I should like to thank Amikam Elad, Chase Robinson and Luke Treadwell for most helpful criticism of earlier drafts of this article. The following abbreviations have been used: AA: see note 61. Aghānī: see note 68. BA: see note 66. BF: see note 61. El2: Encyclopaedia of Islam, second edition. Tab.: see note 2. TB: see note 58. “Turk”, C and L: see note 12. YB: see note 37. YT: see note 25.

¹ Zakeri, M., Sāsānid Soldiers in Early Muslim Society (Wiesbaden, 1995), esp. pp. 265ff.Google Scholar

² al-Ṭabarī, Ta'rīkh al-rusul wa'l-mulūk, ed. de Goeje, M.J. and others (Leiden, 1879–1901) (hereafter Tab.), Glossarium, s.v. ‘bny’.Google Scholar

³ B. Lewis in Elz, s.v. “abnā”’.

⁴ D. Ayalon, “The military reforms of Caliph al-Mu'taṣim. Their background and consequences”, paper read in New Delhi 1964, published (after circulating as a stencilled pamphlet for thirty years) in his Islam and the Abode of War (Aldershot, 1994), p. 6.Google Scholar Ayalon's conclusion was anticipated by Mez, A., Die Renaissance des Islams (Heidelberg, 1922), p. 151; English tr. (Patna, 1937), p. 155.Google Scholar

⁵ Zakeri, pp. 191ff, 265ff.

⁶ EI2, s.v. “Abnā”’, cf. also the references in Zakeri, pp. 328f.

⁷ Zakeri, p. 289. Note also his conviction that the subgroup of Tamīm known as the Abnā’ must have been Iranian by origin (pp. 270ff).

⁸ Its unspoken nature does generate tensions at times, cf. p. 275, where he says that the term abnā’ al-da'wa means “sons of the revolution”. If so, why should they have more in common with abnā’ al-aḥrār than with, say, abnā' al-muhājirūn, abnā' al-'arab or abnā' al-kuttāb? One assumes he would invoke word-play (the Abnā' in the Persian sense were legitimated as Abnā' in the revolutionary sense), but he never confronts the question.

⁹ Zakeri, p. 267, explains abnā' al-aḥrār is an Aramaicism meaning the same as al-aḥrār with reference to Nöldeke, Th., Geschichte der Perser und Araber zur Zeit der Sasaniden (Leiden, 1879), p. 235n. But Nöldeke's Aramaicism is banū 'l-aḥrār, and the frequency with which the gentry are called al-aḥrār in pre-Islamic contexts (e.g. Zakeri, pp. 267, 269n), abnā' al-aḥrār thereafter, suggests that the latter expression is idiomatic Arabic.Google Scholar

¹⁰ Tab. iii, 460.9 (year 159).

¹¹ Tab. iii, 1368.15.

¹² al-Jāhiẓ, , “Manāqib al-turk”, in his Rasā'il, (ed.) ‘Hārūn, A.-S. M. (Cairo, 1965), i, (hereafter C), pp. 630Google Scholar in his Tria Opuscula, ed. van Vloten, G. (Leiden, 1903) (hereafter L), pp. 40fGoogle Scholar= Walker, C. T. H. (tr.), “Jahiz of Basra to al-Fath ibn Khaqan on the ‘Exploits of the Turks and the Army of the Khalifate in General’”, Journal of the Royal Asiatic Society, 1915, p. 679.Google Scholar

¹³ al-Jāhiẓ, , “Fakhr al-sūdān 'alā '1-biḍān”, in his Rasā'il, ‘Hārūn, A.-S. M. (Cairo, 1965), i, p. 220.7;Google Scholar in his Tria Opuscula, ed. van Vloten, G. (Leiden, 1903), p. 82.12.Google Scholar

¹⁴ Jāḥiẓ, “Turk”, C, 8.-3; L, 4.1 = 636.

¹⁵ Jāḥiẓ, “Turk”, C, 26.2; L, 15.2, 7 = 651.

¹⁶ Below, notes 28, 53. This and the following expression may of course also refer to people of civilian Khurāsānī descent. Not all the Iraqi scholars described as abnā' ahl khurāsān necessarily descended from members of the Khurāsānī army (e.g. Ibn Sa'd, al-Ṭabaqāt al-kubrā (Beirut, 1957–1960), vii, pp. 340, 342 (twice), 349, 351–3 (thrice), 355, 490; but some clearly did (348, 350).Google Scholar

¹⁷ Below, note 74; ‘Idhārī, Ibn, Kitāb al-bayān wa' l-mughrib, ed. Colin, G. S. and Lévi-Provençal, E. (Leiden, 1948–1951), i, p. 90.11.Google Scholar

¹⁸ Below, note 72.

¹⁹ Below, notes 27, 29, 32, 33; Tab. iii, 1169.9.

²⁰ Below, notes 26, 58, 63, 74; Jāḥiẓ, “Turk”, C, 8.-3, 77.2; L, 3 ult, 50.3 = 636, 689.

²¹ Below, notes 26, 30, 68; Jāḥiẓ, “Turk”, C, 26.2; L, 15.7 = 651.

²² Pace Zakeri, p. 275, who sets up a contrast between metaphorical “sons of the revolution” of the revolutionaries”.

²³ Aḥmad b. Abī Ṭāhir Ṭayfūr, Kitāb Baghdād (hereafter Ibn Ṭayfūr), ed. Keller, H. (Leipzig, 1908), pp. 143f;Google Scholar ed. al-Kawtharī, M. Z. (Cairo, 1949), p. 80.Google Scholar

²⁴ Cf. Arazi, A. and 'El'ad, A.,’ “L'Epitre a l'armée'. Al-Ma'mūn et la seconde da'wa”, Studia Islamica, LXVI–LXVII (1987–1989), i, pp. 52ff.Google Scholar

²⁵ al-Ya'qūbī, , Ta'rikh, ed. Houtsma, M. Th. (Leiden, 1883) (hereafter YT), ii, p. 437.11;Google Scholar cf. p. 437.3 (wa-man ḥaḍara min … al-quwwād), accepted as historical by Elad, A., “Aspects of the transition from the Umayyad to the ‘Abbāsid Caliphate”, Jerusalem Studies in Arabic and Islam, XIX (1995), p. 99.Google Scholar

²⁶ Tab. iii, p. 498; cf. EI2, s.v. “Bāramika”.

²⁷ Tab. iii, p. 531, reading ibn for ayna (twice) and dawla for walī, following MS C and the edition of M. A.-F. Ibrāhīm (Cairo, 1960–9), viii, p. 175 (where the retention of the second ayna must be a misprint). Compare al-Amin's characterization of the Abnā' as ahl al-sabq ilā' l-hudā (Tab. iii, p. 931.7).

²⁸ al-Dīnawarī, , al-Akhbār al-ṭiwāl, ed. Guirgass, V. (Leiden, 1888), p. 386.Google Scholar

²⁹ Tab. iii, p. 703.15.

³⁰ Tab, iii, p. 732.15.

³¹ al-Kindī, , The Governors and Judges of Egypt, ed. Guest, R. (London and Leiden, 1921), p. 147.Google Scholar

³² Ibn al-Nadīm, , Kitāb al-fihrist, ed. Tajaddud, R. (Tehran, 1971), p. 257.18, cf. p. 120. Ayalon mysteriously infers that al-Rāwandī's followers were called abnā’ al-dawla after their leader's book (“Reforms”, p. 33).Google Scholar

³³ A young man told Hārūn that he was min a'qāb abnā' hādhihi l-dawla, his origin being in Marw and his birthplace in Baghdād (Tab. iii, p. 672). Another young man min al-abnā' received money from the generous al-Faḍl b. Yaḥyā al-Barmakī (al-Jahshiyāri, Kitāb al-wuzarā' wa'l-kuttāb, ed. al-Saqqā, M.’ and others (Cairo, 1938), p. 195.2). Two Abnāwis rebelled in North Africa in Hārūn's reign (below, notes 60, 68).Google Scholar

³⁴ Ayalon, ‘Reforms”, pp. 51ff, 12ff; Hoffmann, G., “Al-Amīn, al-Ma'mūn und der ‘Pöbel’ von Baghdād in den jahren 812/13”, Zeitschrift der Deutschen Morgenländischen Gesellschaft, CXXXXIII (1993).Google Scholar

³⁵ EI2, s.v. “al-Ma'mūn”, cf. also Lapidus, I. M., “The separation of state and religion in the development of early Islamic society”, International Journal of Middle East Studies, VI (1975).Google Scholar

³⁶ Above, notes 12, 23.

³⁷ Tab. iii, pp. 1179 (Ḥarbiyya), 118 of (Abnā'); al-Mas'ūdī, , Kitāb murūj al-dhahab, ed. de Meynard, A. C. Barbier and de Courteille, A. J.-B. Pavet (Paris, 1861–1877), vii, pp. 118f;Google Scholar ed. C. Pellat (Beirut, 1966–79), iv, par. 2801; al-Ya'qūbī, , Kitāb al-buldān, ed. Goeje, M.J. (Leiden, 1892) (hereafter YB), p. 256.8 (Baghdādī masses).Google Scholar

³⁸ For this dīwān, see YB, p. 267.9; cf. also p. 262.10. The expression al-jund wa'l-shākiriyya seems to have been interchangeable with al-abnā' wa'l-shākiriyya (Tab. iii, p. 1510.3, 14).

³⁹ Tab. iii, p. 1510.14.

⁴⁰ Tab. iii, pp. 1579.1, 1582.9.

⁴¹ Tab. iii, p. 1463.2.

⁴² YT, p. 604.5.

⁴³ YT, ii, p. 618.5.

⁴⁴ A fact of which he is aware. He responds that it“ explicitly [sic] conceals numerous references to the military contingents of abnā' under the Umayyads and particularly during the early stages of the ‘Abbāsid revolution in Khurasan” (Zakeri, p. 274). But no such contingents are mentioned in the sources.

⁴⁵ The two nisbas are not interchangeable. The former is the singular of al-abnā' (or thus at least in Jāḥiẓ) while the latter is part of a name.

⁴⁶ I did so myself in Slaves on Horses (Cambridge,1980). Elad even supplies the men with the nisba al-Abnāwī (“Transition”, pp. 105, 123).Google Scholar

⁴⁷ Cf. above, note 12; Lassner, J., The Shaping of 'Abbāsid Rule (Princeton, 1980), ch. 5. The praise of the Turks was allegedly relayed to Ṭāhir, who died in 207/822 (“Turk”, C, 56; L, 35 = 673).CrossRefGoogle Scholar

⁴⁸ “If I have described things correctly, the Banawī is a Khurāsānī; and if the Khurāsānī is a mawlā and the mawlā is an Arab, then the Khurāsānī, Banawī, mawlā and Arab all come to a single class … the Turks are Khurāsānis and mawālī of the caliphs … and (the merit of) the Turk redounds to the whole community … If all the troops knew this, they would become forbearing, ill-feeling would vanish and rancour would die down” (“Turk”, C, 34 (omits “the merit of”); L, 21 = 658; cf. also C, 9ff; L, 4ff, = 637). That the purpose is conciliatory is also argued by Lassner, Shaping, p. 136, who does however envisage the five groups as (semi-imaginary) regiments.

⁴⁹ The Arabs vaunt pre-Islamic characteristics of theirs such as illiterate memorization of poetry and fondness for boasting competitions adjudicated by kāhins; the mawālī boast of their loyalty to their patrons (C, 21f, 23ff; L, 12, I3f = 647, 648f), i.e. neither claims military merit as a group distinct from the Khurāsānīs.

⁵⁰ “Turk”, C, 56.2; L, 35.9 = 673; cf. Crone, Slaves, appendix V, no. 4.

⁵¹ “Turk”, C, 58.7; L, 36.18 = 674.

⁵² He is first mentioned between 142 and 144 (Tab. iii, pp. 145f cf. EI2, s.v. “Muhallabids” for the date).

⁵³ “Dhamm akhlāq al-kuttāb”, in his Rasā'il, ed. ‘Hārūn, A.-S. M. (Cairo, 1965), ii, p. 207.4.Google Scholar

⁵⁴ YT, ii, p. 547; cf. Crone, Slaves, appendix V, no. 17.

⁵⁵ The Arabs claim “most of the naqībs”, the mawālī claim “the chief naqībs”, and the Khurāsānīs claim all twelve of them (“Turk”, C, 14.-2, 22.8, 24.-2; L, 8.9, 12.17, 14.11 = 642, 647, 650). Ayalon wrongly has the Arabs claim all of them (“Reforms”, p. 5); and Zakeri wrongly has the Abnā’ do so (pp. 275f).

⁵⁶ Cf. below, note 76; cf. also C, 31.-3; L., 19.7 = 656 (fa'l-banawī khurāsānī min jihat al-wilāda).

⁵⁷ “Turk”, C, 26.5, 13; L, 15.10, 18 = 651f; Zakeri, p 277.

⁵⁸ Ibn al-Jawzī, , Manāqib al-imām Aḥmad b. al-Ḥanbal, ed. al-Khānjī al-Kutubī, M. A. (Cairo, 1931), pp. 31ff,Google Scholar esp. 15.8, 17.-4, 19.14; al-Khaṭīb al-Baghdādī, , Ta'rikh Baghdād (Cairo, 1931) (hereafter TB), iv, pp. 412ff, esp. pp. 413.n, 415.3, II.Google Scholar

⁵⁹ Ibn Sa'd, v, p. 348; the nisba is given in Ibn Ḥajar, , Tahdhīb al-tahdhīb (Hyderabad, 1325–7), ii, p. 329, s.v.Google Scholar

⁶⁰ Khayyāṭ, Khalīfa b., Ta'rīkh, ed. Zakkār, S. (Damascus, 1967–1968), p. 748, in conjunction with YT, ii, p. 497; Ibn 'Idhārī, Bayān, i, pp. 89ff. The men with whom he rebelled were also Abnā'.Google Scholar

⁶¹ Tab. iii, p. 911; cf. Mas'ūdī, Murūj, vi, p. 475; ed. Pellat, iv, par. 2683, where the suggestion is made by al-ṣa'ālīk min aṣḥabihi wa-hum fityān al-abnā’ wa ‘l-jund. Al-Aghlab b. Sālim was one of the seventy missionaries and a Tamīmī from Marwarrūdh (Akhbār al-dawlat al-'abbāsiyya wa-fīhi akhbār al-'Abbās, ed. ‘al-Dūrī, A.-'A. and al-Muṭṭalibī, A. J. (Beirut, 1971) (hereafter AA), pp. 221; 335;Google Scholar al-Balādhurī, , Futūḥ al-buldān, ed. de Goeje, M.J. (Leiden, 1866) (hereafter BF), p. 233.8;Google Scholar Caskel, W. and Strenziok, G., Gamharat an-nasab, dasgenealogische Werk des Hišām b. Muḥammad al-Kalbī (Leiden, 1966), Register, s.v.).Google Scholar

⁶² “Turk”, C, 28.11; L, 16.17 = 653; Ibn ‘Idhārī, Bayān, i, p. 89; cf. Crone, Slaves, appendix V, no. 15.

⁶³ Tab. iii, p. 840.6; 931.18; Crone, Slaves, appendix V, no. 18.

⁶⁴ Cf. above, note 26; Crone, Slaves, appendix V, nos. 7, 13.

⁶⁵ Alī b. ‘Īsā was shaykh hādhihi 'l-dawla (Dīnawārī, p. 391) and the shaykh and kabīr of the Abnā’ (Tab. iii, p. 931.14), of whom he commanded 50,000 for al-Amīn against Ṭāhir (below, note 160). The Abnā' also honoured al-Ḥusayn b. 'Alī b. 'Isā (Tab. iii, p. 846) and were outraged when ‘Abdallāh b. ‘Alī b. ‘Isā was flogged (Tab. iii, p. 1001.19). They and other members of the family held high office down to the reign of al-Ma'mūn, and some of them eventually joined his side (Tab. iii, p. 882, 904; al-Azdī, , Ta'rikh al-Mawṣil, ed. ‘Ḥabiba, A. (Cairo, 1967), pp. 325, 328).Google Scholar

⁶⁶ AA, pp. 217, 220; al-Balādhurī, , Ansāb al-ashrāf (hereafter BA), iii, ed. ‘al-Dūrī, A.-'A. (Wiesbaden, 1978), pp. 137.5, 169.2, on ‘Isā b. MāhānGoogle Scholar.

⁶⁷ Tab. ii, p. 2004.14; cf. YB, p. 247.19 (with al-Anbārī for al-Abnāwī). The name of Abū Khālid's father was al-Hunduwān (Tab. iii, p. 1002.4). His son Muḥammad b. Abī Khālid was the shaykh of the Abnā' in the Zawāqīl affair, a loyal supporter of al-Amīn (Tab. iii, pp. 843, 848, 883) and the leader of the Abnā'/Ḥarbiyya after al-Amīn's death (Tab. iii, pp. 935, 998ff; YT, ii, pp. 532f, 547) together with his sons Hārūn, 'Abdūs and 'Īsā; 'Īsā inherited his leadership (Tab. iii, pp. 976, 978, 985, 1003f) and played a prominent role in the revolt against al-Ma'mūn (Tab. iii, pp. 1006f; Ibn Ṭayfūr, ed. Kawtharī, p. 79; ed. Keller, pp. 141f). Al-Ma'mūn pardoned him (al-Ya'qūbī, , Mushākalat al-nās li-zamānihim, ed. Millward, W. G. (Beirut, 1962), p. 28 = id.,Google Scholar ‘The Adaptation of Men to their Time’, tr. Millward, W. G., Journal of the American Oriental Society, LXXXIV (1964), p. 341) and employed him against BābakGoogle Scholar (de Goeje, M.J. (ed.), Kitāb al-'uyūn wa'l-ḥadā'iq (Leiden, 1871), p. 361;Google Scholar Azdī, , Mawṣil, pp. 386.3, 387.10). Hārūn held a governorship in Arabia in 226/84 of (Tab. iii, p. 1319). Contrary to what is sometimes stated, al-Ma'mūn's secretary Aḥmad b. Abī Khālid does not seem to have been a member of this family.Google Scholar

⁶⁸ ‘Abdawayh fought on the side of the Abnā’ in Baghdad in 203 (Tab. iii, p. 1035.7) and reappears as ‘Abdawayh b. Jabala min al-abnā’ in Egypt, where he commanded the shurṭa for Ibn Ṭāhir in 211 and became governor himself in 215, moving on to Barqa with the Afshīn in 216 (Kindī, pp. 183, 189f). ‘Abdawayh al-Abnāwī is said to have been a rebel in North Africa in 178/794f and to have come from Herat (Khalīfa, p. 748; cf. Tab. iii, p. 630 [al-Anbārī]); but the rebel is also said to have been ‘Abdallāh b. al-Jārūd (e.g. YT, ii, p. 496; Ibn 'Idhāri, , Bayān, i, pp. 86ff;Google Scholar cf. Ibn al-Athīr, vi, p. 93, who identifies the two, as does Elad, “Transition”, p. 99, note 47). In Barqa we later find Muḥammad b. 'Abdawayh b. Jabala as governor for al-Wāthiq (YT, ii, p. 586); he also governed Ḥims for al-Mutawakkil (Tab. iii, p. 1421; YT, ii, p. 599). 'Abd al-Raḥmān b. Jabala al-Abnāwā defeated a Khārijite for al-Rashīd in 185–801f (Tab. iii, p. 651) and fell against Ṭāhir as commander of 20,000 or 30,000 Abnā’ (Tab. iii, pp. 826–32; cf. pp. 650, 773, 798, 804; Dīnawarī, p. 394). ‘Alī b. Jabala al-Abnāwī, presumably yet another brother, was min abnā' al-shī'a al-khurāsāniyya min ahi baghdād. He was born in the Ḥarbiyya, but he was one-eyed and eventually lost his good eye as well, so he was a poet rather than a soldier (Abū 'l-Faraj al-Iṣfahāni, Kitāb al-aghānī (Cairo, 1927–1974) (hereafter Aghānī), xx, p. 14).Google Scholar If 'Abdawayh was named after his grandfather, the latter will have been 'Abdawayh al-Jirdāmidh b. 'Abd al-Karīm, who used to drive sheep to Marw before becoming one of the “well known commanders” in the revolutionary army (Tab. ii, p. 1957.4,7) and who settled in Baghdad, where a street and a qaṣr were known after him (YB, p. 241.2; BF, p. 296.3; as a member of the ahlal-dawla he cannot be identical with his Abnāwī namesake, as suggested by Elad, , “Transition”, p. 99,Google Scholar note 47). BF describes him as an Azdī, clearly by walā'.

⁶⁹ Tab. iii, p. 845.

⁷⁰ He first appears under al-Manṣūr, who sent him to Armenia (Ibn A'tham, , Kitāb al-futūḥ (Hyderabad, 1968–1975), viii, pp. 233)f.Google Scholar In 148/765 he was in North Africa, where Hāshim b. Ishtākhanj led a mutiny in North Africa that caused Muhammad b. al-Ash'ath to be expelled and 'Isā b. Mūsā to be elevated to the governorship (YT, ii, p. 464; Khalīfa, p. 680; Ibn ‘Idhārī and al-Nuwayrī in Idris, H. R., “L'Occident musulman à l'avenement des Abbāsides d'après le chroniqueur Zīrī de al-Raqīq”, Revue des Etudes Islamiques 39, 1971, 283f). He fell at Āmid in 160 or 162 against the Khārijite 'Abd al-Salām al-Yashkurī (Khalīfa, p. 701; Tab. iii, p. 492.17). The identification was suggested to me by A. Elad.Google Scholar

⁷¹ al-Sam'ānī, , al-Ansāb (Hyderabad, 1962–1982), i, p. 10.9 (drawn to my attention by C. Robinson).Google Scholar

⁷² Aghānī, xx, p. 188.Google Scholar

⁷³ Ibn Ḥajar, iv, p. 163, s.v. “Sulaymān b Dāwūd b. Rushayd”.

⁷⁴ E.g. 'Alawayh al-A'war min qumvād al-abnā' who appears on campaign against Bābak in 220/835f (Tab. iii, p. 1173); Ṣāliḥ b. ‘Aṭiyya al-Aḍjam, another poet min abnā' al-da'wa (Aghānī, xx, p. 157, 2);Google Scholar Kathīr b. Qādira and Abū ‘l-Fil, who appear along with Dāwūd b. Mūsā b. 'Isā al-Khurāsānī as Banawī soldiers (Tab. iii, p. 845); Shu'ayb b. Ḥarb al-Madā'inī, a scholar who was min abnā' khurāsān, or min afnā' al-nās, his mother being min al-abnā' (TB, ix, p. 239, drawn to my attention by C. Robinson); and Aḥmad b. Abī ḥahir ḥayūr, the author of Kitāb Baghdād, who was min abnā' khurāsān min awlād al-dawla and whose ancestor came from Marwarrūdh (Ibn al-Nadīm, , Fihrist, p. 163;Google Scholar TB, iv, p. 211).

⁷⁵ Arazi and El'ad, “l'Epitre”, ii, pp. 33f; cf. above, note 64, for Mu'ādh.

⁷⁶ AṢlī khurāsān … wa-far'ī baghdād … wa-hiya khurāsān al-'irāq … wa-lanā baghdād bi-asrihā (“Turk”, C, 25.-4; 26.1, 2; 28.7; L, 15.2, 6, 8; 16.13 = 651, 653); cf. also Lassner, , Shaping, p. 134.Google Scholar

⁷⁷ “Turk”, C, 26, 28; L, 15ff = 651, 653; cf also abnā' rijāl al-dawla wa'l-mansūbīn ilā 'l-ā'a (C, 8.-3; L, 4.1 = 636f).

⁷⁸ Above, note 30.

⁷⁹ Tab. iii, p. 1140.

⁸⁰ Ibn Sa'd, vii, p. 350 (s.v. ‘Abd al-Jabbār b. 'Āṣim).

⁸¹ Above, note 28; below, note 117.

⁸² Above, notes 60, 68. The Muslim b. Nasr al-A'war al-Anbārī who appears at Barqa in the time of al-Ma'mūn was presumably also an Abnāwī (YT, ii, p. 542). Those mentioned in Egypt had come from Baghdād (above, note 31).

⁸³ Cf. Arazi and El'ad, “l'Epitre”, i, pp. 29, 50. (Al-Jāḥiẓ does not use this expression himself.)

⁸⁴ Above, note 53.

⁸⁰ Ed. Kawtharī, p. 80; ed. Keller, p. 144 (wa-hum qāmū bi-ḥarb amīr al-mu'minīn).

⁸⁶ Above, note 54.

⁸⁷ Cf. “Reforms”, p. 7. The third assumption is tacit.

⁸⁸ Noted by Ayalon, , “Reforms”, pp. 7ff, 11, 31f;Google Scholar Lassner, , Shaping p. 134.Google Scholar

⁸⁹ YB, p. 248.14; Azdī, Mawṣil, pp. 194f, 201; TB, i, p. 85.15; Yāqūt, , Mu'jam al-buldān, ed. Wūstenfeld, F. (Leipzig, 1866–1873), ii. P. 234.Google Scholar s.v. “Ḥarbiyya”.

⁹⁰ Pseudo-Nāshi’, par. 53, in van Ess, J. (ed.), Frühe mu'tazilitische Häresiographie (Beirut, 1971), pp. 35fGoogle Scholar; al-Nawbakhtī, , Firaq al-sh'a, ed. Ritter, H. (Istanbul, 1931), pp. 29f, 42;Google Scholar al-Ash'arī, , Maqālāt al-islāmiyyīn, ed. Ritter, H. (Istanbul, 1929–1933), pp. 21ff.Google Scholar

⁹¹ YB, p. 248.15–17.

⁹² See the references in Elad, , “Transition”, p. 102,Google Scholar note 59.

⁹³ Tab. iii, p. 829.4. They were also praised as “wearers of bracelets and crowns” (Tab. iii, p. 824.15), but not necessarily with reference to their Iranian ancestry; according to Ayalon, al-Amīn was in the habit of bestowing crowns and bracelets on soldiers who had distinguished themselves (“Reforms”, p. 8n, without reference).

⁹⁴ al-Khwārizmī, Mafāti al'ulūm, ed. van Vloten, G. (Leiden, 1895), p. 119. The passage could be about the Yemeni Abnā', but is conventionally taken to refer to the ‘Abbāsid variety, presumably because it gives the nisba as Banawi, otherwise only attested in al-Jāḥiẓ (cf. above, note 45).Google Scholar

⁹⁵ Still, who are the scholars who supposedly adduce “dubious material where ahl al-Shām, ahl al-Urdunn, ahlFilistin [sic], etc., seem sometimes to denote the Arab forces stationed in these areas” (Zakeri, p. 279)? It is a well-known fact that Syrian troops were stationed in Khurāsān, but even the most ardent supporters of the Arab hypothesis stop short of claiming that the 'Abbāsid revolution was conducted by Syrians!

⁹⁶ Daniel took issue with it almost twenty yean ago: in his view, “No one group, racial or otherwise, dominated the da'wa” (Daniel, E. L., The Political and Social History of Khurasan under Abbasid Rule (Minneapolis and Chicago, 1979), p. 36)Google Scholar. See now also his “The ‘Ahl al-Taqādum’ and the constituency of the Abbasid Revolution in the Merv oasis”, Journal of Islamic Studies, VII (1996),Google Scholar with a helpful survey of the history of the ‘Arabist’/“revisionist” thesis at p. 151n.

⁹⁷ Agha, S. S., “The agents and forces that toppled the Umayyad Caliphate”, PhD, Toronto, 1993, p. 415.Google Scholar I am grateful to Aziz Al-Azmeh, Matthew Gordon and Chase Robinson for drawing my attention to this work.

⁹⁸ I hope to publish this material elsewhere.

⁹⁹ Dīnamwārī, pp. 359f.

¹⁰⁰ YB, pp. 240ff.

¹⁰¹ Cf. Naṣr b. Sayyār's celebrated poems (Dīnawārī, p. 360; Ibn A'tham, viii, pp. 162f; Mas'ūdī, , Murūj, vi, p. 62;Google Scholar ed. Pellat, iv, (par. 2286), ‘Abd al-Ḥamīd b. Yaḥyā's image to the revolutionaries in his Rasā'il, ed. ‘Abbās, I. (Amman, 1988), nos. 8, 38;Google Scholar the alleged order to kill all Arabs/speakers of Arabic in Khurāsān (Tab., ii, pp. 1937, 1974; iii, p. 25; Dīnawārī, p. 358; Azdī, p. 107.3), and the denial of charges of cat worship and the like in AA, p. 282.

¹⁰² Similarly Agha, , “Agents”, p. 374.Google Scholar

¹⁰³ YB, p. 249.3, on the plot for himself and his troops in Baghdād. He was the father of Asad b. al-Marzubān, who also participated in the revolution (AA, pp. 345, 370; Tab., iii, p. 16; BA, iii, p. 139).

¹⁰⁴ Below, notes 141–5; YB, p. 249.2.

¹⁰³ On whose revolt after Abū Muslim's death, see BA, iii, pp. 246f.

¹⁰⁶ Both he and his father had participated in the revolution (Narskhakhī, , Tārīkh-i Bukhārā, ed. Schefer, C. (Paris, 1892), p. 64Google Scholar = id., The History of Bukhara, tr. Frye, R. N. (Cambridge, Mass., 1954), pp. 65f).Google Scholar

¹⁰⁷ Pace Zakeri, p. 280.

¹⁰⁸ EI2, s.v. “Sulaymān b. Kathīr”; AA, p. 219.3 and passim.

¹⁰⁹ Cf. Agha, p. 419.

¹¹⁰ Cf. Sharon, M., Revolt. The Social and Military Aspects of the 'Abbāsid Revolution (Jerusalem, 1990), pp. 112ff.Google Scholar

¹¹¹ Their leader was 'Alī b. Ma'qil al-Ḥanafi, credited with a following of 10,000 or 25,000 men (AA, pp. 295, 309f; Tab., ii, p. 2002.6).

¹¹² Thus Sunbādh, Muqanna' (above, notes 105–6) and Isḥāq the Turk, cf. Sadighi, G. H., Les mouvement religieux iraniens (Paris, 1938).Google Scholar

¹¹³ Cf. van Ess, J., Theologie und Gesellschaft, iii (Berlin and New York, 1992), pp. 10ff.Google Scholar

¹¹⁴ Ibn Tayfūr, ed. Kawtharī, p. 80; ed. Keller, p. 143. Ayalon takes muwallad to mean “of mixed descent” (“Reforms”, p. 6; followed by Crone, Slaves, p. 66 and note 472 thereto). But the basic meaning of muwallad is “homeborn slave” (Lane, E. W., An Arabic-English Lexicon (London, 1863–1893), s.v.).Google Scholar The Hispanic Muslims were known as muwallads because they were non-Arabs born in Muslim society; and the Abnā' were clearly muwallads in the sense of Khurāsānīs born in Iraq (cf. Lassner, , Shaping, p. 134).Google Scholar

¹¹⁵ Elad, , “Transition”, pp. 101ff;Google Scholar cf. Arazi and El'ad, “L'Epitre”, i, p. 70, where the Ṭāhirids are declared to be Arabs, although of indigenous (Iranian) origin; Elad, , “Transition”, pp. 91f, 119,Google Scholar where the 'Abbāsid period as the apogee of Arabism (in Goitein's words) is explained as the superiority of Arab culture largely sustained by non-Arabs and where the Arab culture of the non-Arabs among the Abnā’ is stressed.

¹¹⁶ Cf. the following note.

¹¹⁷ Tab. iii, p. 1278.17; cf. pp. 1273.16, 1274.16, where the Arabs and Abn' are listed separately, and p. 1278.5, where the Abnā' are distinguished from the local population as abnā' al-quwwād.

¹¹⁸ The same goes for the troops that Elad sees as “comprised exclusively of Arab tribes”, cf. the fact that his first reference is to an army of 3,000 Basran Arabs and mawālī (“Transition”, p. 107).

¹¹⁹ Cf. Tab. iii, p. 922.18, on the killers of al-Amīn (qawm min al-'ajam); Kindī, Governors, p. 184.1, on ‘Abdallāh b. Ṭāhir's troops in Egypt (they were quwwād al-'ajam min ahl khurāsān); Tab. iii, p. 1142.7; Ibn Ṭayfūr, ed. Kawtharī, p. 144; ed. Keller, p. 266 (a Syrian held al-Ma'mūn to favour the 'ajam of Khurāsān); Azdī, Mawṣil, p. 334.11 (Naṣr b. Shabath was of the same opinion). Al-Ma'mūn entered Baghdad with ‘ajam carrying bows and arrows (Ibn Ṭayfur, ed. Kawtharī, p. 16; ed. Keller, p. 15). When al-Ma'mūn praised the ‘ajam of Khurāsān, a commander in his presence responded that nobody was braver than abnā' khurāsān al-muwalladīn, i.e. the homeborn Khurāsānīs (cf. above, note 114). Ḥumayd b. 'Abd al-Ḥamid also distinguished the Abnā' from the 'ajam (here possibly non-Muslims) of Khurāsān, cf. below, note 147.

¹²⁰ Tab. iii, p. 829.4; cf. Elad, , “Transition”, p. 105,Google Scholar note 68; Hoffmann, , “Pobel”, p. 30,Google Scholar who both infer that the Abnā' saw themselves as Arabs. Note also that Ibn Ṭayfūr wrote a book on faḍl al-'arab 'alā 'l-'ajam (Ibn al- Nadīm, Fihrist, pp. 163f; above, note 74).

¹²¹ YT, ii, pp. 532f, 547.

¹²² Crone, Slaves, note 604.

¹²³ Tab. iii, p. 78; cited in Elad, , “Transition”, p. 103, note 61.Google Scholar

¹²⁴ Tab. iii, p. 80; cited in Elad, , “Transition”, p. 103, note 61.Google Scholar

¹²³ Elad, , “Transition”, p. 102, note 59.Google Scholar

¹²⁶ Azdī, , Mawṣil, p. 286;Google Scholar Tab. iii, p. 555; cited in Elad, , “Transition”, p. 103,Google Scholar notes 58, 62.

¹²⁷ The non-Khurāsānī “Yamāniyya” in al-Manṣur's ṣaḥāba successfully intervened on behalf of a Syrian Kindī/Yemeni who had rebelled with 'Abdallāh b. ‘Alī (Azdī, Mawṣil, pp. 178, 2320. ‘Uqba b. Salm al-Hunāā'ī/Azd/Yemen killed numerous Rabī'a as governor of Baḥrayn and Yamāma to make up for the Yemenis that Ma'n b. Zā'ida al-Shaybānī/Rabī'a had killed as governor of Yemen (YT, ii, p. 463; cf. above, note 52). When Hārūn came to Mosul, the Mosulis made sure to include Anṣārī scholars in their delegation to him, for Abū Yūsuf was with him and he was an Anṣārī responsive to fellow-tribesmen (Azdī, , Mawṣil, pp. 284f).Google Scholar

¹²⁸ Cf. Tab. iii, p. 631.

¹²⁹ Cf. Crone, , Slaves, p. 244,Google Scholar note 432; Elad, , “Transition”, p. 108Google Scholar and note 90 thereto. There was plenty of it in Mosul too, and the Azdī/Yemeni author of Ta'rīkh al-Mawṣil credits al-Manṣur himself with a declaration of pro-Yemeni sentiment so violent that he could hardly have retained non-Yemenis in his service if he had actually made it (Azdī, , Mawṣil, pp. 219ff).Google Scholar

¹³⁰ Elad, , “Transition”, p. 105.Google Scholar

¹³¹ The examples relate to ahl al-dawla rather than Abnā'. Al-Ḥasan b. Qaḥṭaba al-Ṭā'ī/Yemen told the future al-Manṣūr that he could not kill Ibn Hubayra al-Fazārī/Muḍar without stirring up tribal enmity that would engulf al-Manṣūr's own 'askar, hence Khāzim b. Khuzayma al-Tamīmī/Muḍar was sent to do the job (YT, ii, p. 424). Khāzim b. Khuzayma al-Tamīmī's killing of Ḥārithīs/Yemenis in 134 did have repercussions in the army (Tab. iii, p. 76f; cf. Elad, , “Transition”, p. 101,Google Scholar note 57). Abū Dāwūd was accused of 'aṣabiyya, explained as a preference for Arabs (in general) and his own people (in particular) over other ahl al-da'wa, in Khurāsān in 135 (Tab. iii, p. 83). But in the story of how Qutham b. al-'Abbās stirred up tribal strife in al-Manṣūr's army, the Khurāsānīs remain distinct from Muḍar, Rabi'a and Yemen (Tab. iii, pp. 266f; Elad, , “Transition”, p. 102,Google Scholar note 57).

¹³² Elad, , “Transition”, pp. 125ff.Google Scholar

¹³³ Everyone who could not pass for an Arab would present himself as a non-Arab of the most elevated kind, cf. the royal descent claimed by Bashshār b. Burd (Aghānī, iii, p. 135; cf. Zakeri, pp. 3031) and by local rulers in Iran (Bosworth, C. E.,“ The heritage of rulership in early Islamic Iran and the search for dynastic connections with the past”, Iran, XI, 1973). DifFerendy Zakeri, p. 289.Google Scholar

¹³⁴ Cf. the cattle-driver ‘Abdawayh (above, note 68) or Khālid b. Barmak, who owed his presence in Muslim society to enslavement and whose previous status was in any case not that of an asbār, his father being the leader of a Buddhist monastery (cf. Bosworth, C. E., “Abū Ḥafṣ 'Umar al-Kirmānī and the rise of the Barmakids”, Bulletin of the School of Oriental and African Studies, LVII (1994), esp. pp. 269n, 270–2).Google Scholar

¹³⁵ “Turk”, C, 28 (with naḥnu for taḥta); L, 16 = 653; Zakeri, 277. Zakeri also claims that the 'Abbāsids honoured their troops with the tide al-aḥrār, but his examples refer to tenth-century soldiers who were described as aḥrār in the literal sense of free as opposed to ghilmān (p. 276 and note 858).

¹³⁶ Zakeri, pp. 52fF.

¹³⁷ Tab. ii, p. 1550.15; 1606.6 (years 112 and 119); cf. Encyclopaedia Iranica, s.v. “Asāwera”; Zakeri, pp. 112ff.

¹³⁸ Cf. Tab. ii, p. 1243.1, on the Sogdians in 93.

¹³⁹ Cf. Narshakhī, pp. 8, 57 = pp. 10, 59.

¹⁴⁰ Tab. iii, p. 356.8; cf. p. 93.7. For Turār, also known as Utrār and Uṭrār, see al-Muqaddasī, Aḥsan al-taqāsim², ed. de Goeje, M.J. (Leiden, 1906), p. 263.1;Google Scholar Yāqūt, , Mu'jam al-buldān, ed. Wüstenfeld, F. (Leipzig, 1866–1873), i. p. 310.Google Scholar

¹⁴¹ BA, iii, p. 249.7; Yāqūt, , Buldān, i, p. 90.Google Scholar

¹⁴² TB xi p. 438.9. Compare Justi, F., Iranisches Namenbuch (Marburg, 1895), p. 171 (Xarŝah(?)).Google Scholar

¹⁴³ Tab. iii, p. 492.19; cf. ii, p. i960; BA, iii,, pp. 139f (S. b. Rāh), 205–7; cf. Justi, p. 338.

¹⁴⁴ YB, p. 293.18; cf. p. 248.17; Barthold, W., Turkestan down to the Mongol Invasion (London, 1968), p. 95; cf. Tab. ii, p. 1598.2.Google Scholar

¹⁴³ Tab. iii, p. 122; cf. note d and the Addenda et Emendanda thereto (wa-ma'a Jahwar nukhab fursān al-'ajam [minhum?] Zuwāreh wa'l-Ishtākhanfj); BA, iii, pp. 247f (Zubārah al-Bukhārī); YB, p. 245.4 (Rubāwah al-Kirmānī); Justi, pp. 337, 388 (Zuwāreh, from Uzwārak). For Hāshim b. al-Ishtākhanj, see note 70, and Jāḥiẓ, “Turk”, C, 19.6; L, 11.1 = 645, where the Khurāsānīs boast of him (not the Abnā', as Zakeri, p. 296, would have it).

¹⁴⁶ “Turk”, C, 26ff;L, 15f = 651f.

¹⁴⁷ “Turk”, C, 52ff; L, 33f = 671f.

¹⁴⁸ “Turk”, C, 27 ult.; L, 16.7 = 652.

¹⁴⁹ “Turk”, C, 62.4; L, 39.9 = 677.

¹⁵⁰ “Turk”, C, 27f; L, 16 = 652.

¹⁵¹ Ed. Kawtharī, p. 80; ed. Keller, p. 144.

¹⁵² “Turk”, C, 26f; L, 15 = 652; Ibn Ṭayfūr, ed. Kawtharī, p. 80; ed. Keller, p. 144.

¹⁵³ Zakeri, pp. 281, 288; cf. “Turk”, C, 20; L, 11 = 646.

¹⁵⁴ “Reforms” p. 32.

¹⁵⁵ Shaping, p. 135; accepted by Hoffmann, , “Pöbel”, p. 36.Google Scholar

¹⁵⁶ Zakeri, p. 288.

¹⁵⁷ Tab. iii, p. 872 (Fishbein's translation).

¹⁵⁸ Zakeri, p. 287; cf. also pp. 28 jf, where the ‘ayyārim's costume, or rather lack thereof, is adduced in favour of their Iranian descent. (Generally speaking, every non-Arab is an Iranian asbār in this book.)

¹⁵⁹ Cf. Hoffmann, , “Pöbel”, pp. 35ff.Google Scholar

¹⁶⁰ Tab. iii, p. 817, where the troops number 50,000 and are described as ahl baghdād; cf. p. 824.15, where 'Alī b. [Īsā addresses them as ma'shar al-abnā'. He had 40,000 men according to YT, ii, p. 530; Azdī, Mawṣil, p. 323.7, and 40,000 horsemen according to Ibn Kathīr, , al-Bidāya wa'-l-nihāya (Cairo, 1351–8), x, p. 226.15.Google Scholar Cf. above, note 65, for the family.

¹⁶¹ Tab. iii, 827.1, 3; 832.9; Dīnawarī, p. 394 (for the figure 30,000). Cf. above, note 68, for the family.

¹⁶² Tab. iii, 843.14, cf. p. 1007.16.

¹⁶³ Tab. iii, pp. 1007f. Cf. below, note 67, on the family.

¹⁶⁴ Tab. iii, p 831.1,14.

¹⁶⁵ The maḍāyiq could of course refer to cramped urban conditions, but the word means “defiles” in Azdī, Mawṣil, p. 389.-2 (Azerbayjan), and the khanādiq must be trenches dug around camps.

¹⁶⁶ “Turk”, C, 27.6; L, 16.4 = 652.

¹⁶⁷ It was opposed to the Sa'diyya and is first mentioned in 255/868f in connection with the pre-history of the Zanj revolt (Tab. iii, pp. 1745, 1747 and passim.).

¹⁶⁸ Practically the same phrase recurs in the story of Khālid b. Yazīd, the mawlā of the Muhallabids, in al-Jāḥiẓ, al-Bukhalā', ed. al-Ḥājirī, Ṭ. (Cairo, 1958), p. 50:Google Scholar “Ask the Katifiyya, Khulaydiyya, Khurrabiyya, Bilāliyya … about me”. But it is no more informative. For the name Khulaydiyya, see Bosworth, C. E., The Mediaeval Islamic Underworld (Leiden, 1976), i, pp. 43n, 64n (drawn to my attention by A. Elad).Google Scholar

¹⁶⁹ Cf. his reference to the Bilāliyya, who may not have existed in al-Ma'mūn's time, but who certainly did in al Jāḥiẓ's days (above, notes 167–8).

¹⁷⁰ Cf. Bachrach, B. S. in Parker, G. (ed.), Cambridge Illustrated History of Warfare (Cambridge, 1995), p. 90.Google Scholar

¹⁷¹ Tab. iii, pp. 40, 42 (where the shield wall is stationary); Azdī, MawṢil, pp. 128f (where it advances towards the enemy). Compare Hollister, C. Warren, Anglo-Saxon Military Institutions (Oxford, 1962), pp. 132f.Google Scholar