Between diversity and hegemony: Transformations of kinship and gender relations in Upper Yemen, seventh- to thirteenth-century CE

ABSTRACT

The present analysis explores for the first time main changes in gender and kinship relations of Upper Yemen between the seventh and the thirteenth century CE. An initial section outlines main continuities throughout this period. The article's three main sections then investigate, first, transformations of kinship terminologies after the seventh and eighth century. Second, changes in reckoning genealogies, as well as descent and inheritance rules are scrutinized for the early tenth century. Third, structural changes in marriages patterns are outlined for periods before and after the thirteenth century. These investigations are carried out for the region's rural tribal majorities. Moreover, they also include a substantial consideration of the region's main social and demographic minorities. They were elite groups of Persian and more importantly, of West Arabian Muslim origins. In addition, important Jewish, Christian and slave groups of socially subaltern status are taken into account. These analyses conclude with the insight that elite groups eventually succeeded in establishing their fragile regimes, assisted by genealogical discourses and scriptural conventions. Meanwhile, the socially subordinate minority groups came under increasing pressures of concubinate, conversion, and demographic constraints. For the agricultural tribal majorities of the region, these developments amounted to a continuous interplay between an enduring diversity of kinship and gender relations, and a gradual growth of hegemonic influences.

KEYWORDS:

• Historical Upper Yemen
• changing kinship and gender relations
• genealogical reasoning, reckoning descent and inheritance rules
• transformations of marriage patterns
• diversity among tribal majorities
• elite groups' hegemonies & subaltern minorities' constraints

This article discusses transformations in kinship and gender relations between the seventh and thirteenth centuries CE in the Northern highlands of Southwest Arabia (historical Yemen).¹ The analysis focuses on three transitional periods: (1) continuing Arabization and Islamization (seventh to eighth century); (2) tribal consolidation against early Zaydi² rule (ninth to tenth century); and (3) Zaydi consolidation (mid-thirteenth century) (Madelung 2002; Serjeant 1969). For these periods, we scrutinize the prevailing kinship terminologies, dominant modes of reckoning genealogies and descent, and common marriage patterns. These transitions were part of pre-colonial Islamic history, which itself had emerged from complex trajectories of pre-Islamic ancient South Arabia.³ By the sources available, these three transition periods allow to explore how kinship and gender changed in relation to each other and with regard to society at large, as outlined in the introduction to this theme issue.

The analysis will engage with stratified societies, exposed to scriptural traditions and based on mixed economies with agriculture at the core. Changing, fragile elites represented the upper strata, including administrative specialists. Other experts in crafts and manual services were the heterogeneous lowest strata. Between the upper and lower ends of internal hierarchies, the majority populations largely practised agriculture. Particularly in the Eastern and Western fringe areas of the plateaus and highlands, this combined with greater elements of pastoralism and hunting.

Traditions of writing (in languages and scripts other than Arabic) had already existed in ancient South Arabia for many centuries. Together with related versions of fragile statehood, these legacies had been associated most closely with upper and elite strata – that is, legacies that now also were contributing to Islamic history in the area. Simultaneously, Southwest Arabia was not merely connected to its regional prehistory, but also to its wider contemporary present in the overall settings between the Gulf and the Hijaz, East Africa, the Mediterranean, Central (Persia) and South Asia.

Methodologically, this study combines insights from Yemen-related historiography, archaeology and philology with the more recent advances in the anthropological analyses of kinship and gender relations (for example, Moore 2013). As part of these efforts, the authors seek to not discard valuable elements of kinship analyses, but by following the works of Jack Goody (for example, 1990), Maurice Godelier (2012) and others (cf. the editors' introduction), to elaborate and apply them in fields where historians require anthropologists’ supportive collaboration, and vice versa.

Tribal and non-tribal affiliations

Determining the social meaning and content of names and designations for majority population sections and their members first requires a consideration of their primary subsistence activities.

By the seventh century, the primary zones of agriculture in the region had shifted from the Eastern steppe fringes to the highland and mountain regions and from large-scale irrigation to an enhancement of the small-scale irrigation that had already existed in the highlands for quite some time. When Islam gained first influence in the Yemen during the seventh century, the main agricultural legacies of ancient South Arabia had already mostly fallen into ruin. The traditions of large-scale flood irrigation by great dams, prevailing in ancient South Arabia and its state cultures for 1600 years, with most of its centres in the plains to the East, had disintegrated around the turn from the sixth and seventh centuries (Hehmeyer 2019). The final collapse of the great dam of Ma’rib, the centre of Saba (late sixth century) is famously mentioned in the Qur’an (34, 15–17).

Survivors and descendants of the former population in the low-lying Eastern centres had become nomads or were emigrating with the Islamic conquests to Northern Arabia and North Africa (Robin 1991), while most moved to the fertile highlands and mountains. Intersecting but different techniques of small-scale cultivation had preceded irrigation with large dams in the Eastern plains but also in the highlands and mountain regions where they dated back perhaps even to the third millennium BCE. The majority population in the seventh century thereby lived in highlands and on mountains that became one of the most fertile regions not merely of South Arabia but of Western Asia in general. Internally, this promoted the emergence of local alliances against outside interests and regional rivalries.

These agricultural priorities implied several social consequences. Most importantly, the need for regular cycles of small dam and terrace repair required meticulous observance and local cooperation, combined with well-experienced efforts toward improvement of soils and seeds across generations learning from accumulated experience (Kopp 1981). Households as the centres of mastering these tasks represented the logical core units of production and consumption. Besides ethnographic sources, a few textual (Gingrich and Heiss 1986, 172) and archaeological indications testify to the same basic point. Since the seventh century at the latest, the agricultural majorities of the Yemen mountain and highland regions had been organized in household units, or in what others have addressed as household or domestic economies (Mundy 1995) or domestic modes of production (Sahlins 1972).

All available evidence indicates that these agricultural households were the social and demographic majority’s basic unit, simultaneously linked with others through connections of marriage and descent, and often grouped in hamlets and villages. Pressures, tensions and competition from outside for these fertile agricultural resources, and internal frictions among the majority of land-holding households themselves, could be further reduced by associating territorial neighbours into larger groups of co-resident inhabitants. These territories of associated neighbours had invisible borders that kept outsiders’ claims at bay and prioritized claims by residents within those borders. The authors thereby continue an argument made in an earlier issue of this journal (Gingrich 2015), addressing the relative continuity of territorial neighbourhood associations linked with collective or generic names since early Islamic centuries in the Yemen. The continuous link between a collective name and a fairly stable rural territory lies at the core of the Yemeni (and Qur’an’s) Arabic noun qabīla, which can scarcely be translated by any other term than tribe. Paul Dresch (1989) was the first to explicitly point out these long-lasting links, subsequently elaborated by a number of other authors (for example, Brandt [2012]; Varisco [2017]). Minor or major groups of previous residents emigrated, and likewise, new in-coming settler groups were accepted. Yet the actual spatial contours did not change very much. The relatively stable boundaries of these collective territories and their respective names thus functioned as ‘containers’ for changing social contents associated to very fertile lands.

The rather frequent emigration and immigration processes had consequences of fusion, fission and reconfiguration. If they did not migrate to other parts of Arabia and North Africa but stayed elsewhere in Upper Yemen, those who left their territorial tribal residence and its affiliation were confronted with three other possibilities. Either they were hierarchically marginalized into groups of inferior status, or they could maintain some kind of tribal status to enter another tribe if they were accepted there. Alternatively, they could also amalgamate with other such groups, if they were large enough, into a newly configurated quasi-tribe. Yemen’s primary tenth-century historian, al-Hamdānī ([1368] 1949, 1968, 1980), mentions five relatively large quasi-tribal groups of this kind for Upper Yemen (Heiss 1998, 125–165), that is, heterogeneous groups of various backgrounds but without any, or any clear, genealogical background.

Among the tribal territories and their fairly stable boundaries, three major tribal spheres of unequal size may be distinguished for the seventh and eighth centuries. Some of the tribes of Upper Yemen by then had already existed for longer (Hamdān [Ḥāshid and Bakīl] on the plateau and the mountains in their immediate Western vicinity), some had established themselves more recently (Khawlān b. Quḍā’a in the Sa’da basin), while others were still in their early formative phases, that is, in the least accessible westernmost mountains (Meissner 1987).

The name of a person from highland Yemen’s majority population in those days would therefore by necessity include a reference to their household, and to the name of a tribe or tribal section the household was affiliated with. Simultaneously many – especially among the tribal and quasi-tribal elites in the region – saw themselves as descendants of ancient South Arabian polities’ eponymous ancestors. A tribal territory’s single name thus would often be identical with that of a pre-Islamic eponymous ancestor. This is testified by many individual names of historical persons, as featured in the works of al-Hamdānī and a few other earlier or later Yemeni authors. During the early decades of Islam’s rise in the Yemen, that past was a few centuries distant and had not been forgotten. Through their imaginary connection to eponymous ancestors, tribal boundaries thereby emphasized some basic shared identification and status for resident insiders, while they primarily helped to reduce external claims on local resources and to strengthen elements of internal cooperation. Main forms of peacefully gaining access to agricultural land thus were the channels of inheritance and marriage.

Complementing this outline of social settings in the Upper Yemeni highlands also requires a short consideration of religious and ethnic minorities, despite the scarcity of historical accounts. Records mention the existence of Christians and Jews in Yemen from around the third century. A Persian elite known as Abnāʾ is documented since they made Yemen a Sasanian province at the end of the sixth century. The strongholds and overall situations for each of these communities differed. Yemeni Jews did not always have a minority status. Elite members and at one time the last pre-Persian ruler had embraced Judaism. Jews’ pre-Islamic presence also included a wide range of political, scholarly, commercial or crafts’ expertise, but in later periods also that of rural villagers and tribal protégés (Newby 1988). Christians seem to have lived together more collectively in smaller or larger communities such as Najran, with specializations in certain crafts, trades and agricultural professions (Nebes 2008). In their major centres Christian semi-autonomous communities lived under the protection of surrounding tribes, while maintaining far-reaching connections to other places of Christian learning in Arabia and Byzantium. Especially Christians but also Jews had important transcultural functions as translators (Irfan 1979). Persians entered the Yemen mainly on a military mission fighting the Aksumite forces, which were about to conquer Yemen. After their eventual success, Persian officials continued their rule and administration in Yemen. The Sasanian powers brought a workforce for mines, irrigation specialists, soldiers and administrative staff from Persia to their new satrapy (‘province’) of Yemen. Probably the early Islamic learned stratum in Yemen also partially developed from the administrators. Many Persians stayed in Yemen after its incorporation into the Islamic domain, and intermarriage with Yemeni women was not uncommon. Henceforth, their descendants resided in Ṣanaa, Ṣaʿda, and smaller towns or villages in the Yemeni highlands. They benefited from an elite status, held property and continued to be active in Yemen’s mining and trade (Kommer, forthcoming).

These basic features of tribal majority and of non-tribal affiliations would largely endure from the seventh to the mid-thirteenth century. Against this background, the following three sections will now examine dimensions of kinship and intersecting gender relations for the three transitional periods addressed in the introduction.

Kinship terminology in Upper Yemen by the seventh/eighth century

Several norms and preferential speech practices for kinship terminology are addressed in ancient South Arabia’s epigraphy and in the Qur’an. This is why early Arabization and Islamization in the region have to be taken into account here. The section’s main argument will seek to demonstrate the emergence of a wide range of kinship terminologies in Upper Yemen, between the two poles of cognatic and Sudanese terminologies.

For the landscape of al-Jawf in Eastern Upper Yemen, sources already indicate a prevalence of Arabic for about 300 CE (Robin 1997, 181). Vernacular forms of Arabic were gaining significance in Upper Yemen since then. The last Ḥimyarite inscription is dated 554 CE (Brunner 2005, 143). By the seventh century, the extent of Arabization must already have reached a fairly wide albeit uneven influence in the upper highlands and the Western mountains. Minority languages such as Ḥimyarite⁴ and Persian continued to be spoken and read among fractions of the elites, or Hebrew and Aramaic among the relevant monotheist minorities. Yet various combinations of vernacular Arabic with remaining elements of Ḥimyarite were most common (Robin 1991, 83). With Arabic as the scriptural version of revelation in the Qur’an, Islamization certainly gave Arabization a decisive new impetus and some institutional basis around mosques and in the few cities. This also reinforced distinctions between scriptural standard versions of Arabic and a heterogeneous variety of vernacular forms spoken (and occasionally also written) in most rural areas.

When the last (fourth) Persian governor of Sanaa had converted to Islam in 628, this contributed to the claim that Yemen ‘embraced Islam in one day’. In fact, that day was followed by three and a half centuries of various phases of opposition, apostasy, unrest and instability. It therefore seems safe to address the arrival and establishment of Yemen’s first Zaydi Imam in the late ninth century as the beginning of a second wave of Islamization in Upper Yemen and the Western mountains with Sa’da as the main Zaydi theocratic power base (Heiss 2014). In Upper Yemen that second wave had its main support in the few cities and some additional rural strongholds. Similar in that regard to the first wave of the seventh century, it encountered widespread support but also substantial dissent and opposition.

Between the seventh and the ninth centuries, Arabization was therefore somewhat more advanced and more widespread than Islamization in the region. Existing levels of Arabization facilitated the first introduction of Islam in the early seventh century while the introduction of Islamic norms and values further promoted scriptural Arabic in social groups that were its strongest supporters. The thesis that early Islamization in the Yemen unfolded primarily ‘from above and from below’ (Gochenour 1984) has been widely accepted in Yemeni studies. This implies that large factions among the existing elites, but also wider segments among the lowest social strata largely supported Islamization,⁵ while by comparison the agricultural and tribal majorities seem to have been slower and sometimes even reluctant to fully embrace it. Inside tribal hierarchies, again, the tribal poor and the tribal elites may have been more enthusiastic than others. For the elites, the promotion of socio-political coherence and integration by the new belief system offered added value, while equality for all before God implied promises of justice and upward social mobility for people of low status and the poor.

These insights have implications for kinship and gender in their terminological and legal dimensions. They lead to the suggestion that, together with Arabization and Islamization, a major specific form of kinship terminology was enhanced or took over in some parts of Upper Yemen, which socio-cultural anthropology classifies as the Sudanese system (Godelier et al.1998, 9). If Islam was gradually accepted from above and from below, and if this spread from the few urban centres to some tribal or rural elites (and perhaps also to some among the low-status minorities), then this also promoted the partial dissemination of Sudanese kinship terminology in Arabic. By consequence, around the eighth century, Arabic kinship terminology and its Qur’anic interpretation and usage most probably gained influence in elite residence areas such as Sa’da and Sanaa, as well as among some of the major tribal leaders’ residence, and in other centres of crafts and commerce.

Among the tribal majorities, however, only some would have followed up at one end of a wider spectrum. At its other end, elements of older kinship terminologies certainly also continued to exist and thrive. These older terminologies seem to have been formally cognatic: based on the analyses of epigraphic evidence by various Russian specialists, Walter Dostal (1989) identified a substratum of formal cognatic terminology, against a gradual elite prevalence of patrilineality among Ancient South Arabia’s kings and royal dynastic families. In this perspective upon the eighth century, the unavoidable co-existence of ‘ancient’ and ‘very recent’ forms of kinship terminologies would identify the two extremes of a wider range of heterogeneous tribal kinship terminologies for Yemen’s highland and mountain regions, with a spectrum of various vernacular and local intermediate forms.

As their main criterion, Sudanese kinship terminologies include collateral bifurcation one generation above Ego – that is, a clear distinction among terms of reference between Ego’s relatives on the mother’s side from those on the father’s side. That distinction might be somewhat blurred in the second generation before/above Ego, but it is continued for Ego’s own generation and for those that follow. The relevance of kinship terminology for actual kinship practices and relations should not be exaggerated; several authors have emphasized its inherent limitations. Yet if compared to prevailing forms of kinship terminology in other language-encoded systems, some important differences stand out, highlighting the specificities of (South) Arabian contexts. In its Arabic versions the Sudanese system explicitly emphasizes the distinction between father’s and mother’s side, while seniority within each side and generation is not addressed by any primary terms.

That explicit distinction between father’s and mother’s side has the obvious potential for more generalized differences and hierarchies between male and female. This goes together with a non-descriptive, but classificatory terminology grouping together seniors and juniors within the same generation – and by consequence, with limitations to establish hierarchies through them. Southwest Arabic kinship terminologies of the Sudanese type therefore minimize hierarchies and encourage equal status within generations, but allow for clear terminological distinctions and hierarchies among male and female (Godelier 2012; Parkin 1997).

The basically Sudanese terminological distinction between the patri-and matrilateral side was not necessarily translated dogmatically into recurrent interactions and daily speech routine. For instance, that distinction could be blurred and attenuated by practices of teknonymy – widespread anywhere in South Arabia’s ethnographic present – that is, referring to parents by the names of their children (kunya), and vice versa to children by the name of either one of their parents (by patro- and matronymy). This left much room for diverse practices, some of them solidifying into diverse local terminological traditions.

Cognatic terminologies tend to indicate somewhat wider social spaces for female agency, by emphasizing equivalence rather than distinction between male and female. The existence of Sudanese terminology during early Islamic times in Upper Yemen should not, however, be ignored or played down. First, it facilitated distinguishing elites from the rest through patrilineality. Second, this promoted claims in any succession of office among the regional elites. Third, it implied some specific proximity to the early Qur’anic emphasis – in a linguistic and legal sense – on the male side of the world and thereby could indicate a corresponding religious affinity. It is therefore probable but remains to be proven in detail that these patrilateral elements continued and grew in relevance during the first Muslim centuries in Upper Yemen, especially among the tribal and non-tribal elites. In moderate or more radical forms, this implied the conceptual and legal de-valorization of women’s status in corresponding elite sectors of Upper Yemen’s hierarchical societies. It would apply not only for the non-tribal elites claiming superior status to the tribes, but also for those tribal elites associated with them (for comparable cases see also Duindam 2021).

To some extent, this also had its indirect and direct repercussions on the religious minorities of Jews, Christians, Zoroastrians and followers of local belief systems in Yemen. Jews were the largest religious minority in Yemen.⁶ With the coming of Islam and the formation of the new albeit fragile Islamic state, interreligious relations had to be renegotiated and ways of cohabitation reorganized. The Prophet’s Medina model already included ahl al-dhimma (‘protected people’) and corresponding taxation. This was followed by the establishment of the Zaydi Imamate in Northern Yemen at the end of the ninth century, by a similar treaty for Jews, Christians and the surrounding tribes and communities. Socially, Yemeni Jews had much in common with local majority Arabs. To an extent they continued to share with them Arabic language and dialect skills, and their lives also were closely interconnected through economic activities. Outside of urban centres they lived alongside to one another. Although religious and family life was strictly separated between the communities, considerable (South) Arabic influence can be traced in the otherwise Hebrew-dominated ritual terminology of Yemeni Jews (Tobi 1999, 209–210). Unlike in Islam, polygynous marriages were not tolerated in Jewish and Christian doctrines. In addition, having concubines was at least normatively condemned by these two doctrines. Through corresponding practices observing the doctrines in this regard to a certain extent, fewer children in Christian and Jewish households also affected demographic dynamics and micro-economic conditions. Christians and Jews preferred marriage inside their own socio-religious communities. Minority female hypergamy was acceptable though from Muslim perspectives, and whether on a voluntary basis or under pressure, it seems to have occurred repeatedly. While in theory at least a Jewish or Christian wife could maintain her original belief after marriage, the children of such a hypergamous marriage would be Muslims. In the long run, this is one of several factors contributing to the gradual shrinking of Jewish and Christian minorities in Southwest Arabia, and the eventual disappearance of Zoroastrianism.⁷

Statements regarding other non-tribal and low status groups and their social and kinship relations are difficult to make. Slaves were already relevant enough in the Yemeni highlands under early Zaydi influence, to be addresses by the first Zaydi Imam (early tenth century) in his collection of legal decisions and opinions (Moorthy-Kloss 2019, 49, note 141). In line with what has been outlined here for the Yemen’s second wave of Islamization, these slaves in all likelihood worked in wartimes for military purposes (Moorthy-Kloss 2019, 72) and during peaceful periods, in the fields and domestic spheres of non-tribal and tribal elite representatives. If so, relations to their homes and families being cut off they barely had any relatives but had to cope with growing elements of gendered abuse and its consequences. It is even more uncertain whether low-status Muslim groups, later collectively identified, for example, as Ahl al-Thulth or Banu ‘l-Khums, already existed on the plateau at the time.⁸ Whenever they emerged, their social origins may have been ex-tribal emigrants, converts (from Christianity and Judaism) and freed (children of) slaves.

This concludes our discussion of a heterogeneous range, between cognatic and Sudanese extremes, of kinship terminologies during Upper Yemen’s first wave of Islamization (seventh/eighth century) among regional majorities, and of Islamization’s potential effects upon denominational and social minorities. We may now turn to the next transitional period for which, beyond terminology, the work of al-Hamdānī permits some crucial insights into other important aspects of kinship and gender relations.

Genealogies, descent, and inheritance around the early tenth century

Anthropology insists on basic analytical differentiations between genealogy, descent and inheritance, and on the fact that there is ‘no overriding necessity why in any given society’ they ‘should always be homologous in character’ (Goody 1990, 363). Descent rules usually follow general local interests with fewer elements of standardization. Genealogies may intersect with descent but usually relate to special group interests with more formalized trajectories of transmission.

In his major works, the ‘Yemen’s tongue’ of the tenth century al-Hamdānī (280–334/893–945) presented a complex series of overviews on tribal genealogies of Southwest Arabia through basically patrilineal terms. They primarily related to the eponymous ancestors of specific tribal groups during or before his time. The tribal genealogies connected such an eponymous ancestral name of the more recent past through various links and chains of mostly male names with the ancestral source of South Arabian tribes, that is, Qaḥṭān. From the authorities and sources the author quoted as evidence, these patrilineal chains of tribal ancestors leading back to Qaḥṭān were obviously not even common knowledge among al-Hamdānī’s readers from the small literate upper tribal and quasi-tribal leadership groups. Wherever possible, the author linked a larger tribal group collectively, through their leaders and their past, by genealogical proximity to other tribes in the area. The relevant sections in al-Hamdānī’s work thereby sought to systematize and disseminate specialized genealogical knowledge of tribal elites, leading and representing large tribal entities on their behalf. The explicit patrilineal dimension in these genealogies was in line with two elements already addressed here. It was compatible with the regional versions of Sudanese kinship terminology, and it built upon the older legacy of patrilinearity among pre-Islamic royal South Arabian dynasties. Moreover, on an abstract level the Ahl al-Bayt’s establishment in the Sa‘da region since the ninth century also was compatible with these tendencies: although the new Zaydi rulers represented opposing genealogical interests, they also upheld formalized, detailed patrilineal genealogical chains.

The Ahl al-Bayt were descendants of the Prophet Muḥammad. In Northern Yemen they represented the elitist recruitment base for the Zaydi Imams’ succession and support. By their genealogical adherence to the Prophet’s family, they represented a Northwest Arabian background. Since the conquests in early Islamic times, the genealogical differentiation between Southern and Northern Arabs had become widely apparent. At the time of al-Hamdānī and his teachers in the preceding generation, this difference gained new political significance when the first Zaydi Imam and his companions arrived in the Yemen to stay and rule. In that context, al-Hamdānī’s patrilineal descent-chains of long, written South Arabian genealogies also had a competitive (anti-Zaydi) dimension to them. They were equipped with references to the glorious South Arabian past alluded to in the Qur’an: personal names and toponyms from pre-Islamic inscriptions (for example, al-Qalīs, the church in Sanaa) were inserted in the genealogies, giving South Arabian genealogies a distinctive feature compared to those of North Arabia. In sum, these South Arabian tribal genealogies provided a means of collective identification through tribal leaders’ knowledge, among one another as well as by contrast to outsiders. In that sense, they also informed and legitimized tribal populations’ social coherence and social status whenever main aspects of their contents were publicly referred to.

Among the rare women who appear in the genealogies are the brides in intertribal marriages. Referencing them apparently served the documentation and maintenance of cohesion and solidarity at the highest social levels of South Arabian tribal federations.⁹ The Ḥimyar were a case in point, discussed in the second volume of al-Hamdānī’s al-Iklīl (while Khawlān are addressed in the first and Hamdān in the tenth volume). The genealogy of Ḥimyar implied problems, since it had to incorporate groups that did not originally pertain to tribal federations while a core group probably were descendants of the last pre-Islamic South Arabian rulers and their families, as suggested by their name. The author and his interlocutors (re-)constructed the quasi-tribal entity of Ḥimyar along the lines known from (other) tribal federations, to give this Ḥimyar genealogical construct its logical base. In such a case, mentioning women in genealogies might come in handy to highlight connections that could be overlooked, or, in our view, to construct connections through bilateral elements of kinship that could not otherwise be made. A bilateral or even bilinear emphasis may at times emerge for the present purpose of consolidating a group, but this may also occur as the re-activation of a covert legacy from the past. This may suffice to introduce the somewhat detailed discussion of relevant text sections.

Just as for al-Hamdānī himself, elaborating the genealogies of the Ḥimyar already had been work in progress for Abū Naṣr al-Ḥanbaṣī, al-Hamdānī’s authority for the Ḥimyar (himself of a noble Ḥimyarī family), and for other experts referenced by the author. The procedure of combining credible specialists’ testimony with the author’s own analysis aimed at providing the required pristine varnish for the unique characteristics of South Arabian genealogies. Additional old sources were available for the Ḥimyar genealogies. In Sa‘da the sijill obviously were lists of an at least partly genealogical content¹⁰ in the possession of Ḥimyar residents among the Khawlān in Sa‘da. Interested visitors could possibly consult with these owners and have them look up specific genealogical relations. Having lived in Sa‘da for 20 years, al-Hamdānī was certainly well informed about the sijill and used it himself when writing volume 1 and 2 of his Iklīl. How he gave these lists special consideration as sources is illustrated in the following sections.

Al-Hamdānī begins his quote from the sijill with a careful protective announcement showing that he was not certain whether the genealogy presented in the sijill was still actual or superseded by newer versions mirrowing changed tribal political conditions: ‘The groups [buṭūn] of al-Ṣadif on the authority of the inhabitants of Ṣa‘da who are the owners of the sijill, read before a genealogist of al-Ṣadif.’ He goes on to emphasize his own role in this genealogical research:

I read before Muḥammad b. Zaghlab b. al-Ḥārith b. Muḥammad al-Ṣadifī of the offspring of Almā b. al-Ṣadif, one of the inhabitants of Dammūn of al-Hajarayn in Ḥaḍramawt, the genealogy [nisba] of the Ṣa‘dīs, and I corrected it on his authority.

The al-Ṣadif group discussed here by al-Hamdānī in fact may well have originally been a group without ‘tribal’ affiliations,¹¹ but at the time of writing its representative is the author’s expert interlocutor. In short, the sijill is corrected or updated by someone who had to know: Muḥammad b. Zaghlab himself was a Ṣadifī from Ḥaḍramawt, lending additional credibility to the authority of the sijill and to al-Hamdānī’s text at the same time.¹²

In the course of the long quotation (with additions by al-Hamdānī and Muḥammad b. Zaghlab) two passages occur where women appear to an extent that by far surpasses the usual procedure of the genealogists. One of them refers to Khuwār b. al-Ṣadif:

Among the offspring of Asad b. Ju‘sham are Umm ‘Abdallāh, and among her offspring is ‘Aydān of Ḥaḍramawt, of dhū l-‘Urf, and āl Abī ‘Alī of Banū Rabī‘a; further on Ni‘ma, who was the mother of Ḥujr b. Muḥammad b. ‘Umar b. Ḥujr, and ‘Abda, who was the mother of Shārūkh b. ‘Uqba b. Ḥayyāt b. Asad b. Ju‘sham the Younger, and the mother of Yaḥyā b. Muqsim al-Şadifī of Banū Baḥrī, and the mother of Mirdās b. Dāfi‘ of Banū Ju‘sham of al-Aḥrūm.¹³

According to this sijill (as read by Muḥammad b. Zaghlab or al-Hamdānī), a man named Asad had no male offspring, only daughters. The first daughter is referred to by her kunya (teknonym), while the second and third are quoted by name. We read nothing of the husbands of the first two; the first daughter has no individually named offspring, only groups as a whole (denoting a rare case of collective matrilineal descent), whereas the other two daughters have individual sons. Subsequently, no daughters are named, so we are back to a common patrilineal genealogy.¹⁴ The third daughter of Asad is shown to have married three times into other branches of the al-Ṣadif group: her marriages contributed to the cohesion of the group as a whole. In passing at least, a strange lapsus in this genealogy should be noted: ‘Abda obviously married her ibn ‘amm (FaBrSo) and had a son Shārūkh whose great grandfather is shown as a son of Asad, but initially Asad was introduced as having had only daughters.

The second passage is taken from the same genealogical context of al-Ṣadif, the sijill, presumably with additions by Muḥammad b. Zaghlab and al-Hamdānī:¹⁵

‘Abdallāh b. ‘Imād sired al-Ḥaḍramī, <al-Ḥaḍramī sired>¹⁶ fourteen men and three women who were al-Ṣa‘ba, Umm Farwa und Umm ‘Amr, the daughters of al-Ḥaḍramī. al-Ḥaḍramī b. ‘Imād married four women, among them ‘Ātika bt. Wahb b. ‘Abd al-Dār b. Quṣay; she bore him Abū Hidm b. al-Ḥaḍramī, Sha‘ba b. al-Ḥaḍramī and al-Ṣa‘ba bt. al-Ḥaḍramī. He also married Zuhra bt. Mālik b. ‘Amr b. Majda‘a b. Jaḥjabā, then from al-Aws [from the northern genealogical context], and she bore him ‘Āmir, al-‘Alī, Abū ‘Amr, Mālik, ‘Abd Mālik, al-Nu‘mān und ‘Amr, the sons of al-Ḥaḍramī. And he married al-Ghuḍūb bt. ‘Afīf b. ‘Awf b. ‘Abbād b. Yarbū‘ b. Naṣr b. Mu‘āwiya b. Bakr b. Hawāzin. ‘Afīf was the one who on the battle-day of Ṣayḥa¹⁷ cried out, and it threw al-Ḥibālī down because of his loud voice. She bore him al-Ḥārith, ‘Ubayda, ‘Abd al-Ḥārith and ‘Imād, the sons of al-Ḥaḍramī. And he married Umm Shurayḥ bt. Ja‘far b. Sa’d, who was a woman of Tujīb, from the offspring of al-Ashras. She bore him Shurayḥ, Umm Farwa und Umm ‘Amr, the children of al-Ḥaḍramī. All the children of al-Ḥaḍramī married among the ashrāf [noble men] of Quraysh, they [themselves] and their children, and the ashrāf of Quraysh married among them.

The example thus illustrates a fairly harmonious combination of isogamy with hypergamy, with unusually explicit references to women’s position in these settings. Again, one father is concerned, who himself had fourteen sons and three daughters. He married four women: the first belonged to the wider genealogical context of the Prophet, the second was from al-Aws, an important tribal group many of whom lived in Iraq. The third was a woman from the Hawāzin, another tribal group. The fourth was from Tujīb (from the genealogical context of Kinda). What is striking here is that all four women came from the Northern Arab genealogical sphere. This was certainly an advantage when all children married among the Prophet’s family network of Quraysh, including male marriages ‘upward’ later considered unlawful by some because of kafā’a, that is, the adequateness of marriage partners from their social/genealogical origin according to Qur’an interpretations. But at this time, the genealogy of Quraysh was probably still seen as a normal tribal genealogy, albeit in the upper ranges, similar if not even equivalent to the genealogy of their partners in intermarriage.

These two sijill-based passages explicitly mentioning women evidently serve different purposes. The first shows a father who has only daughters, but the group of his offspring has to continue. The second shows a noble family with marriages outside of their own group, even including intermarriage with the Quraysh. In that case their own lofty status position could be underlined by mentioning the women’s role in ‘big’ marriages.

One sign of the age of the sijill may indeed be this unusually high, explicit occurrence of women and marriages. Another indicator of the document’s age concerns the use of the Quranic term rahṭ, denoting a group ‘consisting of [the] nearer relations’ of a man, that is, his kindred,¹⁸ for which al-Hamdānī himself mostly uses nafar. The usage of rahṭ instead of nafar occurs only in this particular quote from the sijill. This confirms that al-Hamdānī is indeed quoting another author’s text with an older vocabulary and content.¹⁹ At any rate, al-Hamdānī’s quotes from the sijill, with the text’s strong bilateral and on one occasion bilineal emphasis on men as well as women, differ markedly from the generally more restricted patrilineal orientation in his overall representation for tribal genealogical connections.

Within these overall representations in tenth-century highland and Upper Yemen, written genealogical patrilineal reasoning thus fulfilled important social functions for many leading tribal groups. First, it did to an extent marginalize women, but was flexible enough to integrate them (with bilateral and bilineal inserts by means of old documents) for group interests. Second, it put tribal South Arabian leaders on an equal footing with the Prophet’s descendants, that is, the Ahl al-Bayt. A third dimension was added to these male group-centred and interregional elements, that is, that of basic equal status among and between the various tribes. Common reference to South Arabian ancestry, within a kinship idiom that does not place any privilege upon seniority among ancestral siblings, thereby provided a rationale for contrasting equivalence among and between the tribes themselves.

Contrasting equivalence defined basic intertribal relations through differing names and territories of basically equivalent status. It also was fundamentally relevant for interactions between members of different leading tribal groups, and to a lesser extent also for other individual persons of different tribal affiliations as a basic dimension of their social status. Yet beyond these basic status considerations, the details of tribal elites’ genealogical reasoning did not necessarily play any regular role for normal tribal majority household members. From the perspective of any individual agricultural household in the tribal domains of the time in this region, the existence of such detailed, common patrilineal genealogical awareness and knowledge cannot, and in fact should not, be assumed a priori. Members of majority households would have known about these records, but they could not access them without permission, and in fact they could not read them since they were usually illiterate. The few written genealogical records a tribal leadership group stored and guarded in their own realm would therefore be utilized in ways that were similar to how al-Hamdānī described others’ access to the sijill among the Ḥimyar. The average tribal member or interested traveller would have to ask the tribal chiefly family for the possibility that (sections of) these records would be read out to them (unless they had access to the relevant hand-written copy of al-Hamdānī’s book).

Anthropological reasoning (Goody 1988; Gellner 1988) and evidence from the ethnographic present suggest that average ninth/tenth-century adult household members usually did not master reading or writing. What would matter for them would be descent, combined with just one crucial genealogical element, that is, ‘plus’ the overall tribal name. These largely verbally transmitted ‘descent plus’ oral registers would focus on a local household affiliation – that is, at most three or four names that would connect the name of their house and local village group to that of the relevant tribal subsections and at the top of this very short chain, the genealogically informed name of the entire tribal territory and/or tribe. For the average tribal household member these short chains would primarily represent not a genealogical statement about the past, but a statement about territorial adherence and social belonging in the present. These indicators lead to proposing the following model.

Regular tribal household members used short verbal descent chains of status affiliation that could but did not have to be encoded in patrilineal terms. Tribal elite members by contrast used written records of detailed, usually patrilineal genealogies (but occasionally including bilateral or bilineal elements as reflected in al-Hamdānī’s sijill reference). These were two different genres – one, mostly in writing, was centred among the elite, imbued with claims of longevity; the other was mostly verbal, it was decentralized among tribal households and more open to shorter changes of one kind or another. The two genres were interrelated, and although their content differed they could intersect to an extent. One set of these resources could be accessed and used to inform and inspire the other and their respective representatives. Yet in everyday life the genealogical records of the tribal elites were at quite a social distance from local household members’ concerns. The coexistence of these partially interrelated genres did not imply any binding necessity that the patrilineal idiom in the elites’ genealogical records had its direct patrilineal correspondence in the majority households’ short chains of status-related names. Within the wide range of kinship terminologies among the tribal majorities, the short chains of personal status affiliation were essentially compatible with any variant in the regional inventories of kinship terminology. At the average tribal household level, ‘short chains’ of names therefore indicated territorial and sub-group adherence, while descent was recognized primarily through the respective personal proper name and a nisba.

The general absence of any elaborate patrilineal kinship genealogy among regular tribal households corresponded to another important element: Islamic rules of inheritance are essentially bilinear. Property is passed on through men, but also through women albeit in smaller shares. In theory, Zaydism allows even a larger circle of female heirs receiving inherited shares than most Sunni orientations would recognize (Madelung 1965). Yet some caution is appropriate with regard to the actual impact of these scriptural rules on the ground.

Beyond pious endowment (waqf) and testament (wasiya) (Hovden 2019), inheritance in the Muslim world always raises the question of women’s various status positions in general and of female property rights in particular. Despite the lack of any direct source evidence for the tenth century, a number of indicators suggest that adult women’s positions among the tribal majorities in Northern Yemen – with all the heterogeneity and variation this necessarily included – comprised a certain occurrence of bilinear inheritance and were thus relatively strong: that is, if compared in the region itself during the twentieth century, to women’s positions among the Ahl al-Bayt and tribal elites (for example, Dresch 1989) in the Yemen’s North, or to those among pastoral nomads to the Southeast (Dostal 1967) and Northeast (Cole 1975). The underlying factors differed but led to similar results. In addition to questions of theological standing, status and property, patrilineal principles of inheritance were also emphasized wherever succession was at stake – for leaders of family clusters, as scholars, or as candidates for higher office. Women had no access to these offices except on rare occasions – mostly as daughters, sisters or widows succeeding certain male Zaydi scholars and Sunni holy men (al-Sharjī 1986, 305, 343). Otherwise, the elements of succession, and of large property, substantially favoured patrilineal succession among the tribal, scriptural and theocratic elites.

By contrast, several elements suggest that bilinear inheritance did regularly occur among the tribal-agricultural majority. First, some of the leading experts testify for ancient South Arabian legacies that no available evidence indicated women might have been excluded from access to property (Beeston 1957; Maraqten 2008), mirrored in the pre-Islamic occurrence of cognatic kinship terminology reconstructed from epigraphic evidence. The simultaneous continuation of cognatic terminologies and bilinear inheritance would have represented a combination to which Islam would have no basic objection. Second, wherever Sudanese terminology and patrilineal descent gained influence between the seventh and the tenth centuries, this transformation did not have any necessary effects upon bilinear elements of descent and inheritance. Third, where Islam already had an impact on the ground after the second wave of Islamization in Upper Yemen had set in, this would include additional support for a pre-existing inheritance tradition from a new side. By contrast to early Islam’s impact in Northern and Northwestern Arabia for women’s roles, in Southwest Arabia it seems more likely that Islam emphasized such continuities in women’s inheritance rights that already had been quite strong by then.

To sum up this section, the tribal versions of a domestic mode of production in Upper Yemen between the seventh and the tenth centuries display a number of crucial kinship features. From a top-down perspective, strong elements of elaborate patrilinear genealogies among the elites provided factors of contrasting group status equivalence and coherence. From a bottom-up perspective, household members most probably identified with short chains of ‘descent plus’ names emphasizing status and territory with links to the respective tribal name. Inside many tribes, different but interacting genres of descent and genealogy simultaneously contrasted with the fixed, elaborate patrilineal genealogies of the Ahl al-Bayt. At the domestic level of agricultural household units, a range of terminologies and descent-reckoning practices between bilateral cognatic and Sudanese patrilineal forms prevailed, associated with the common though not exclusive practice of bilinear inheritance. This allows us to move to the last of three transitional periods for assessing marriage patterns in the region.

Changing marriage patterns in Upper Yemen by the thirteenth century

This section will combine the findings gained so far with some insights on marriage as interactive and transactional processes among households and as investments in domestic production ‘pools’, but also as social variations of marriage alliances before and in the course of the thirteenth century.

Marriage transactions in Eurasia have been aptly characterized by Jack Goody as ‘the first’ in ‘a series of mechanisms and strategies of continuity and heirship which […] help to organise the domestic economy and domestic life more generally’ (Goody 1990: xviii, see also Segalen 2021). Marriage transactions are better understood as resources pooled into a household as a direct or indirect dowry, rather than being misunderstood – through an undifferentiated usage of the term ‘bridewealth’ – as payments for bridal services. The stratified agricultural systems of pre-industrial Eurasia allowed for highly differentiated forms of marriage transactions through which women became endowed with property rights. Martha Mundy (1979, 1995) exemplified this for certain Yemeni ethnographic contexts. Those transactions, which were in line with Islamic law and were contributed by the groom and his family, did not necessarily stay with the bride’s father, but were at least partially passed on to her (Goody 1990, 364). The argument provides an adequate starting point for this section.

Local and regional endogamy between tribal households was an unavoidable tendency if tribal boundaries were crucial for keeping outsiders out. Marriage transactions, and inheritance by children from a marriage, preferably had to be kept among residents from the same villages or tribal sections. First marriages usually were celebrated at some limited time period after puberty, often being arranged by the couple’s parents (see also Parkin 2021 ). The majority of unspectacular, close, rarely documented, or ‘little’ marriages (Bourdieu 1977) thus took place within such nearby local contexts. In rural Upper Yemen after the tenth century such local marriage transactions prevailed that corresponded to the Qur’an’s request for a maḥr (‘bridewealth’) to be transferred upon marriage. The Qur’an probably had not introduced the maḥr to Southwest Arabia as a totally new phenomenon. Rather, it standardized and legitimized some existing forms, including the main transaction from the groom’s family to the bride’s. By the term maḥr it became effective as indirect dowry, and as essential for completing a marriage. From a ranking by al-Hamdānī of three main ‘forms of property’ one can infer that among the common tribal people, marriage transactions mostly included land and livestock, while less often also entailing money. The actual transfer to the young married woman of parts of the maḥr, and of her inheritance shares, often could be delayed to a later point, which indeed represented a form of ‘banking’ (Goody 1990, 373f.) for periods of greater need, such as in case of divorce.

Marriage among equals, that is isogamy, was favoured by a structural, twofold emphasis on status equality. One was the avoidance of any hierarchies among brothers or among sisters in Sudanese and cognatic terminologies and in bilinear inheritance norms. A second factor to the same effect was the basic equivalence of status between members of tribal groups. Isogamy therefore was an almost self-evident, non-exclusive but frequent practice within preferential local endogamy for ‘little’ marriages. Very close marriage practices, such as those between the children of two brothers (that is, the ‘Arab’ or bint al-‘amm marriage famous in Northern Arabia and North Africa), were never particularly common in Southwest Arabia – which in fact confirms the very limited regional relevance of patrilineal descent groups among common tribal majorities. But other versions of repeated, close forms of isogamous intermarriage are mentioned by al-Hamdānī, for example, when in the first and tenth volumes of his Iklil he refers to the two tribal sub-groups of the B. Harb and the Murr (rather than merely to their respective elites) as being ‘in-laws to each other’ (Heiss 1998, 132f., 149, 184: n. 17). The example is not a case of badal, that is, the direct exchange between two men of their sisters for avoiding the maḥr (disapproved of in the Zaydi realm like elsewhere in the Muslim world), but an instance of regular, reciprocal intermarriage patterns between members of two tribal groups. Since a bride-giving family was seen as transforming itself into an inferior social position vis-à-vis the groom’s family, the Banu Harb and Murr intermarriage example demonstrates how this could be balanced out.

Isogamy would also find additional encouragement in those Qur’an legacies arguing that the kafā’a norm expects a bride not to be married into a social status below her household of origin (Gingrich 1995). This marriage norm strongly endorses isogamy, while allowing bridal hypergamy yet placing a taboo upon hypogamy. It is doubtful, however, whether this particular norm met any uncontested acceptance in tenth- to thirteenth-century Yemen. Our discussion of al-Hamdānī’s second sijill example already indicated how the observance of kafā’a required some consensus between the families involved about each other’s social status. It seems that similar to other Qur’an based marriage norms (for example, concerning polygyny), kafā’a was merely loosely observed or ignored even by some among the elites before the mid-thirteenth century (Heiss 1998, 130). It was a matter of historically situated, contested interpretation. This may also be examined through an ethnographic lens on certain family traditions of the late twentieth century among the old chiefly families of Upper Yemen, whose histories in some cases go back to the twelfth and thirteenth centuries. In their records, intermarriage with families from their own tribes was the rule rather than the exception. That also included the occasional marriage of the bride from a tribal chief’s family into a regular tribal family (Gingrich 1989; Brandt 2014). In some of these contemporary cases as well as in some of the twelfth and the thirteenth centuries, this seeming violation of the kafā’a norm may precisely emphasize that the chief’s family is ‘equal to’ rather than ‘superior to’ the others. In principle, these practices would relate to differing layers of intra-tribal hierarchies, for which variations across regions and history were the rule rather than the exception.

Isogamy therefore could never be any strict standard rule among the tribal communities. Instead, it had to be a contested and complex sequence of debates and processes reviewing the potential partners’ and their families’ tribal social status, their legal positions, and their standing in terms of honour and reputation. Between families to be connected through marriage, any equality of material wealth was almost impossible in view of many inherent forms of economic inequalities between mountain farmers’ households. Next to inheritance, however, marriage was a key mechanism of peaceful property transfers.

The distinction between isogamous and hierarchical marriages was a blurred one, as we saw, particularly before Zaydism achieved a somewhat more stable hegemony by the mid-thirteenth century. Yet there were nevertheless relatively clear cases of hypergamy before that time too. In one instance, al-Hamdānī mentioned a tribal elite group whose status was so high that they found it difficult to find any marriage partners at all in view of the high transactions involved (Heiss 1998, 46). The point indicates not only that for the elites – in contrast to priorities for local tribal endogamy among the majorities – marrying ‘out’ was essential for demographic as much as for political reasons. In addition, the example emphasizes that expenses for ‘big’ marriage transactions among the elites were much higher than among regular agricultural households, and that they involved direct dowry. Other cases reported by al-Hamdānī also indicated the more widespread occurrence of polygyny among the elites, that is, in contrast to normal tribal households. In addition to formal polygyny by male elite members with free but subordinate women, the possession of female slaves as concubines is reported for leading Zaydi male circles at least since the twelfth century by Ibn Ḥātim (Moorthy-Kloss 2019, 70). Out of wedlock affiliations by enslaved concubines to patriarchs in Zaydi and tribal elite families, as well as (formally free) offspring from these relations, thus had become an undeniable and systemic element of elite polygyny and of gendered stratification by the late twelfth century.

Preceding sections demonstrated the much more explicit observance of patrilineal principles among these tribal and non-tribal elites of Upper Yemen, in their genealogical reasoning and in succession. A basic recognition of scriptural Islamic interpretations of marriage norms also was somewhat more widespread among them. Most of their marriages were either isogamous or hypergamous. Their fewer households were wealthier, larger and included polygyny and concubinage. These would become the location of more frequent, spectacular, documented, political ‘big’ marriages. By the interplay of all factors involved, women lived under more strictly observed conditions of seclusion and subordination inside these households already before the thirteenth century.

By the early thirteenth century, elite marriage patterns began to enter a new, influential transformation. The Zaydi Ahl al-Bayt had not managed to establish their rule in any stable way in Upper Yemen before that time. After their initial installation in Sa’da (early tenth century), many successors to the first Zaydi Imam encountered increasing dissent from various sides, among them the Muṭarrifiya movement. The Ahl al-Bayt leadership overcame these phases under Imam al-Manşūr ‘Abdallah b. Ḥamza, who died in 1217 (Heiss and Hovden 2016). After his reign a stable denominational and social Ahl al-Bayt hegemony in Upper Yemen no longer could be ignored. The establishment of special hijra enclaves all across Upper Yemen and in several parts of the Western mountains was a crucial part of this transition. This indicated a much more profound impact of Islamic principles in their version of Zaydism led by Imams from Ahl al-Bayt for subsequent periods, as compared to previous centuries (Gochenour 1984; Madelung 2012).

By the thirteenth century, social relations among the leading circles of the region were going through a significant shift in terminology, indicating actual processes of transformations. In al-Hamdānī’s work, the author’s terminology avoided any usage of the term ‘Ahl al-Bayt’, which reflected his general scepticism towards increasing Zaydi influence. The biographer of the first Imam, by contrast, an open supporter of the Zaydi Ahl al-Bayt and a partial contemporary of al-Hamdānī, used more friendly and supportive terms for the group (including Ahl al-Bayt). Yet in spite of their opposing views, both authors agreed in one respect: in the tenth century, nobody in Upper Yemen and the Western mountains yet referred to the Ahl al-Bayt by the term under which they became known much later, that is, Sāda (masters, lords). In contrast, in his pro-tribal writing al-Hamdānī (in al-Iklīl) routinely addressed the tribal elites of Upper Yemen as Sāda (Heiss 1998, 29; Heiss 2005), without ever using the term shaykh, common from more recent literature and from ethnographic evidence. Sāda, as a public designation for members of the Ahl al-Bayt by others, came into public usage only after their enduring stabilization was accomplished in the early thirteenth century (Heiss 2005). In turn, that new elite hegemony interacted with existing kinship relations. Once it was accepted who the superior ‘Lords’ of Upper Yemen were, regular hypergamous relations to them could potentially honour the bride-givers as well. The growing hegemonic effects upon kinship relations in the region eventually included a relative increase in hypergamous marriages from some parts of the tribal population to the tribal elites, and from the tribal elites to the Ahl al-Bayt. By consequence, this must have led to the gradual emergence of a cluster of tribal leaders more closely associated with the Ahl al-Bayt through marriage relations, while others were ignored or chose to ignore the Ahl al-Bayt in this regard.

These spiritual and social ties conveniently were enacted through the growth of hijra enclaves. More or less enduring realms of core connections in the Zaydi networks of tribal influence emerged, as opposed to the more volatile and the more peripheral spheres. By contrast, other tribal realms had more or less stable affiliations to other ‘galactic’ centres of power (Tambiah 1977; Gingrich 2014) in the region (for example, Banu Mālik of Khawlān with Sunni lowland centres, or the Muslim tribal sections of Banu’l-Hārith with Najran). Beginning with the mid-thirteenth century, the Northern Yemeni highlands and their adjacent regions thus may be envisioned as a galactic space, with a few mobile centres of gravity and several smaller planets and satellites with their more irregular movements around them. Zaydism representing the major constellation within this mobile space, hijra settlements provided nodes for social ties and networks among Zaydism’s non-tribal and tribal elites. Many tribal elites, however, continued to shift their loyalties in between these core networks from one centre of gravity to the other in pursuit of their relative autonomy (Figure 1).

Figure 1. Galactic polities and elite intermarriage patterns in Northern Yemen after 1250 CE.

The emergence of these Zaydi core networks in turn elevated some of the long-standing ruling tribal elites to allied tribal dynasties, with analogous marriage patterns among their own tribal majorities favouring bridal hypergamy. From a long-term perspective following 1250, the ensuing results among the tribal elites under increasing Zaydi influence in Upper Yemen and Asir would be a gradual growth of social stratification, a progressive implementation of patrilinear principles in terms of dynastic elite formations, and a corresponding intensification of women’s seclusion and subordination.

In these new marriage patterns, only the tribal elite family gave a daughter to the groom’s Ahl al-Bayt family, while the opposite never took place. The establishment of this type of ‘big’ hypergamy presupposed an increasingly hierarchical shift with regard to gender relations among the tribal leaders and the Ahl al-Bayt, through politically pre-arranged alliances for which brides were expected to move ‘out (of their village) and up’ into a different social stratum. That shift interrelated with the consensual practice of those in power among the dominant tribal and Ahl al-Bayt families, legitimized by the kafā’a principle.

Under these conditions, the tribal bride would always move into the Ahl al-Bayt family and virilocally reside there as her new husband’s first or additional wife. The marriage contract established, confirmed or re-activated existing ties of loyalty and alliance between the families involved, thereby including elements of reciprocity. The Ahl al-Bayt family would provide good connections and scholarly, legal, administrative or political services to the tribal elite family, who would usually commit to ‘protect’ the Ahl al-Bayt family within and beyond the hijra in the tribal realms (Puin 1984).

Such a marriage contract would always imply an ultimate symbolic statement, built on the principle that bride-givers acknowledged their irreversible social inferiority to bride-takers, that is, that this relationship could not be reciprocated by means of marriage in the foreseeable future. This implied that tribal leaders’ families publicly posed as the socially inferior bridal elite group, and Ahl al-Bayt families would gladly celebrate their recognition as the socially superior groom elite group. If marriages of this ‘big’ type occurred before 1250, they were individual cases of limited symbolic significance. As a solidified element within the wider cluster of elite marriages, these forms of hypergamous marriages could only become a more widespread phenomenon (continued into the ethnographic presence) after Zaydi Ahl al-Bayt hegemony had become more firmly established in Upper Yemen and the Western mountains.

Conclusion

This study’s results point out that in the tenth century the Zaydi Ahl al-Bayt in their entirety had not yet reached a social and spiritual standing that would have allowed them to be addressed in public (even by their own supporters and followers) as the lords or Sāda of anybody in Northern Yemen. To reach such a wide, publicly acknowledged legitimacy would require additional efforts. They achieved some stable foundation when those who had themselves formerly been addressed as ‘lords’ of the tribes were now to publicly acknowledge their own reciprocal inferiority to the new ‘lords’. In addition to religious and political rituals, the structural establishment of hypergamous marriage relations, between the more or less loyal clusters of the tribal elites and the one or other faction of the Ahl al-Bayt, was an excellent building block in elaborating this publicly acknowledged hierarchy: through hypergamous marriage patterns, the tribal elite families and previous ‘lords’ of al-Hamdānī’s time would publicly recognize the superiority of the new Sāda of Upper Yemen.

On the other hand, the structural diversity of tribal lives among the regular rural household population would be going through less dramatic transformations, but could continue to evolve more gradually along diverse trajectories. In the affluent tribal highlands of Upper Yemen and its Western mountains, the existing diversity of tribal social configurations was eventually transformed into zones of core Zaydi influence, as opposed to those with intermittent and those with little or no Zaydi influence. Structural heterogeneity of kinship relations in the tribal domains of Upper Yemen would continue, with varying fields of women’s agency including more tribal female power in the vast peripheries. This was facilitated by elements of cognatic kinship terminology, short chains of descent reckoning with bilateral elements in them, and bilinear inheritance options. Eventually, however, a more dynastic and explicitly patrilineal genealogical hegemony would manage to establish itself inside this tribal diversity – as volatile and fragile as it would continue to be – in the form of increasing female seclusion as part of emerging hierarchical and patrilineal elite relations to new ‘galactic’ centres of normative legal gravity and ideological power.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 ‘Yemen’ refers to pre-Ottoman Southwest Arabia. These regions were somewhat larger than late and post-colonial territories and states known by the same name. This text is an outcome of the Austrian Science Fund’s SFB programme (VISCOM – SFB F42-G18), a cooperation project between the Austrian Academy of Sciences and the University of Vienna, in which the authors participated between 2011 and 2019. These three institutions’ support is hereby duly acknowledged. The authors are grateful for comments on earlier drafts of this text by Eirik Hovden (Bergen, Norway), Christina Lutter, Magdalena Moorthy-Kloss, David Westacott (Vienna), by this journal’s editor and peer reviewers, and by their other colleagues at the Austrian Academy of Sciences’ Institute for Social Anthropology’s Yemen Studies group (Marieke Brandt, Lisa Lenz-Ayoub and Alexander Weissenburger). The assistance by Eva Kössner (Vienna) for the elaboration of Figure 1 is thankfully appreciated.

2 Except for Anglicized forms (e.g. Imam, Sa’da, Zaydi, Qur’an) this text uses IJMES transliteration rules for Arabic terms.

3 On a few occasions, this text is supported by fieldwork insights during the ‘ethnographic present’ by Gingrich and/or Heiss in Northern Yemen and Southwest Saudi Arabia between 1980 and 2002.

4 The term Himyarite designates not only one among the Ancient South Arabian polities, which came to dominate the final phase of that era. It also refers to the non-Arabic, Semitic language (and related vocabulary as well as scripts) that survived those polities for centuries. In the tenth-century CE, Ḥimyar also designated a quasi-tribal entity of particularly high status, which was attributed to their claim for partial descent from the Ancient Himyarite elite.

5 In this initial discussion, Islamization is addressed as the gradual introduction and standardization of relevant values and terminologies. Among ensuing social practices, inheritance and marriage will be addressed in subsequent sections of this text. As for taxation, Sima and Korotayev (1994) argued that with regard to the ‘Tenth’ (‘ushr) Islamization may have simply continued and codified certain pre-Islamic Southwest Arabian practices.

6 Judaism was prevalent among major tribal associations of Ḥimyar in central Yemen, Kinda in Hadhramaut (Lecker 1995, 635–636), among nomadic groups in other parts of Arabia (Newby 1988, 12), and since centuries before Islam in Najrān, according to some sources since Sabaean times (Lecker 1995, 636). Most other sources relate the beginnings of Yemenite Judaism to the conversion of Dhū Nuwās, the last tubbaʽ (‘king’) of Ḥimyar who ruled from the end of the fifth century for almost four decades. His missions against Ethiopian and Yemeni Christians, and the ensuing Ethiopian Christian reinvasion of Yemen (which in turn provoked a Persian invasion) contributed to the demise of the Ḥimyar polity in 525 (Nebes 2008; Tobi 1999, 34).

7 It seems that for a number of reasons the Muslim populations grew more rapidly over time than Christians and Jews as a general demographic tendency. To an extent, this was due to more children from polygamy (and concubines), and from unilateral patterns in inter-denominational marriages. Another cause related to the early Islamization of most land-owning elites, thereby also gaining socio-political influence. This may well have heightened tacit or explicit pressures toward conversion.

8 On non-tribal minorities and low status groups in later periods see Henninger (1989).

9 One example is the marriage between Bata‘ of Hamdān and Jamīla of Ḥimyar, see al-Hamdānī ([1368] 1949, 13). These marriage constructions gave South-Arabian genealogies horizontal cohesion, whereas the sequence of generations ensured vertical, diachronic coherence.

10 The word sijill has an interesting etymological history. As a diminutive of the Latin word signum, sigillum originally designated a small picture, and in a more restricted meaning a seal. The term sigillum travelled across several regions and eras, changing its meanings on the way as usual in such cases. It was adopted, for example, into several Germanic languages (English seal, German Siegel, Dutch zegel and from there, Indonesian segel) and into other idioms as well. The word also travelled to the Eastern Roman Empire. In the Byzantine realm it was adopted into administrative terminology, denoting a treaty or a pact. Its meaning thereby was transferred from the original seal to the whole document under which the seal was fastened. From Byzantine Greek σιγίλλιον the word then became absorbed into Arabic, where it referred to (drafts of) treaties or protocols by Qāḍī’s (Heiss 1998, 48–57). In his South Arabian tenth-century usage of the term, al-Hamdānī seems to apply it without any additional explanation in a self-evident manner. This could indicate that the term was widely known among the reading elites of his times and regions, and that it retained some of its official and quasi-legal meaning.

11 al-Hamdānī (1980, 36–54) comprises the total genealogy of al-Ṣadif.

12 al-Hamdānī (1980, 51–53). On this occasion, al-Hamdānī quotes the version of Almā b. al-Ṣadif’s genealogy taken from the sijill of the inhabitants of Ṣa‘da. He carefully adds the variants contributed by Muḥammad b. Zaghlab.

13 Al-Aḥrūm is a plural denoting offspring of Ḥuraym b. al-Ṣadif, a brother of Ḫuwār. The genealogy of Ḥuraym b. al-Ṣadif is given by al-Hamdānī (1980, 39–42). The quotation is from al-Hamdānī (1980, 37).

14 These shifts between patri- to matrilinearity and back indicate a form of ambilinearity, a sub-version of bilinearity.

15 al-Hamdānī (1980, 44).

16 We inserted these words; the next line ‘al-Ḥaḍramī b. ‘Imād married’ shows that they are missing.

17 Al-Ṣayḥa means ‘the cry’; as mentioned by al-Hamdānī the term gave the battle its name. Otherwise nothing is known about it.

18 See as an example Murra b. Udad, the rahṭ of al-Af‘ā al-Kāhin, whose name translates as ‘the viper, the diviner’, al-Hamdānī ([1368] 1949, 2) and cf. Lane (1863–1893, 2421c).

19 Cf. Lane (1863–1893, 2825a), where nafar can be the same as rahṭ.