width, breadth, and height. It is also the term used by some Arab genealogists for "family tree" (see Ħamādah 2000;Wehr 1976:297) instead of the generic and prosaic compound noun, shajarat nasab ("genealogy tree"). The same term, dawħah, with the accompanying image of a majestic tree with branches spreading out around a massive trunk, is used today by the Khaṭāṭibah tribe in the Jordanian village of Kufrinja, where I conducted fieldwork in 1994 (see the webpage about the genealogy and history of the Khaṭāṭibah tribe posted by Ṭālāl Muħammad Maħmūd al-Khaṭāṭibah at http://alkhatatbah.blogspot.com/p/blog-page_10.html).

Conclusions
My discussion of the formation of the Sudanese Rashāyidah tribe sheds light on the formation of other Bedouin tribes. The Rashāyidah case is not exceptional. The composition of the Rashāyidah tribe is comparable to that of 46 other tribes in Jordan, Palestine, and Syria, as well as some tribes in Saudi Arabia and Sudan. This means that the history of Bedouin tribes is not just a matter of movement from one territory to another. It is also a matter of core lineages assuming leadership positions and offering protection to the peripheral lineages that they attract during the period of migration. The process of attachment and detachment is central to the historical formation of Bedouin tribes generally. To model these processes, the concentric model presented at the beginning of the paper is of greater utility than the branching hierarchy model of lineages often used in ethnographies of Bedouin societies. What is more, it may represent the Bedouin conceptualization of their tribal structures more accurately than the branching hierarchy model does.
Why should we limit ourselves to Bedouin tribes when applying this concentric model? Bedouin societies are historically associated with nomadic pastoralism, even if they no longer practice it. They are less likely than sedentary societies to have formal, legal claims to land. In contrast, local descent groups that are not Bedouin usually have a more direct connection to land and immovable property (see, for example, Antoun 1972). For this reason, a sedentary agnatic group's ties to other local agnatic groups are not so easily broken or fabricated. As Barbara Aswad argues, sedentary rural societies do not permit "foreign" lineages to graft themselves onto established, politically powerful or landowning tribes (see Aswad 1971:52).
We must always bear in mind that the "attachment" that I am discussing is not just an abstract connection between one descent category and another. It is a concrete association between a local descent group and other local descent groups. In Bedouin societies, local descent groups may have claims to use plots of land for grazing and subsistence agriculture but these claims are not always exercised. In practice a household or a group of households may not use a particular plot because of varying climatic conditions (especially rainfall). Furthermore, such claims do not amount to formally acknowledged and jurally recognized rights. This means that if a nomadic pastoralist lineage breaks away from a clan or tribe it does not forfeit jural rights to exclusively cultivate and inherit land because it did not have such rights to begin with. At most it forfeits claims to use particular lands seasonally. A local lineage may be willing to forfeit these claims when they are not equal in value -from the lineage members' point of view -to the gains that they can make by changing descent group affiliation. (For discussions of this among Libyan "Bedouin," see Behnke 1980and Peters 1960, 1990.) Hence it is easier for the members of a Bedouin tribe or clan to cut their ties and attach themselves to other descent groups than it is for sedentary cultivators to do this. One would not expect the concentric model to apply to tribal societies in Yemen, for example, where tribes do not move seasonally and where they tend to have exclusive control over and rights to agricultural territories (see Adra 1982:104;Dresch 1994;Varisco 2017:231, 236, 240-41;however, see Brandt 2016:118, 132, 136, 137).

Transliteration of Standard Arabic in Roman Characters
The system for transliterating standard Arabic in Roman characters that has been adopted here (see Table 4) is very close to the systems used by the U.S. Library of Congress and the International Journal of Middle Eastern Studies. One difference between these systems and the one used in this paper concerns the voiceless pharyngeal constricted fricative ‫,ح‬ which is repre-sented here as /ħ/ rather than as an "h" with a dot below it: /ḥ /. The motivation for this change is to separate this symbol from four other symbols that have dots below them (/ṣ /, /ḍ /, / ṭ /, and / ẓ /) and that represent a set of "emphatic" or pharyngealized consonants. Although these consonants are, like /ħ/, pharyngealized, their points of articulation are palatal, alveolar, or interdental -i.e., much farther front than the point of articulation for /ħ/ (see Ingham 1994;Procházka 1988). To make it clear that /ħ/ does not really belong to the set of "emphatic" consonants, it seemed wise to use a distinctive symbol for it. Throughout, I have changed the renderings of Arabic words that appear in the works by MacMichael, von Oppenheim, and some of the other authors cited in this paper so that they conform to this system. One reason for doing so is to make them consistent throughout. The authors cited frequently reproduce the colloquial pronunciations of words rather than the forms used to write them in Arabic; thus, they provide transcriptions of speech rather than transliterations of written words. This often yields several inconsistent ways of representing the same tribe or lineage name and introduces some confusion into the ethnographic literature. Furthermore, it breaks the link between these names and their written forms in standard Arabic, making it more difficult for Arabic speakers to recognize them. Note that I have omitted all case endings to make the transliterations more readable. This is common practice.