ALLIANCES AMONG THE ARABS*

Arabian society of pre-Islamic and early Islamic times consisted of units whose members shared certain legal responsibilities. They may be called co-liable groups. Originally the boundaries of any given co-liable group matched those of a certain descent group (that is, a group consisting of agnatically related males). However, the boundaries of all the groups were constantly shifting, due to segmentation on the one hand, and alliances on the other. Three types of alliances may be discerned. The paper discusses one of them in detail, using the term "hosting alliance" to designate it. This type of alliance amounted to integrating outsiders into co-liable groups.

from the same distant ancestor.Thus the more distant the ancestor, the larger the descent group, and the greater the number of segments included in it.All Arabs considered themselves as being ultimately descended from two distant ancestors, in two different male lines, so that the genealogical scheme may be represented approximately as two pyramids.Descent groups are typically called "Banü so-and-so", i.e. "the descendants of so-and-so".It should, however, be noted that not every name mentioned in the genealogies stands for a founder of a descent group, and that the recorded genealogies are not always genuine (some would say: are always not genuine).As will be shown below, groups were sometimes formed by alliances, not by segmentation; but these groups too were eventually integrated into the genealogical scheme of agnatic descent groups.Furthermore, the sources convey the impression that pure agnatic descent groups only existed in theory; in practice, descent groups constantly absorbed outsiders who retained their own genealogy, at least temporarily.However, the basis and the bulk of almost any given social unit was a (genuine or alleged) agnatic descent group.Therefore, when I use this term I mean units that are considered by the sources to be descent groups, regardless of their actual origin and their attached members.
The sources preserved the names of many descent groups, which apparently varied in size and in their level of segmentation.It is often clear that a given descent group was an entity that comprised a number of independent parts.We may call such a descent group a comprehensive descent group.In the genealogies, the ancestor of the comprehensive descent group would be far removed up in the male line, whereas the constituent parts would be called after various descendants of that ancestor.Students of Arabia and Islam commonly refer to the comprehensive descent groups as "tribes".Such are Quraysh, Tamïm, 'Amir, Tayyi', Asad, Bakr, 'Abs, Dhubyan, to name but a few.I shall follow this common usage, and apply the term "sections" to the descent groups constituting these tribes, regardless of their level of segmentation.Due to the hierarchical structure described above, the distinction between tribes and sections is often blurred, so that my terminology is in fact arbitrary, but I shall use it for want of a better one.^ "Northern" tribes).When there are several possibilities in various versions of the same account, the smallest framework is probably the one that was involned in the events described.This means that even when a tribe (or a large section) is mentioned as having taken part in a certain activity, it is usually only members of one (or a few) of its constituents who were involved.^ We are thus left with no real definition of an Arabian tribe in the period discussed.

Shared responsibility among agnates
In pre-Islamic society the male adult members of certain descent groups shared legal responsibility.They were accountable for each other's offenses.At its most extreme manifestation this rale meant that they jointly sought revenge, or received blood-money, when one of them was killed by an outsider; conversely, they were all exposed to vengeance, or obligated to pay blood-money, when one of them killed an outsider.The obligation of mutual assistance applied not only in matters of blood revenge but in less extreme situations as well.^ The framework within which the rights and obligations of members were effective may be called a co-liable group, to borrow Emanuel Marx' terminology.Since outsiders constantly joined descent groups and shared liability with them, it follows that co-liable groups were not in fact identical with agnatic descent groups.^ However, joining a co-liable group involved a specific legal mechanism (to be discussed below), therefore we may assume that the original, or at least theoretical, boundaries of any given co-liable group matched those of a certain descent group.The mies by which co-liable groups were formed are unknown.The material at hand does not disclose ^ Landau-Tasseron, "Asad", esp.1-5; idem, "Tayy".See also Obermann, "Early Islam", 249 on the use of names, and 244-50 for a general description of the tribal system.
(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.eswhether they came into being on the basis of a certain genealogical depth, or mutual consent of the members, or a decision by the elders, or external public opinion, or any combination of these or other factors.^ Also unknown are the actual boundaries of liability, that is, the lines dividing between co-liable groups.We know thousands of names of tribes and sections, but we do not know how each of them actually operated in relation to the others.Not only is the source-material inadequate to supply such data in detail, but also the boundaries between the groups were not in fact rigid.As shown by the Tamîmï example above, members of different sections in a tribe could feel obligated to assist one another in certain situations.However, the men of a tribe never constituted a single co-liable group, and we cannot tell which sections within each tribe fiilfiUed this function at any given point in time.Where modem Bedouin are concerned, contemporary anthropologists can ask their infoiiiiants for data and attempt to draw the lines demarcating one co-liable group from another.^^ Modem anthropologists may indeed encounter certain difficulties, but they can at least both observe the groups they study and interrogate their members.As for the past, such methods are altogether inapplicable.The tribal system was (and is) dynamic, the processes of segmentation and alliance effecting constant changes in the composition of descent groups.Thus co-liable groups, sections and tribes were constantly dissolved and newly formed, but the sources only give indirect and sporadic data about such matters.Moreover, they also confiise the names of smaller and larger sections as explained above.It is thus not possible to define the actual boundaries of past co-liable groups.It is only possible to point out that they were defined by legal resposibility rather than by genealogy, and that they absorbed outsiders by a certain t3^e of alliance.^ Cf.Stewart, Texts, 1/88, where a Bedouin says, "blood-money pacts have been made between people who've risen five generations from each other, and people who've risen six generations..." This comment amounts to a statement that co-liable groups ("blood-money" groups) are formed by both a certain genealogical depth and mutual agreement ("pacts").
'^ See e.g.Marx, Bedouin, chapters 7, 8. Marx' terminology is diiferent from mine.He defines the boundaries of certain co-liable groups and indentifies their constituent parts, which he terms "sections" (see e.g.115, 125); in my terminology a section would be larger than a co-liable group.

Types of alliances
Genealogy was not the only adhesive force in Arabian society.Alliance, hilf fulfilled this fiinction as well.The term hilf covers various types of agreements.The root hlf signifies oath, and any agreement bolstered by an oath can be referred to as hilf.Indeed alliances are called /^//f precisely because they involve an oath.The close connection between the two, and the precedence of the meaning "oath" to that of "agreement", is indicated by the report about the Fudül alliance (hilf al-fudül).In pre-Islamic times, certain Qurashï sections agreed to protect the interests of strangers coming to trade in Mecca.It is reported that "they exchanged oaths in the name of God, saying, "we shall help those who are wronged'... they concluded an alliance agreeing not to allow injustice in Mecca...".' ^ The Fudül was not merely a sworn agreement to help the oppressed, but an alliance, as proved by a later incident.During the caliphate of Mu'áwiya, al-Hasan b. 'Alï died, and his brother al-Husayn intended to bury him next to the Prophet.The Umayyad governor, Marwan, strongly opposed this project, and so al-Husayn rallied the Fudül alliance.The sections Hashim, Taym, Zuhra, and Asad assembled at al-Husayn's call, equipped with their weapons, to back him up.^2 It is in fact difficult to distinguish between hilf^s a sworn agreement on co-operation, and hilf as an alliance.The verb tahálafü, meaning both "they exchanged oaths" and "they concluded an alliance", sometimes covers treaties and agreements upon limited co-operation, because they were fortified by oaths; the taking of oaths, however, is not always explicitly mentioned.Among other things, the treaty of Hudaybiyya between the Prophet and Quraysh is sometimes referred to as hilf as are the pre-Islamic treaties of the îlâf the boycott of Quraysh against Banü Háshim, Muhammad's treaties with pagan tribes, a treaty between Bedouins and Kurds at the end of the 9th cen-'' wa-tahálafü bi-allàh al-qá'il la-nakûnanna yadan li'1-mazlüm...fa-summiya hilf al-fudül...tahálafü wa-ta'âqadû an là yuqïma bi- Akhbàr Makka 5/195;Ibn Kathïr,.Al-Jâhiz, Hayawân 4/471, states: "wa-asl al-hilf wa-l-tahâluf innamâ hmva min al-hilf wa-l-aymân ".Also the explanations in the classical dictionaries strongly imply that hlfzs, an oath preceded hlf as an agreement, contrary to Tyan's view, see Tyan,"Hilf" 388.'2 Al-Dhahabï, Siyar 3/276.
(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.estury CE, etc. ^^ It appears, then, that the term hilf is indeed rather confixsing.^^ A distinction is often made between tribal alliances on the one hand, and pacts between one individual and another, or an individual and a descent group, on the other.^^ The notion that individuals habitually forged alliances is based on two facts.First, the formal act of concluding alliances was often performed by individuals; the common oath formula was uttered in the singular form even though, by necessity, it applied to co-liable groups.^^ Secondly, the sources constantly refer to individual allies, and the status of being an ally is mentioned in both biographies and isnáds as an identifying feature of individuals.^' ^ The profusion of references to individual allies is misleading.More often than not, it turns out that an allied individual was in fact part of an allied section.For example, the ancestor of the Qariz '^ For Hudaybiyya and the ilâfsQC al-Fâkihï, Akhbâr Makka 5/208-9; Ibn Abî Shayba,MusannafimOAOl;318 (Hudaybiyya) ;Ibn Sa'd,Tabaqàt 1/75,78,71 (îlàj).The boycott of Quraysh against the Banû Háshim is designated tahaluf, see Yáqüt, Mu'jam al-buldân 1/170, 3/347; Ibn Kathïr, Al-bidàya wa-l-nihâya 3/84.Muhammad's treaties with pagan tribal groups are called hilf, see Ibn Hanbal, Fada'il 2/640.For the treaty between Bedouins and Kurds to fight al-Mu'tadid, see al-Taban, Ta 'rîkh (Beirut) 5/609.Other cases: the military alliance of local Syrian Bedouins against the Fatimids in the 5th/11th century, see Ibn al-'Adïm, Bughyat al-talab 5/2239, 9/4078; the agreement among the people of Khurasan to fight Abu Muslim, see Ibn al-Jawzî, al-Muntazam 7/271.It is sometimes clear that the agreement was bolstered by an oath, e.g.377,7/97,8/373,399,9/61,64,73. ^^ Reinert's statement (Das Recht,p. 22), that hilf'is a purely political agreement between two equal groups, ignores the diversity of the term and is therefore misleading.So is ; he sees hilf2LS a political, mostly temporary treaty between tribes,j/wâr as a protection agreement between individuals, and wala' as a pact between an individual and a conmiunity.See also Smith,Kinship,[53][54][55][56] •^ Crone, Roman, Provincial and Islamic Law, 51; Kister, "On strangers and allies", 142; Tyan, "Hilf" 338-9.Tyan's distinctions between types of alliances is blurred and self contradictory.See also Goldziher,Muslim Studies 1/65;Juda,Mawülí,p. 11, mentions alliances of individuals, but as a whole, his discussion (4-18) lumps together various types of agreements with no proper distinction between them.See also the previous note.
'^ See e.g.Ibn Habïb, Munammaq, 240, and below.On ceremonies related to alhances, including written documents, see Smith,Kinship,[53][54][55][56][57][58][59][60][61][62][63]Juda,Mawàlï,[2][3][4]  'Abd al-Hàrith from Zuhra, a section of Quraysh.But the alliance involved the whole of the Qàriz section (àl Qâriz) on the one hand, and the Zuhra section on the other.The Qàriz section was part of yet another alliance called the Ahàbïsh; the latter was in fact a confederacy of discrete sections which was allied to Quraysh in pre-Islamic times, forming an auxiliary military force for Mecca.^^ A simpler, perhaps more typical case is that of 'Utba b.Ghazwàn, a Companion oiF the Prophet from the Banù Màzin b.Mansùr tribe.'Utba is defined as an ally of the Qurashï section Banù Nawfal, and it would seem that his case is one of an alliance between individuals.However, according to the data supplied by the scholar Ibn Habib (d. 245/859), 'Utba's status was shared by his descent group, Banù Nusayb.^^ Nusayb was 'Utba's ancestor 4 generations removed, so that this alliance with Banù Nawfal involved much more than one individual.The same applies to many allies of Quraysh known to us, such as the famous Companions 'Abdallah b.Mas'ùd, al-Akhnas b.Sharïq, 'Abdallah b.Jahsh, al-'Alá' b. al-Hadramï.Each of these, and other allies of Quraysh, when checked, turns up with brothers or other agnates who shared the same status.Sometimes people who entered an alliance brought their previous allies into the new relationship, ^o Indeed the reports of the allies of Quraysh as recorded by Ibn Habib indicate that alliances always involved groups, not individuals.21 It may have happened that a man left his own descent group and allied himself with another.Such an individual might in time have offspring, who would be included in the alliance as well.Thus Ibn Sa'd states that the grandfather of the Companion Khàlid b. al-Bukayr concluded an alliance with the grandfather of 'Umar b.
•^ The alliance of Qàriz: Ibn Habib,Munammaq,239;Ibn Sa'd,Tabaqât 5/58 When one Ibn Nassar negotiated his affiliation with a certain group, the options were to "join as a brother or as one among several sub-groups" (i.e.sections).^5 in view of the above, it seems to me that a distinction between group-alliances and individual alliances is not significant.A distinction between types of covenants should rather be qualitative, not quantitative, and I therefore suggest the following classification.First, there were alliances which amounted to co-operation between sections for limited, clearly defined purposes, such as specific military enterprises, or consolidation vis-à-vis political rivals.The parties to such alliances never merged with one another, nor did any of them assume the name or use the nisba of the other.Thus we are told that the Hanifa and the Taghlib tribes allied themselves with one another against the Bakr tribe; ^^ Banù Malik, a section of Thaqîf, allied itself with the Tamîmï section Yarbù' against a confederacy from their own tribe, the Ahláf of Thaqîf.^^ Certain sections of Quraysh allied themselves to one another against other sections of the same tribe, who in turn formed a "counter alliance".The reasons for this particular division were economic and political; the groups were 22 Ibn Sa'd, Tabaqàt 3/388; Ibn 'Abd al-Barr, Istí'áb 2/426.2^ Naqà'id 2/653.See also Juda,Mawàlî,5,6,11 (but the individual allies whom he mentiones were in fact part of groups, as I have just explained).
2"* Al-Yaman was not the father but an ancestor several generations removed; there are, however, other accounts about Hudha3dfa, see Ibn Qutayba,Ma'àrif,114. 2^ See also the argument that follows in that discussion, "here are sub-groups with only one or two members...", Stewart, Texts 1/88-89.Further explanation on individuals will be offered below, the chapter on "the parties to alliances".
2^ Al-Isfahânî, Aghànî 11/48.The alliance did not apply to the whole of the tribes but only to sections thereof, cf above.called the Mutayyabün ("the Perfiimed") and the Ahláf ("the Allies").We should, however, beware of confiising specific purposes with limited duration.Thus Tyan adduces the Mutayyabün and the Ahláf as examples of temporary alliances, but the oaths taken by the parties clearly show that the time limit set for these covenants was "for ever" ^^.Indeed, the awareness of these alliances within Quraysh was deep and durable, as shown by Kister.^^ Secondly, the sources mention alliances between more or less equal groups that appear to have agreed on general co-operation.^^ Alliances of this kind sometimes led to a merging of the parties although, as far as I am aware, merging was never something explicitly agreed upon but rather something that occured gradually.^^ The process may have been as follows: independent sections conclude an alliance; the confederacy sometimes assume an aggregate name, or is given one by outsiders; later, the descent groups constituting the confederacy claim common descent, often interpreting the aggregate name as indicating a common ancestor.Thus certain sections which pledged themselves to act in co-operation "like finger-joints" (barâjim, pi. of burjuma), were called al-Barájim, "the Finger-joints", and we are told that they all had a male ancestor in common, Hanzala (of the Tamïm tribe).^^ Goldziher 28 Tyan,Hilf,338,vs. Ibn Hisham,Sîra 1/263;Ibn Sa'd,Tabaqât Mil.cf.Juda, Mawâlî, chapter 1 note 32, where he adduces an example of temporary alliance (cited from Bakri, 1/41), whereas the text is clear about the original intention of the parties to stay allied forever.See also Goldziher, Muslim Studies 1/68-9, and Obermann, "Early Islam", 245, both distinguishing between temporary and permanent alliances.Generally, however, the classifications of alliances are not clear.
(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.esadduces this example to argue that "confederacies could spring up between the lineages of a great tribe who were brought together by common interests"; presumably he means sections which had drifted apart, and wished at some point to come together again.^^ Perhaps the argument is correct, but I suggest here another possibility.It seems to me that alliances for specific purposes (type 1 above) could be concluded within one and the same tribe, as the examples of the Qurashi Mutayyabûn and Ahlàf shows.But a confederacy within one and the same tribe seems to me redundant.Had these sections, which became "The Finger Joints", really believed themselves to be agnatically related, they would not have needed to conclude an alHance and assume an aggregate name; they could have united under the name of their common ancestor, even if they had already drifted apart due to segmentation.^^ I therefore suggest that the Tamïmï genealogy of the Barajim groups is not the original one.That I found no indication of the earlier genealogies of any of the "Finger-joints" sections, is probably due to the incomplete state of our sources.Another possibility is that in the case of al-Barájim the process of merging was complete, and the original genealogies forgotten, before the Arabian genealogies were put in a written form.In contrast, two groups, Banü Bishr and Banü Ya'nuq, concluded an alliance and were jointly called "Al-Adïm", "the piece of leather", because they wrote their agreement on parchment.^^ I did not find al-Adîm as a name of an ancestor, nor did I come accross any claim of these groups to common descent.Again, this may be due to the state of the sources or to mere accident.But it is also possible that the tribal genealogies crystallized and were put into writing before these two specific groups completed their merging.If they ever became one group with alleged common ancestry, that group was formed too late to be recorded in the classical genealogy books.Another case is the group called al-Ribáb, which comprised six sections.Their aggregate name reflects the ceremony performed in order to ratify the alliance: hands were dipped in finit juice (rubb).They too are considered by our ^^ Goldziher, Muslim Studies 1/68.^' ^ But cf.Stewart, Texts 1/87-8.For legal purposes, that is, the drawing of the boundaries of co-liable groups, common ancestry is not generally adequate; the agnates usually conclude a pact among themselves, cf.Marx,Bedouin,143,where   extant sources as agnatically related, although their ancestors are not all brothers.^^ Dipping of hands, this time in blood, occurred in the alliance which produced the Khath'am tribe.According to a certain account, Khath'am was the name of the camel whose blood served in the ceremony by which the alliance was formed.Yet "Khath'am" is also taken to represent an individual, the founder of this group.The camel, originally part of the ceremony of the alliance, became "a camel that was in possession" of the ancestor.The "real name" of the ancestor is sometimes given as "Half b.Aftal"; Half, not surprisingly, derives from the same root as "alhance" (hlf).^^ Similarly, the tribe called Madhhij is reported to have aquired its name from "a tree by which they took the oath of alliance", but the name aquired a genealogical meaning as well, and it is recorded as "Madhhij, son of Yuhábir, son of Malik", Since the word "Madhhij" was not widely recognized as a proper name, a comment was added, "they are descendants of Madhhij, whose name was Malik b.Udad".^^ Many examples may be added, and indeed certain scholars consider confederation, not segmentation, as "the original type of Arab tribe formation"; but, as far as I can judge, both mechanisms were at work simultaniously, constantly forming new groups.3^ The third type of alliance is the attachment of people to descent groups other than their original ones.As a rule, it was small sections that were accepted as allies by larger, presumably stronger ones.More often than not, these smaller sections came to live with their allies, and also added the latter's nisba to their own.As will be explained below, the smaller or weaker section in fact was accepted into ^^ For more examples see Juda, Mawâlî, 7, 10.Goldziher, Muslim Studies 1/65-70, speaks of the process of alliances turning into (allegedly) descent groups, but does not explain the transformation of an aggregate name into an alleged ancestor.He also blurs the distinction between merging such as described here, and the absorption of weaker groups by stronger ones, which I consider as the third type of alliances.Tyan, "Hilf", 339, states that according to AI-Hqd al-fand most of the Arab tribes were formed by alliances, but he cites no reference; the Kitab al-nasab in the Iqd, however, consists of genealogies recorded as hierarchies of genuine descent groups.See also Obermann, "Early Islam", 245 note 4.
(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.es the co-liable group of the larger or stronger one.I therefore call this kind of attachment "hosting alliances", the weaker sections that joined stronger ones "guest allies", and the absorbing sections "host allies"."^^ Sometimes a guest section first forged the alliance, and later forged a genealogy, to integrate itself into the male line of the host section.This process is described by the terms dakhalüji,.., "they entered into" (the name of a host section follows), and 'Hntasabufi... '\ "they traced their pedigree in" (or: "they claimed descent from)" (the name of a host section follows).For example, the section Banù Aklub b.Rabï'a b.Nizár concluded an alliance with the Khath'am tribe, whereupon they joined the Khath'am in their dwelling place.The Aklub later changed their own genealogy, replacing Nizár with 'Ifiris; they claimed that they were descendants of Aklub b.Rabï'a b. 'Ifris, 'Ifris being a name in the male line purportedly descending from Khath'am.^^ Describing the vicissitudes of the tribal system the geographer Bakrï remarks, wa-l-qawm haythu wada'û anfusahum, "the people's places [in the genealogical scheme] are determined by themselves".42 A systematic, deliberate distinction between types of agreements is lacking in the sources.Thus it is, for instance, difficult to determine whether the ceremonies described in the reports are pertinent to all types of alliances or only to some of them.In what follows I focus on hosting alliances, although I cannot be sure that my description does not sometimes apply to other types of alliances as well.I discuss the material under two categories, namely, the rights and obligations of allies, and the precise identity of the parties to alliances.These categories do not exist as such in the sources, whether works of hadfth, jurisprudence, Qur'án exegesis, history or adab.

Rights and obligations of allies
Hosting alliances entailed specific rights and obligations, which were incumbent upon all the parties to the agreement.There is, first, ^^ By contrast, Marx,Bedouin,[129][130][131][132]distinguishes   indirect evidence of this fact, namely, the pejorative terms zanim, mulsaq, and mulsaq bi-ghayr hilf.Zanim is an outsider informally staying with a descent group, and also: someone known to be an evil person, or a troublemaker (yu 'rafu bi-Usharr).The term mulsaq, attached, literally: glued, is explained as someone staying with a descent group without being related to it by genealogy nor by alliance.
In Arabian tribal society where genealogy was the main claim to status, this term had a strong negative connotation.The same applies to mulsaq bi-ghayr hilf, attached without an alliance.^^ A differentiation between the allied status and other categories also follows from comments found in the sources, such as "I think that [the family Sayhán] was not allied to Banù 'Abd Manàf but merely stayed with them"."^"^ There was thus a distinction between people informally residing with a group other than their own, and allies who, in fact, did not necessarily reside with their partners.^^ This distinction favors the allied status as opposed to informal relationships, presumably because the former was more secure for being institutionalized, that is, having definite, firmly guaranteed rights and obligations."^^ The direct evidence pertinent to the rights and obligations of allies is here examined under two categories.One of them may be termed scholarly evidence, consisting of statements by scholars who define and describe alliances.The other is narrative evidence, by which I mean reports of steps taken by allies in specific situations of crisis.Accounts of the stipulations made in concrete cases would belong in this category, but they are mostly given in the form of the traditional oath of alliance, so that they do not add to the knowledge derived from the scholarly evidence.

Scholarly evidence:
The scholarly statements are often general, saying that parties to alliances pledge themselves by oath to adhere to one another (an yalzama ba'duhum ba'dan), to give succor to one another (nusra, i'ma), to act as one body in matters of defence and support (amruhumá wáhid fî al-nusra wa4-himâya).^'^ Sometimes, however, more specific stipulations are reported in scholarly statements; these are 1) mutual inheritance and jointly owned property, and 2) common liability.
1) Mutual inheritance, and sometimes joint property, are mentioned in scholarly statements about hilf; obviously, hilf in these cases means a hosting alliance, not a political or military treaty.However, it is not clear whether mutual inheritance was part of the rights pertinent to the institution, or a specific type of agreement between two individuals.The latter is implied by certain discussions of inheritace laws, and one may even encounter the term hilf aUtawàruth.^^ This may be due to the fact that inheritance is always connected to the death of an individual.Indeed it is not specified exactly how a mutual inheritance arrangement worked, if the alliances were concluded between groups.Presumably it applied to the representatives of the two parties, or only to property jointly owned by each one of the parties, such as houses."^^ However, references to "a covenant about mutual inheritance" are rare, and such a stipulation is usually recorded as part of a broader agreement.A typical formula of alliances reads, "my blood is your blood, my property is your property, you will give me succor and I shall give you succor, you will inherit from me and I shall inherit from you".^^ This formula stipulates both the sharing of property and mutual inheritance.The matter of inheritance is raised in connection with a phrase in Qur'in 4:33, namely, "and those with whom you have agreements by oath give them their due" (wa-lladhma 'àqadat (ox 'aqadat) aymanukum  fa-âtuhum nasîbahum).The connection with the Qur'àn arouses the suspicion that inheritance by alliance was an invention of the exegetes rather than a real practice.The following interpretation is in particular unconvincing: It is explained that in the Jahiliyya, a man would ally himself to others promising that he would assist them in securing their rights, but without their having any obligation towards him; the verse abolished this unjust practice.Obviously, this is a concocted elucidation of "give them their due".^^ There is, however, evidence that seems to show that mutual inheritance was indeed practiced.First, a specific share, a sixth, is mentioned for the ally, and this lends the reports some credibility.^^ Secondly, the subject is also discussed independently of the verse.The hadlth "there is no alliance in Islam" was adduced in order to abolish the practice of inheritance by alliance.^^ Thirdly, the formulae of alliance associate inheritance with the duty of paying blood-money.This association is attested for agnates, ^^ and there is no reason to reject it for allies, given that alliance simulates agnatic relations.Fourthly, and perhaps more convincingly, certain jurists, mainly Hanañs, accepted the practice of inheritance by alliance.^^ This seems to prove that the practice really existed, since jurists are more likely to legitimize an existing practice than to invent and defend one that a) never existed, and b) does not fulfill any urgent need.
52  ing outsiders into agnatic co-liable groups.Like the agnaticallyrelated members, the allies were supposed to bear the consequences of the offenses of their partners.The document composed by the Prophet, the 'ahd al-umma, contains a clause saying "a man should not bear the guilt of his ally".^^ The existence of such a clause indicates that common liability of allies was the rule in pre-Islamic Arabian society.
Statements by Muslim scholars reveal that common liability was a major theme of hosting alliances in the Jáhiliyya.^^ Allies were supposed to protect their partners, to avenge their blood if spilled or to compensate their agnates with blood-money, as well as pay blood-money on their behalf when they were involved in homicide.However, genuine tribesmen were not supposed to be killed in retaliation for the murder of allies.^^ There are various versions of the oath taken on the occasion of alliances, such as, "my blood is your blood, my unavenged blood is your unavenged blood (hadmi hadmuka), my vengeance is your vengeance, my war is your war, my peace is your peace, you will inherit from me and I shall inherit from you, you shall be sought for my offenses and I shall be sought for yours, you shall pay blood-money on my behalf and I shall pay blood-money on your behalf.^^ The reason why the formula is pronounced in the singular form is apparently that the ceremonies were performed by individuals.The commitments, however, were meant to join the allies in one co-liable group, as the 56 Ibn Hisham, Sira 3/33.5^ Al-Hanafí, Mu'tasar 2/109; Ibn Hajar al-'Asqalànï, Fath al-bârî 10/502.An ally, as well as a protected individual (j^^)^ could be punished for the crimes of their partners, ibid.12/211.Co-liability applied to allies, and to other attached people as well ('adid, on which see "Alliances in Islam"), al-Shaybânï, Mabsût 4/661-3, see also Juda,Mawàlî,Bedouin,clearly distinguished between alliances and co-liable groups; Bedouin allies in the Negev do not share liablity (but he does not distinguish between types of alliances).
5^ Ibn Hajar al-*Asqalànî, Fath aUbârî 7/111.5^ damî damuka wa-hadmí hadmuka wa-tha 'rî tha 'ruka wa-harbT harbuka wa-silmï silmuka wa-tarithum wa-arithuka wa-tutlab bi wa-utlab bika wa-ta 'qil 'annî wa-a 'qil 'anka, al-Qurtubî, Tafsîr 5/166, see also al-Tabarï, Tafsîr 5/53; al-'Azîmâbàdî, 'Awn 8/96; Ibn Qudâma, Al-mughnî 6/205; 'Abd al-Razzáq, Musannaf 10/305; see the references in Juda, Mawàlî, pp. 3, 5. Juda follows the usual (and erroneous) translation of hadmî hadmuka as "my destruction is your destruction".Reinert, Das Recht, 22, records one, very deficient form of the oath, under walà\ clientage, and not under hilf.See also Smith,Kinship,56.(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.esphrase "my war is your war" clearly indicates.Another pointer in the same direction is the version that runs "whoever kills one of my co-liable group, it is as if he killed one of yours (man qatala waliyyl fa-qad qatala waliyyaka).^^ The most obscure phrase in the formula is hadmî hadmuka.The Prophet used the phrase when he concluded his agreement with the Ansar-to-be at the ' Aqaba, and Muslim scholars offered various explanations.^^ Some of these explanations are obviously reflections of the particular relations between Muhammad and the Ansar, such as the fact that he came to live among them.For example, hadmï hadmuka, or hadamï hadamuka, is interpreted as "my house is your house" (manzil), because a house may be ruined (qad yuhdamu) and therefore it may be called "ruin" (hadm, hadam).The dictionaries, however, also record that hdm is equivalent to hdr, a term applied to blood revenge that was waived.The content of the oath clearly indicates that the latter was the original meaning of the phrase.There is not much sense in stating "my blood is your blood, my house is your house, my vengeance is your vengeance" etc., or "my blood is your blood, my grave is your grave, my vengeance is your vengeance", and so on.There is much more sense in a statement which runs "my blood is your blood, my unavenged blood is your unavenged blood, my vengeance is your vengeance" etc.Another indication that this is the correct interpretation is the proposition made by 'Abd al-Rahmán, of the Qurashï section Taym, to the famous jurist Malik b. Anas (d. 179/795).He proposes only two conditions, an yaküna damunâ damuka wa-hadmunà hadmuka.These two elements must be complementary, not disconnected, meaning that in matters of blood revenge the two parties would act together, whether in seeking revenge or waiving it.^^ Whatever the meaning of hadml hadmuka, common liability between allies is amply established by the rest of the formula.

Narrative evidence:
The general terms of mutual assistance occur also in reports of particular alliances.^^ In addition, reports about the conduct of allies under various circumstances provide evidence of the effects of alliances.A few examples foUlow.
The section Sayhan, originating in the Muharib tribe, was allied to the Umayyads, of the Quraysh.The Umayyad caliph Mu'awiya intervened on behalf of one of tíiem, the poet ' Abd al-Rahman b.Artát b.Sayhan; he annulled the punishment which the govemor of Medina inflicted upon the poet for wine-drinking.Since the punishment had already been carried out, the point of Mu'awiya's measure was the rehabilitation of the poet's reputation.^ However, Mu'awiya, while protecting his halfJbn Artát from the govemor of Medina, did not hesitate to punish another one of his allies when he thought it appropriate.^^ Mu'awiya's son, Yazïd, asked his father to assign allowances to the orphans of the Qurashî sections Jumah, Sahm and 'Adî.Mu'awiya inquired why Yazïd was concerned about them, whereupon Yazïd replied that they were allied to him, and had moved to his residence.^^ It should be noted that in pre-Islamic times the sections Jumah, Sahm and 'Adï belonged to the intra-Qurashï alliance called the Ahlàf, whereas Yazïd's family, the Umayyads, belonged to the rival alliance, the Mutayyabùn.^^ Another remarkable point about this report is the implication that Yazïd was committed to an alliance of which his father was unaware.
The case of the poet Ibn al-Mufarrigh al-Himyarï is somewhat complicated.He was aUied to one of the Umayyad families, that of KhaUd b.Asïd.However, he joined the entourage of the govemor ' Abbád b.Ziyad, son of Ziyad b.Abïhi.Ziyád, a former govemor of Iraq, was related to the Umayyads because the caliph Mu'awiya had recognized him as his brother.When Ibn Mufarrigh got himself into trouble with his patron, Ziyád's son, his Umayyad ally Khálid b.Asïd (among others) refused to help him.He therefore sought the protection (jiwàr) of the father of Ziyád's daughter-in-law.For some reason, the notable Talha al-Talahat, himself a Khuzâ'ï (but related to Quraysh on his mother's side), pressured the Umayyad allies of Ibn Mufarrigh to assist him, saying "he is your ally, he is counted among you, he is one of you" (hallfukum wa-'adidukum wa-rajul minkum).An expedition comprising members of several sections, Umayyads included, appealed to the (Umayyad) caliph Yazïd b.Mu'awiya on behalf of Ibn Mufarrigh.The appeal involved a reference to Ibn Mufarrigh's status as an ally, and Yazîd eventually complied and rescued the poet from the sons of Ziyàd.There are indications that the intervention on behalf of Ibn Mufarrigh was motivated by hatred of the sons of Ziyad b.Abïhi rather than by codes of social institutions.^^ Intervention on behalf of allies was not restricted to the royal family.During Muhammad's raid on Muraysi' an ally of the Banù 'Awf (of the Ansàr) quarreled with an employee (ajïr) of 'Umar b. al-Khattâb.The Ansàr responded to the tribal war-cry uttered by their ally (yà la4-ansàr).So did, however, the Muhâjirùn, called up not by an ally but by an employee.^^ Ibrahim b.Muhammad b.Talha and his section the Banù Taym of Quraysh stood by their allies, the descendants of the Companion Suhayb b.Sinan, in a time of crisis.When the caliph Hisham b. 'Abd al-Malik decided to reduce the pension of Suhayb's descendants from the level of allies to that of clients (mawâlî), the Banù Taym protested by reñising to collect their own pensions.As a result, the caliph withdrew his decision.^^ There were also cases where the allies did not live up to their obligations.The Prophet's uncle, 'Abbas, was captured by the Muslims at Badr together with his nephew(s) and an ally.Upon the Prophet's request, he paid ransom for himself as well as for his nephew(s), but withheld the payment for his ally.Eventually he handed it over only because he feared for his good reputation.'^K ^^ Al-Isfahânï, Aghânî 18/262-94; Ibn Qutayba, Al-shi'r wa-1-shu'arà' (Cairo) 1/360-65; al-Jumahî, Tabaqât, pp.143-4.The rivalry is reflected in Ibn Mufarrigh's poetry, see Ch. Pellat, in Mélanges Louis Massignon, Damascus 1957, 3/195-232   The foregoing examples illustrate the general concept of nusra, assistance, undertaken by allies.As for the specific obligations, we may again consider the evidence under 1) mutual inheritance and jointly owned property, and 2) common liability.
1) Narrative evidence of mutual inheritance and joint property is sparse.A report of an agreement between two Jâhilî poets, Durayd b. al-Simma and Mu'áwiya b. 'Amr, is sometimes interpreted as an illustration of inheritance between allies; ^^ this interpretation, however, is mistaken.The two poets promised one another not an inheritance, to be bequeathed by the one who dies first, but an elegy, to be composed by the one who dies last (read an yarthiyahu, not an yarithahu).They are indeed also reported to have undertaken mutual responsibility for blood revenge, but the elegy is the core of the anecdote.The story goes on narrating that Mu'áwiya b. 'Amr was murdered, and Durayd composed an elegy on him; no more is said about blood revenge.In fact al-Isfahânî adduces the anecdote as an introduction to the elegy, which he records, among other verses by Durayd, in the latter's biography.Another dubious report states that when the caliph Abu Bakr died, a sixth of his property was bequeathed to one of his allies.The scholar reporting it remarks, "I found no one who had heard about it".Indirect evidence of mutual inheritance is supplied by the anecdote about Kinána b. 'Abd Yâlïl, a notable fi*om the Thaqïf tribe.The reason for his reñisal to convert to Islam was that he rejected the idea that a Qurashî should be entitled to his inheritance (là yarithunî rajul min quraysh).The report may be fictitious, but it reflects the way some people grasped the act of conversion to Islam: it was seen as an alliance with Quraysh.Furthermore, it attests to the rule of inheritance between allies because, had no such rule existed, this report would have made no sense.
The only report about inheritance which has a real ring of truth is the one concerning the daughter of the Companion Sa'd b. al-Rabî'.She complained that she received nothing of the estates of her father and her uncle, who had both been killed at Uhud.Their allies, she alleged, had taken the whole of both estates.I see no reason to dismiss ^^ Juda, Mawali, p. 5;Crone, Roman, Provincial and Islamic Law, 53, both quoting al-Isfahânï, Aghànî (Cmo 1927-72)  this report, and, again, it would not have been fabricated if inheritance between allies had been an unknow practice.
Jointly owned property is attested by one piece of evidence.When the allies of the Umayyads, Banù Ghanm, emigrated to Medina with the Prophet, the Umayyad leader Abu Sufyan seized their houses in Mecca and sold them.According to 'Umar b.Shabba Abu Sufyan had a right to do so, on account of the alliance between the two sections.^^ 2) The narrative evidence of common liability is more substantial.The obligation of allies to bear the consequences of the offenses of their partners is demonstrated by the following incident.Companions of the Prophet were taken prisoners by people from the Thaqîf tribe.Muhammad retorted by seizing an ally of Thaqîf, a man from the tribe of 'Uqayl.The 'Uqaylî asked why he was taken, having done no wrong, and the Prophet replied "I seized you because of the crime of your allies'" (bi-jarirat hulafà'ika).Notwithstanding, the Prophet stipulated in 'ahd al-umma that "a man should not bear the guilt of his ally".^4 Illustrations of alliances functioning in matters of blood revenge are listed in what follows.
Abu Uzayhir, from the Daws tribe, was allied to the [co-liable group of] the Umayyad leader, Abu Sufyan b.Harb.Abu Uzayhir was killed by people of the Qurashî section Makhzum, but Abu Sufyan refrained from avenging his blood in order to avoid bloodshed within Quraysh.Instead, he sent blood-money to Abu Uzayhir's family.
(c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.esblood-money to the Hadramî family and to compensate them for the lost property, rather than avenge 'Amr's blood.The brother of the deceased, who was also allied to 'Abd Shams, protested strongly and threatened to transfer his alliance to another Qurashï section (but he never did).Abu Jahl, of the rival Qurashï section Makhzùm, attempted to exploit this affair in order to undermine the prestige of the 'Abd Shams section.
Solidarity with allies continued into Islamic times in a truly Jâhilï spirit, sometimes with a complete disregard for the newly introduced values and institutions.The ally and client of the Makhzüm section, the Companion 'Ammàr b.Yásir, was beaten by the caliph 'Utmán's men.The Makhzùmïs decided to kill the caliph if ' Ammar should die as a result of his injuries; they ignored the Jáhilí rule that no genuine tribesman is to be killed in retaliation for the killing of an ally, let alone the Islamic rule forbidding bloodshed among Muslims; neither did they pay any regard to the sanctity of the caliphate.^^ The Jáhilí rules of co-liability influenced the Prophet's conduct as well.His Khuzà'î allies were assaulted by allies of Quraysh, whereupon the Khuzâ'ïs appealed to Muhammad's assistance, even though they were not yet Muslims at the time.The conquest of Mecca by the Prophet was the outcome of that appeal.^^ A case where the guest-, not the host-ally shouldered responsibility is that of Hujayr b.Abu Iháb, an ally of the Qurashï section Banü Nawfal.When his ally and brother in law, al-Hárith b. 'Àmir, was killed by Muslims, Hujayr bought a Muslim captive and handed him over to the dead man's son, who executed him.
Co-liability within the injuring rather than the injured party is illustrated by the case of the Banü al-'Arába, from the Sulaym tribe.They were involved in homicide within their own tribe, and were protected from blood revenge by their ally 'Ata' b.Abu Jalïd al-Khuzâ'ï al-Himyarï.' Atâ' promised to pay the blood-money on their behalf.^^ ^^ The story of Abu Uzayhir: Ibn Habib,Munammaq,[199][200][201][202][203][204] In conclusion, it may be stated that, as far as the rights and obligations of allies are concerned, the scholarly and the narrative evidence is not in conflict even though the narrative evidence is deficient where mutual inheritance is concerned.Finally, it should be mentioned that the parties to an alliance could make special stipulations that did not apply in all alliances.Such stipulations are not covered by scholarly statements.For example, it is reported that al-Damùn b. ' Abd al-Malik from Hadramawt, a fugitive from blood revenge, allied himself to the Thaqîf tribe in pre-Islamic times.It was stipulated that he pay for the construction of a wall around the town of Thaqîf, al-Tá'if, and that the two parties to the alliance exchange women in marriage.^^ Exchange of women between allied parties was common (although I found no scholarly statement to that effect), but paying for a wall around Ta'if is obviously peculiar to this particular alliance.
The identity of the parties to alliances Since hosting alliances involved rights and duties, it was important to know who the parties were in any given case.It seems, however, difficult to acquire such information from our sources.The names of individuals and the names of their descent groups are mentioned interchangeably as partners to alliances.For example, Muhammad b. 'Abdallah b.Jahsh (a Successor (tàbi ') is once defined as one of the allies of the Qurashï section Banü 'Abd Shams, another time as an ally of Harb b.Umayya (b. 'Abd Shams).The descent group of this Successor, the Banù Ghanm, is designated alternatively as allies of the Banü 'Abd Shams, of Harb b.Umayya (b. 'Abd Shams) 'Ubaydallah (or his son) alone or upon other members of the Taym section as well?The alliance obviously persisted, and was sometimes called after 'Uthmán b. 'Ubaydallah, even generations after his death.References to Malik's descendants as allies of 'Uthmán occur as late as in the third century A.H. Did the rights and obligations rest only upon the offspring in male line of 'Uthmán b. 'Ubaydallah or upon other Taymis as well, or upon Taym as a whole?A somewhat different example is that of Ja'wana b.Sha'üb, who is said to have been an ally of one or the other of Muhammad's uncles, Hamza and al-'Abbas.This seems to imply a boundary: If Ja'wana was allied either to one man or to the other, then it would appear that the rights and obligations entailed by the alliance belonged to either one or the other.But sometimes Ja'wana is called "an ally of the Banû Háshim", the section that included both Hamza and al-'Abbás.^^ Various sections at various levels of segmentation within one and the same tribe, as well as the whole tribe, are mentioned interchangeably as the parties to one and the same alliance.For example, the allies of Ghazwán b. 'Utba are variously said to be the Qurashi sections Banü 'Abd Manáf, and Banü Nawfal b. 'Abd Manáf ^^ The former designation technically comprises the latter, as well as the Banü Háshim, the 'Abd Shams and the Banü al-Muttalib, each of them apparently a co-liable group in its own right.Another version has it that 'Utba b.Ghazwán was allied either to the Banü 'Abd Shams or to the Banù Nawfal (who were both sections of 'Abd Manàf).^^ Thus it is difficult to know whether only Banü Nawfal were a party to that particular alliance, or another section of 'Abd Manàf, or the whole of 'Abd Manàf Similarly, the Khuzà'î section of Sibà' is defined once as allies of the Qurashî section Banù Zuhra, another time as allies of the Banù 'Awf b. 'Abd 'Awf from Zuhra.^^ Another example is that of al-Miqdàd b. 'Amr, who was not only allied to, but also adopted by the Zuhrï al-Aswad b. 'Abd Yaghùth.The sources often refer to al-Miqdàd as an ally of the Zuhra section as a whole.^^ The formulation in the case of the small descent group Sayhàn, originating in the section Jasr b.Muhàrib, is puzzling.They were allies of Harb b.Umayya and "were considered as part of [the group of Harb] in particular, and of the Banù Umayya in general" (bi-manzilat ba'dihim 'indahum khâssatan wa-Hnda sa IT bam Umayya 'âmmatan).^^ In Islamic times, long after Harb b.Umayya had died, a quarrel broke out among the Umayyads on account of one of the Sayhànï allies, the poet Ibn Artàt.The governor of Medina, the Umayyad Marwàn b. al-Hakam, punished Ibn Artàt for drinking wine, whereupon the caliph Mu'àwiya scolded Marwàn, saying, "you beat him [...] when he was allied to [my father] Abu Sufyàn, by God, had he been allied to your father al-Hakam you would not have punished him".^^ This indicates a division within the Umayyad section into independent co-liable groups.Yet according to another opinion, the Sayhànîs were attached not to a mere section of the Umaj^^ads (Abu Sufyàn and his son) but to the whole of Banù 'Abd Manàf (which comprised the whole of the Umayyads as well as other sections).It is recorded that "[the Qurashî section] Banù 'Abd Manàf loosened their relationships with the Sayhànîs; the latter are [now] not allied to the Banù 'Abd Manàf but merely reside with them".^^ Perhaps the ^^ The text reads wa-banü 'abd Manàf tuqwf hilfahum, wa-hum 'indi a'izzà'uhum wa-laysil bi-ahlâfihim, read a 'râ'uhum for a Hzzà'uhum, al-Isfahânî, Aghânï(Bemxt) 2/236.See also Ibn Habib,Munammaq,252, where they are defined as Hdâd, i.e. "counted with" (for administrative purposes, see Landau-Tasseron, "AlHances in Islam").The text could be read tuqaMnvi, "Banù 'Abd Manàf buttress [the alhance]", but I think that this reading (c) Consejo Superior de Investigaciones Científicas Licencia Creative Commons 3.0 España (by-nc) http://al-qantara.revistas.csic.esSayhànîs were rejected by their allies (whoever they were) precisely because of the shame caused by the behaviour of Ibn Artât.Be that as it may, the data concerning the Sayhànîs seem to be contradictory.It is not clear whether they were attached to one Umaj^ad family, to the Umayyad section as a whole, or to the larger descent group, 'Abd Manâf.
Occasionally, clear specifications of the parties to alliances occur in the sources, expressed by the terms thumma or khassatan.The poet Yazîd b.Mufarrigh al-Himyarî was "an ally of Quraysh, and, more specifically (thumma), of the family of Khálid b.Asid", one of the families of Banu Umayya.The Asadï section Banù Ghanm was allied to Banu 'Abd Manáf, but, according to another version, to Banù Umayya specifically (hulafa'banïUmayya khassatan duna banî 'Abd Manáf).Thumma sometimes indicates an individual.Sa'd b.Khawla was allied to the Qurashi section 'Amir b.Lu'ayy and, more specifically (thumma), to Abu Ruhm b. 'Abd al-'Uzza.^^ The confiision about the precise identity of the parties to alliances in fact reflects the confusion about the boundaries of descent groups and co-liable groups, explained in the beginning of this article.The only additional problem is this: Whereas genuine tribesmen always had a group of agnates responsible towards them, it appears as if allies could have individuals responsible towards them.This is puzzling, for two reasons.First, it has been shown above that alliances concerned groups, not individuals.Secondly, the nature of the rights of allies makes alliances a matter for groups, not individuals.If one of these rights was retaliation for spilled blood, then obviously a group, not an individual, must be responsible to implement it.By way of settling this seeming contradiction I suggest the following: A guest ally in fact had to be tied to an individual within the hosting co-liable group in order to determine his place within the hierarchy of succor.By this term I mean the order in which the members of a co-liable group were called upon to act.In matters of blood revenge, and perhaps in other matters a well, the responsibility rested first and fore- most upon the closest male agnates, then on the farther circle of agnates, and so on (of course a large circle of agnates could be called upon already at the first stage of action).From an external vantage point, the members of a given co-liable group were seen as equals, indeed interchangeable, to the extent that any one of them was prone to be killed for the crime of another.From an internal point of view, there was a hierarchy, or rather a network of relationships determining the stage at which any member of the group had to help any other member.This explanation also accounts for the fact that Hujayr b.Abî Ihàb mentioned above helped the son of his murdered ally to avenge his father's blood, but did not himself carry out the act.^^ An ally who joined a co-liable group had to have his specific place within that network.This explains the role of the individual host who actually performed the ceremony of concluding the alliance.This individual host ñmctioned as the closest agnate in relation to the guest allies.The relationships thus created were sometimes named after the individual host even generations after his death.^^ The same individual, and afterwards one of his descendants, formed a point of reference for the guest allies when the need arose for action.My conclusion is corroborated by the following anecdote.The Qurashî nobleman 'Abdallah b.Jud'an was host ally of the family of Mas'ùd b. 'Amr al-Qârî.When 'Abdallah was on his deathbed his allies asked him to renounce the alliance, because he had no son.'Abdallah consented and the family of Mas'ùd transferred their alliance to another section of Quraysh.^^ The point of the allies' request must be, that when a host ally dies, a son takes over the responsibility towards the guest allies.Further evidence of this is the case of the son of Abu al-Jalîd mentioned above; he declares, "[Banù al-Ghazála of Sulaym] allied themselves to my father, therefore I shall pay the blood-money on their behalf.^^ Presumably, if he could not afford to pay, his co-lia-^' That the Hanafis insisted on the equahty of all members of any given co-liable group seems to vindicate my point, see al-Jassâs, Ahkâm 3/197.See also Marx,Bedouin,Kinship,  ble group would offer help.This means that obligations towards the guest allies rested in fact on a host co-liable group, not on an individual.The case of 'Amr b. al-Hadramî (and his family) clearly proves this argument.'Amr was "allied to 'Utba [b. Rabï'a] to the exclusion of the rest of Quraysh", but when he was killed by Muslims and the matter of blood revenge was raised, the whole of 'Utba's co-liable group, Banù 'Abd Shams, was involved.^^ In conclusion, the term hilf, translated as "alliance", may refer to a variety of agreements and attachments.The type which I have termed "hosting alliance" applied to descent groups, whereby one absorbed another, weaker or smaller one, to share co-liability with it.One person (and afterwards his male descendants in male line) within the hosting section would act as if he were the closest agnate of the guest section.The most clearly defined rights and obligations were shared legal responsibility and mutual inlieritance, but assistance was given in a variety of circumstamces.Needless to say, the real situations were much more complex than any schematic description can convey.Bibliography