1. P.F. de Moraes Farias has recalled preliminary discussion in Dakar, well over a decade ago in his “Great States Revisited,” JAR, 15(1974), 483n.Google Scholar This acute and stimulating review article first drew attention to several points which are developed below. About the same time Coulibaly, M. published “L'attaque de Ghana (XIe siècle),” Afrika Zamani [Yaounde], 2 (April 1974), 55–57.Google Scholar This article, too little known, denies (as we do) that any Almoravid conquest of Ghana ever occurred. Our thanks go to Dr. Farias, and to John Hunwick, Harry T. Norris, Lamin Sanneh, and Thomas Whitcomb for much help and advice. Hunwick's article, “Gao and the Almoravids: A Hypothesis” in West African Culture Dynamics, ed. Swartz, B. and Damett, R. (Hague, 1980), 413–30CrossRefGoogle Scholar, has been of special value to us.

2. Cuoq, J.M., ed. and tr., Recueil des sources arabes concernant I'Afrique occidentale du VIIIe au XVIe siècle (Bilād al-Sūdān) (Paris, 1975)Google Scholar; and Hopkins, J.F.P. and Levtzion, N., ed. and tr., Corpus of Early Arabic Sources for West African History (Cambridge, 1981).Google Scholar Fr. Cuoq has immeasurably lightened by his pioneering work the labor of anyone working on the early Arabic sources for western Africa, but it seems likely that the Corpus will in the long run prove more thorough and reliable.

3. This is the date al-Bakrī gave; see Cuoq, , Recueil, 92.Google Scholar This date, or one year earlier, is confirmed by other sources; ibid, 126, 223, 237.

4. The dates of Abū Bakr's career have been the subject of considerable discussion; see de Moraes Farias, P.F, “The Almoravids: Some Questions Concerning the Character of the Movement During Its Period of Closest Contact With the Western Sudan,” BIFAN, 29B(1967), esp. 848–49.Google Scholar The date of the conquest (1076) is taken from al-Zuhrī, a dubious passage (see below). The date of Abū Bakr's death, on the other hand, seems firmer, for it agrees with numismatic evidence. See Hazard, H.W., The Numismatic History of Late Medieval North Africa (New York, 1952), 60–62.Google Scholar

5. Trimingham, J.S., A History of Islam in West Africa (London, 1962), 29–30.Google Scholar It is interesting that Ghana, sacked and fallen, is still presented in the above quotation as ruling over many people. The passage continues with a further elaboration of the repercussions of the conquest. Many other authors follow roughly the same line; see, for example, Fage, J.D., An Introduction to the History of West Africa (Cambridge, 1962), 22Google Scholar, and Awe, B., “Empires of the Western Sudan: Ghana, Mali, Songhay” in Ajayi, J.F. Ade and Espie, I., A Thousand Years of West African History (Ibadan, 1965), 61.Google Scholar In the hands of enthusiastic amateurs the story of the conquest becomes more and more vivid, circumstantial, distorted, and inaccurate; see, for example, Rosenthal, R., The Splendor That was Africa (New York, 1967), 18–19.Google Scholar

6. Cornevin, Robert, Histoire des peuples de l'Afrique noire (Paris, 1960), 238.Google Scholar

7. Ibid, 246. 1088 is presumably Abū Bakr's death but, apart from the fact that he was killed while fighting the sūdān, there is no hint anywhere in the Arabic sources, so far as we know, that this had anything to do with Ghana and its independence.

8. Ibid, 246.

9. Cuoq, , Recueil, 42–80.Google Scholar Of 21 pages with pre-al-Bakrī references to Ghana, the index misses eight.

10. Passing mention should be made of al-Bakrī's contemporary, Ṣāʿid b. Aḥmad Ṣāʿid, who died in 462/1070. He mentioned Ghana as a people and without elaboration; Cuoq, , Recueil, 109–10.Google Scholar Two editions which Cuoq does not mention are those of Blachère, R., Kitāb ṭabāḳat al-umam (Livre des catégories de nations) (Paris, 1935)Google Scholar in French, and an Arabic version published at Najaf in 1967: a Sūdān reference which Cuoq omits occurs on pp. 44 and 14 respectively; there is a very slight zanj reference on pp. 95/60 (also on pp. 44/14).

11. Cuoq, , Recueil, 115ff.Google Scholar, gives the relevant passages from al-Zuhrī. He relies on the Arabic text prepared by Hadj-Sadok, M., “Kitāb al-djaʿrāfiyya: mappemonde du calife al-Ma'mūn, reproduite par Fazārī (IIIe/IXe s.) rééditée et commentée par Zuhri (VIe/XIIe s.),” Bulletin d'Etudes Orientales, 21(1968).Google Scholar We have consulted this Arabic, and base our own translation directly on it. There are at least nine known manuscripts of al-Zuhrī's Kitāb; the Paris MS., from which Hadj-Sadok works, though the oldest, it the most corrupt, and somewhat condensed; see Hopkins, and Levtzion, , Corpus, 93–94.Google Scholar

12. Variant readings among the manuscripts give also k.nāwah and al-Habashah. Janawah and the European Guinea both derive from the Berber word for blacks; the sense is thus exactly that of the Arabic word sūdān; see Cuoq, Recueil, 119n1. Al-Zuhrī did not include a definite article with these names, such as Janāwah, Lamtūnah, etc.

13. One manuscript reads Misr--Egypt or Cairo--for the Maghrib.

14. The manuscript text reads amīr Massūqah, or in one case amīr al-mu 'minīn.

15. One manuscript reads: “and they waged jihād, and raided, and spent in the way of God, and directed themselves towards Mecca.” It is not clear from Hadj-Sadok, “Kitab,” para. 336, whether this is an addition to the text of the other manuscripts or a replacement for something in them.

16. This particular passage is in Cuoq, , Recueil, 119Google Scholar, and is para. 336 in Hadj-Sadok, , ”Kitab,” 182.Google Scholar

17. Cuoq, Recueil, 119n3; so do Hopkins, and Levtzion, , Corpus, 98.Google Scholar

18. Compare note 4, above, particularly the reference to Farias, , “Almoravids,” 848–49.Google Scholar

19. Cuoq, , Recueil, 119n4.Google Scholar A late fourteenth-century (A.D.) author quoting al-Zuhrī gives the name as Abū Yahyā b. Abī Bakr, thus complicating the matter still further.

20. The Ghana passage and Yahia Ténéméra are in Arnaud, R., “La singulière légende des Soninkés: traditions orales sur le royaume de Koumbi et surs divers autres royaumes soudanais,” Bulletin du Comité de l'Afrique française (1912), 157–58.Google Scholar The French here gives “Maures,” adding this footnote: “Souraké, en soninké, pluriel Sourakou.” For Yahia son of Yennoumas see Marty, P., Etudes sur l'Islam et les tribus du Soudan, 3: Les tribus maures du Sahel et du Hodh (Paris, 1921), 329, 333.Google Scholar Further details are given by Norris, H.T., Saharan Myth and Saga (London, 1972).Google Scholar

21. Norris, H.T., “New Evidence on the Life of ʿAbdullāh b. Yāsīn and the Origins of the Almoravid Movement,” JAH, 12(1971), 255CrossRefGoogle Scholar, quoting from ʿIyād's, Tartīb al-madārik wa taqrīb al-masālik li-maʿrifat aʿlām madhhab Mālik, ed. Maḥmūd, Aḥmad Bakīr (Beirut 1967), 2:780.Google Scholar The passage is translated in Cuoq, , Recueil, 125Google Scholar, but this particular Arabic usage does not emerge clearly there; khurūj seems to be rendered “expansion.” Hopkins, and Levtzion, , Corpus, 101Google Scholar, read “at their first appearance.”

22. Ibid, 98.

23. al-Khaṭīb, Lisān al-dīn ibn. Iḥāṭah fī akhbār Gharnāṭah (Cairo 1973), 1:404–09Google Scholar; for this possibly vital reference, we are indebted to H.T. Norris.

24. Cuoq, , Recueil, 120Google Scholar; Hopkins, and Levtzion, , Corpus, 98Google Scholar; Hadj-Sadok, , “Kitab,” 182Google Scholar, para. 337. The contrast between raiding here and respect in the next passage is curious.

25. Cuoq, , Recueil, 120Google Scholar, reads here “tout amīr de Ghāna des Djināwa.” We do not know if this is a slip or a deliberate correction of the Arabic text as published by Hadj-Sadok. Hopkins, and Levtzion, , Corpus, 99Google Scholar, read “every amīr of the land of Janāwa.”

26. Cuoq, , Recueil, 120Google Scholar; Hadj-Sadok, , “Kitāb,” 181Google Scholar, para. 338.

27. Mauny, R.A., “The Question of Ghana,” Africa, 24(1954), 204–05CrossRefGoogle Scholar; see also his Les siècles obscurs de l'Afrique noire, (1970), 146.Google Scholar Presumably, he draws on al-Idrīsī.

28. Cuoq, , Recueil, 120Google Scholar; para. 338 in Hadj-Sadok, “Kitāb.” Hopkins, and Levtzion, , Corpus, 98–99Google Scholar, read “NSLĀ” for “Silla.” Either the geography or the nomenclature in al-Zuhrī seems askew, since it is generally agreed that Silla well west of Ghana, and Tādimakka far east of Ghana.

29. Lewicki, T., “Un état soudanais médiéval inconnu: le royaume de Zāfūn(u),” Cahiers d'études africaines, 11(1971), 518CrossRefGoogle Scholar, citing this para, of al-Zuhrī.

30. Cuoq, , Recueil, 120n1.Google Scholar

31. Farias, , “Great States,” 480–81.Google Scholar

32. Cuoq, , Recueil, 96, 107.Google ScholarHunwick, , “Gao and the Almoravids,” 427–28Google Scholar, discusses Tādimakka in detail; struggling, like Lewicki and Cuoq, to insert al-Zuhrī into the picture of Almoravid dominance and subordinate Ghana, he comments that “it is more probable that Ghana supported the Sanhaja than vice versa.” But if we accept this inversion of al-Zuhrī we are left with the puzzling question: why should al-Zuhrī have falsified the historical record to enhance the prestige of black Ghana, and to play down that of the white Almoravids?

33. This passage is not in Cuoq, Recueil, The full Arabic text is given in Abdel-Hamid, , Kitāb, 217Google Scholar (in the Arabic). The full French translation, in Fagnan, , L'Afrique, 195Google Scholar, says that the people of Silla, now Muslim, “abjurèrent autrefois entre les mains” of the ruler of Takrur, which is the direct opposite of al-Bakrī and of the Arabic text given by Abdel-Hamid. For the full references of these editions of the Kitāb al-istibṣār see below, note 51.

34. Or, according to a variant reading, about twenty days.

35. Cuoq, , Recueil, 121n3Google Scholar, suggests that we should rather read Zāfūn(u), which is a possible variant for al-Zuhrī and the same name which is found in al-Bakrā and Yāqūt. On p. 185n2, following Lewicki, Cuoq, Recueil, suggests that Qarāfūn or Zāfūn(u) should be identified with Diafounou, a region on the upper Kolimbiné, in the cercle of Nioro; see Lewicki, , “Etat Soudanais,” 511–12.Google Scholar Hopkins and Levtzion prefer Zāfūn, but add that Diafounou is west, not east, of Ghana, , Corpus, 99, 389n24.Google Scholar

36. This passage is presumably the origin for Delafosse's reference to an Arabic author, according to whom “l'empereur de Ghana faisait avec succès la guerre aux Almoravids campés au Nord-Est de sa capitale, entre celle-ci et Rayoun ou Araouân, qui était 'la ville du désert la plus proche de Sidjilmassa et de Ouargla';” Delafosse, Maurice, Haut-Sénégal-Niger (3 vols.: Paris, 1912) 2:41.Google Scholar Delafosse said that the author, whom he describes as anonymous, wrote the ‘Kitâbou-l-jarafiya’, and was cited by Cooley. The only mention we have found of such a work in Cooley does not give any of this information; Cooley, W.D., The Negroland of the Arabs (London, 1841), 19, 19n.Google Scholar

37. Or, according to a variant reading, outside the rite of the people of the Sunnah.

38. Cuoq, , Recueil, 121–22Google Scholar; para. 340 in Hadj-Sadok, “Kitāb.” The final Arabic is:

Hopkins, and Levtzion, , Corpus, 99Google Scholar, read: “They are attached to the town of Ghāna because it is their capital and the seat of their kingdom.”

39. Ibid, 98-100.

40. Cuoq, , Recueil, 126Google Scholar; Hopkins, and Levtzion, , Corpus, 102Google Scholar; Norris, , “New Evidence,” 256Google Scholar; see also note 21 above.

41. Cuoq, , Recueil, 133.Google Scholar There is a passing reference to the Almoravids in the context of a route from Cairo across North Africa, which appears in the early edition by A. Dozy and J. de Goeje (Arabic pp. 162-63, French pp. 193-94), and in the still unfinished Arabic edition which has been appearing in Rome since 1970 (344-45).

42. Cuoq, , Recueil, 133n1.Google Scholar

43. Cuoq, , Recueil, 137n2Google Scholar; Hopkins, and Levtzion, Corpus, 112Google Scholar, read: “All the lands we have described are subject to the ruler of Ghāna, to whom the people pay their taxes, and he is their protector.”

44. Cuoq, , Recueil, 135–36Google Scholar; Hopkins, and Levtzion, , Corpus, 111Google Scholar, prefer Tīraqqā.

45. Cuoq, , Recueil, 133.Google Scholar

46. Corpus, 105, 390n8.

47. Cuoq, , Recueil, 167Google Scholar; Hopkins, and Levtzion, , Corpus, 89.Google Scholar

48. Cuoq, , Recueil, 169.Google Scholar The phrasing of the passage about the kings is a little odd:

Cuoq renders this: “Parmi leurs rois on dit que se seraient convertis à l'Islam (ceux de) cinq tribus, dont la plus proche est Ghāna.” Hopkins, and Levtzion, speak of “kings of five of their tribes” (Corpus, 132).Google Scholar But it could possibly be read, more literally: “and among their kings there have converted to Islam, so it is said, five tribes, the nearest of them Ghana.” In this case, the meaning might be either “the kings of five tribes,” or “five tribes of kings,” that is, presumably, five royal tribes, or perhaps better, clans. Journal asiatique, 207(1925), 41.Google Scholar

49. Cuoq, , Recueil, 171Google Scholar, citing Hess, R.L., “The Itinerary of Benjamin of Tudela,” JAR, 6(1965), 15–24.Google Scholar Benjamin spoke of Zawila “in the land of Ghana” (17); Hess points out that this is an error (22), but it may once again be a distant reflection of Ghana's continuing pre-eminence in the Sudan countries. See also Hopkins, and Levtzion, , Corpus, 135, 393n2.Google Scholar

50. Cuoq, , Recueil, 171–75Google Scholar; Hopkins, and Levtzion, , Corpus, 91.Google Scholar

51. We have consulted the Arabic text, published by the University of Alexandria in 1958, and prepared by Saad Zaghloul Abdel-Hamid, Kitāb al-istibṣār fi aʿjā'ib al-amṣār: description de la Mekke et de Médine, de l'Egypte et de l'Afrique septentrionale par un éarivain maracain du Vie siècle de l'Hégire (XIIe s. J.C.): texte arabé annoté, publié avec une traduction de la partie relative aux Lieux Saints et à l'Egypte. The portion untranslated by Abdel-Hamid is available as L'Afrique septentrionale au XIIe siècle de notre ère: description extraite du Kitab el-istibçar et traduite par E. Fagnan, Recueil des notices et mémoires de la Société archéologique du Département de Constantine. 4th ser., 2 (Constantine, 1899). The African portions, considerably abbreviated, appear in Arabic and French in Kamal's, YoussoufMonumenta Cartographica Africae et Aegypti, vol. 3, fasc. 4, 908–15, published in 1934Google Scholar; Kamal in turn worked mainly from manuscript 2225 in the Bibliothèque nationale and from a published version which we have not seen, but which he defines as “Description de l'Afrique par un géographe arabe anonyme, texte arabe publié pour la première fois par Mr. Alfred de Kremer, Vienne 1852.” See also Hopkins, and Levtzion, , Corpus, 137ff.Google Scholar

52. Abdel-Hamid, , Kitāb, 209 (in the Arabic)Google Scholar; Fagnan, , L'Afrique, 180–81Google Scholar; Kamal, , Manumenta, 914r-vGoogle Scholar; Cuoq, , Recueil, 176.Google Scholar For “chief,” Abdel-Hamid gives zaʿīm, Kamal aẓīm.

53. Corpus, 141; the brackets are their own.

54. Abdel-Hamid, , Kitāb, 219–20 (in the Arabic)Google Scholar; Fagnan, , L'Afrique, 199Google Scholar; Kamal, , Monumenta, 915rGoogle Scholar; Cuoq, , Recueil, 177Google Scholar; Hopkins, and Levtzion, , Corpus, 146.Google Scholar

55. The formerly accepted chronology put the foundation of Marrakesh at about 1062; see, for example, Terrasse, Henri, Histoire du Maroc (2 vols.: Casablanca 1949), 1:222.Google Scholar The revision is argued by Farias, , “Almoravids,” 848–49.Google Scholar See also Hopkins and Levtzion, Corpus, 395n32.

56. This letter quoted in the Kitāb al-istibṣār may raise certain difficulties for the revised chronology, according to which Yūsuf was never amīr of Aghmāt.

57. Abdel-Hamid, , Kitāb, 111 (in the Arabic)Google Scholar; Fagnan, , L'Afrique, 5Google Scholar; Cuoq, , Recueil, 175Google Scholar; Hopkins, and Levtzion, , Corpus, 138.Google Scholar

58. They may most conveniently be pursued through the excellent indices in Fagnan, L'Afrique, and Abdel-Hamid, Kitāb; both Kamal, Monumenta and Cuoq, Recueil, are incomplete for this purpose.

59. Cuoq, , Recueil, 179.Google Scholar The Arabic text appears in Analeotes sur l'histoire et la littérature des Arabes d'Espagne, par al-Makkari, ed. Dozy, R., et al (2 vols.: Leiden 1858–1861), 2:72Google Scholar, line 24. See also Hopkins, and Levtzion, , Corpus, 372.Google Scholar

60. Corpus, 433n9.

61. Delafosse, , Haut-Sénégal-Niger, 2:165–66.Google Scholar

62. Cuoq, , Recueil, 181.Google Scholar

63. We take our translation of this passage from Hopkins, and Levtzion, , Corpus, 170.Google Scholar The Arabic is to be found on 2:908 of Yāqūt b. al-Ḥamawī, ʿAbd Allāh, Muʿjam al-buldān, ed. Wüstenfeld, F. (6 vols.: Leipzig, 1866–1873).Google Scholar The opening Arabic is a little curious:

This seems to indicate that it was the land of the sūdān which was near the Maghrib, while the “vast province” adjoined the land of the veiled people. Cuoq, , Recueil, 185Google Scholar, agrees precisely with Hopkins and Levtzion here. The veiled people are the mulaththamūn or Almoravids.

64. The entry is headed Zāfūnu, and appears in its alphabetical place in each of the various editions of the muʿjam. There is another reference under the heading Ghadāmisu, a town of the Maghrib, to the south of which one comes to the land of the sūdān after Zāfūnu. This seems to place Zāfūnu in the Sahara between Ghadames and the Sudan countries, a location somewhat confirmed by the description in the principal passage of the nomadic way of life in Zāfūnu. This lesser reference is in Hopkins, and Levtzion, , Corpus, 172Google Scholar, from Wüstenfeld, , Muʿjam, 3:776Google Scholar, but is not in Cuoq.

65. Hopkins, and Levtzion, , Corpus, 145Google Scholar; 398n9; Abdel-Hamid, , Kitāb, 218.Google Scholar This passage is not in Cuoq, Recueil.

66. Lewicki, , “Etat soudanais,” 506.Google Scholar

67. Farias, , “Great States,” 480.Google Scholar

68. Cuoq, , Recueil, 187Google Scholar; Hopkins, and Levtzion, , Corpus, 173.Google Scholar They add (399n21) that Ganāwa is the Berber name for the Sudan; Wüstenfeld, , Muʿjam, 4:307.Google Scholar

69. Cuoq, , Recueil, 188.Google Scholar We have been unable to check Cuoq's references here; Hopkins, and Levtzion, , Corpus, 152–53.Google Scholar

70. Cuoq, , Recueil, 192Google Scholar; Hopkins, and Levtzion, , Corpus, 159–60.Google Scholar We have consulted both editions of the Kitāb al-kāmil fi'l-ta'rīkh which Cuoq mentions, the Arabic edited by Tornberg, C.J. (Ibn-el-Athiri: Chronicon quod perfectissimum inscribitur, [14 vols.: Leiden, 1867–1876)Google Scholar, and the French translation of the North African material by E. Fagnan in his Annales du Maghreb et de l'Espagne, using, however, the 1898 edition. The Tornberg passages which Cuoq attributes to 8:74-75 are in fact 9:425-28.

71 Cuoq, , Recueil, 191Google Scholar; Hopkins, and Levtzion, , Corpus, 158–59Google Scholar; Tornberg, , Chronicon, 9:425Google Scholar; Fagnan, , Annales, 462.Google Scholar Ibn al-Athīr named the pilgrim, and Cuoq suggests that this must be a mistake for Yaḥyā b. Ibrahīm. However, since ʿIyāḍ al-Sabtī (see p. above), who was obviously in possession of full information, also spoke of a pilgrim named Jawhar, Cuoq's suggested mistake seems improbable. Norris, , “New Evidence,” 259–60Google Scholar, discusses the problem of the identity of Jawhar the pilgrim in some detail.

72. Cuoq, , Recueil, 193Google Scholar; Hopkins, and Levtzion, , Corpus, 160Google Scholar; Tornberg, , Chronicon, 9:427Google Scholar; Fagnan, , Annales, 465.Google Scholar

73. Cuoq, , Recueil, 194Google Scholar; Hopkins, and Levtzion, , Corpus, 161Google Scholar; Tornberg, , Chronicon, 9:428Google Scholar; Fagnan, , Annales, 466.Google Scholar The date is too early; see above, p. and n.

74. Tornberg, , Chronicon, 14:774.Google Scholar

75. Cuoq, , Recueil, 196Google Scholar, but we have taken our translation from Hopkins, and Levtzion, , Corpus, 368–69.Google Scholar Al-Darjīnī's account survives only as quoted by al-Shammākhī, who died in 928/1522.

76. Cuoq, , Recueil, 196–97Google Scholar, gives only a brief selection of the material; Hopkins, and Levtzion, , Corpus, 164–66Google Scholar, are very much more complete. We have also consulted the Beirut edition (1968-72) of the Wafayāt al-aʿyān wa-anbā' abnā' al-zamān, in eight volumes, the eighth being entirely indices; and the English edition by de Slane, MacGuckin, Ibn Khallikān's biographical diotionary (4 vols.: Paris, 1842–1871).Google Scholar

77. Cuoq, , Recueil, 201.Google Scholar He uses the Arabic edition of al-Qazwīnī's Athār al-bilād wa adhbār al-ʿibād published by F. Wüstenfeld in 1848 under the title El-Cazwini's Kosmographie. An index to this volume has been prepared by Kowalska, Maria, “Namenregister zu Ḳazwīnī's ‘Ātār al-bilād’,” Rocznik Orientalistyczny, 29(1965), 99–115Google Scholar; 30(1966), 119-34. Although personal and tribal names are listed, Abū Bakr b. ʿUmar, Yūsuf b. Tāshfīn and the Lamtūna all do not appear.

78. But which Hopkins, and Levtzion, , Corpus, 179–80Google Scholar, include; they translate namir as “panthers.”

79. Wüstenfeld, , Kosmographie, 2:37–38.Google Scholar Al-Qazwīnī's source for the hollow trees was a faqīh, Abū 'l-Rabīʿ Sulaymān al-Multānī. The city of Multan, as al-Qazwīnī's own description of it makes clear (81-82), is in India, and for an Indian to be an authority on the road from Sijilmasa to Ghana would be a remarkable illustration of early mobility within the Muslim world. However, Abū'l-Rabīʿ is cited on a number of other occasions (p. 73, line 4 from bottom; p. 100, line 5; p. 117, line 11 from bottom; p. 139, lines 7-8; and p. 182, line 5), invariably for North African material. It seems at least possible that al-Multānī is a misreading of al-Lamtūnī (or even, though much less credibly, al-Mulaththamī). Lewicki, T., Arabic External Sources for the History of Africa to the South of Sahara (Warsaw, 1969), 76Google Scholar, suggests Milyānī, from Milyāna in Algeria.

80. Cuoq, , Recueil, 201ff.Google Scholar Cuoq relies upon Ibn Saʿīd's book, Kitāb bast al-ard fī'l-ṭūl wa'l-arḍ, an edition of which was published in Tetuan in 1958, edited by J. Vernet-Ginès. We have also consulted this work. Cuoq does not refer to Ibn Saʿīd's Kitāb al-mughrib fī ḥulā al-maghrib, which has not yet been published in its entirety. Those published portions which we have been able to consult appear not to contain material relevant for us, although a volume of selections concerned with Cairo, , Al-nujūm al-zāhirah fī ḥulā, ḥaḍrat al-Qāhirah, ed. Naṣṣār, Ḥusayn (Cairo, 1970)Google Scholar does refer several times (67, 143, 183, 314) to the Sudanese community in Cairo.

81. Cuoq, , Recueil, 205, 206Google Scholar; Hopkins, and Levtzion, , Corpus, 186.Google Scholar In the passage translated by Fagnan, E., Extraits inédits relatifs au Maghreb (Algiers, 1924), 19Google Scholar, there is mention of gold, slaves, and musk from Ghana. The importance of Sudanese people, most of them presumably slaves or of slave origin, in Mediterranean society also appears in Ibn Saʿīd's anthology of poetry compiled in 1243 (see Arberry, A.J., tr, Moorish Poetry [Cambridge, 1953]).Google Scholar

82. Cuoq, , Recueil, 205, 206Google Scholar; Hopkins, and Levtzion, , Corpus, 186.Google Scholar

83. Recueil, 219-24; Corpus, 216-32.

84. Ibid, 231; Kitāb al-bayān al-mughrib fī akhbār al-andalus wa'l-maghrib, ed. Iḥsān ʿAbbās (Beirut, 1967) 4:25; not in Cuoq.

85. Ibid, 4:29-30.

86. Cuoq, , Recueil, 224Google Scholar; Hopkins, and Levtzion, , Corpus, 232Google Scholar; ʿAbbās, Iḥsan, Kitāb, 4:26.Google Scholar

87. Miranda, A. Huici, Al-Bayān al-mugrib…por ibn ʿIdārī al-Marrākusī (Tetuan, 1953–1954) 1:150n.Google Scholar This is a Spanish version and it appears that the Arabic rendering of Ghānah is a little odd.

88. ʿAbbās, Iḥsān, Kitāb, 4:13Google Scholar; the English is from Hopkins, and Levtzion, , Corpus, 221.Google Scholar

89. Cuoq, , Recueil, 236.Google Scholar Of the editions which Cuoq has consulted we have seen the French of Beaumier, A., Roudh el-kartas: histoire des souverains du Maghreb (Paris, 1860)Google Scholar and the Arabic and Latin of Tornberg, C.J., Annales Regum Mauritaniae (Uppsala, 1843–1846).Google Scholar The references for this passage are Beaumier, , Roudh, 175Google Scholar, and Tornberg, , Annales, 1:81/4 and 2:110.Google Scholar See also Hopkins, and Levtzion, , Corpus, 242.Google Scholar

90. Cuoq, , Recueil, 236Google Scholar; Beaumier, , Roudh, 176Google Scholar; Tornberg, , Annales, 1:81Google Scholar, lines 5-4 from bottom, and 2:111; Hopkins, and Levtzion, , Corpus, 243, 409n20.Google Scholar

91. Cuoq, , Recueil, 238–39Google Scholar; Beaumier, , Roudh, 187–88Google Scholar; Tornberg, , Annales, 1:86Google Scholar, lines 15-16 and 2:118; Hopkins, and Levtzion, , Corpus, 247.Google Scholar

92. Tornberg, , Annales, 2:119Google Scholar, errs in suggesting that Shaʿbān began on 7 April in A.H. 480.

93. Cuoq, , Recueil, 239Google Scholar; Beaumier, , Roudh, 189–90Google Scholar; Tornberg, , Annales, 1:87Google Scholar, line 6ff; 2:119; Hopkins, and Levtzion, , Corpus, 247–48.Google Scholar

94. Cuoq, , Recueil, 233Google Scholar; Hopkins, and Levtzion, , Corpus, 239.Google Scholar

95. The mountain of gold is presumably the source of Delafosse's reference to the Almoravid annexation and to their political domain of all the dependencies of Ghana as far as and including the gold-bearing mountains of Bambuk; Delafosse, , Haut-Sénégal-Niger, 2:54.Google Scholar Delafosse himself does not mention any source, but this was his almost invariable practice.

96. Fragments historiques sur les Berbères au moyen-âge: extraits inédits d'un recueil anonyme compilé en 712-1312 et intitulé Kitab Mafakhir al-Barbar, ed. Lévi-Provençal, E., Arabic text (Rabat, 1934), 2.Google Scholar We are grateful to Dr. H.T. Norfis for bringing this reference to our attention.

97. Cuoq, , Recueil, 239Google Scholar; Hopkins, and Levtzion, , Corpus, 195-203, esp. 201–02.Google Scholar

98. Cuoq, , Recueil, 239–47.Google Scholar Cuoq's material, not quite complete, is taken from the Arabic text published by M.A.F. Mehren, Cosmographie de Chems-ed-din Abou Abdallah Mohammed ed-Dimiohqui; originally issued in St. Petersburg in 1866, this was reprinted in Leipzig in 1923. The index misses references to Ghana on pp. 89 and 154. There is a French translation, also by Mehren, under the title Manuel de la cosmographie du moyen âge, originally published in Copenhagen in 1874, and re-issued in Amsterdam in 1964. We have consulted all three. See also Hopkins, and Levtzion, , Corpus, 204–14.Google Scholar

99. For geography, see Cuoq, , Recueil, 239–47 passimGoogle Scholar; Mehren, , Cosmographie, 19, 22, 88-89, 110-12, 133, 240–41Google Scholar; Mehren, , Manuel, 20, 105-07, 136-39, 173, 341–43.Google Scholar For gold see Cuoq, , Recueil, 243, 244, 247Google Scholar; Mehren, , Cosmographie, 50 111, 167, 238, 268Google Scholar; Mehren, , Manuel, 55, 138, 225, 338, 388.Google Scholar For hippopotami see Mehren, , Cosmographie, 90Google Scholar; and Mehren, , Manuel, 109.Google Scholar For parrots see Mehren, , Cosmographie, 15AGoogle Scholar; and Mehren, , Manuel, 205–06Google Scholar, and compare Cuoq, , Recueil, 242, 246.Google Scholar

100. Ibid, 245; for al-Bakrī see ibid, 98, for the Istibṣār, ibid, 177. For the same passage in Mehren, see Cosmographie, 240, 1. 7, and Manuel, 341.

101. Cuoq, , Recueil, 244Google Scholar; Mehren, , Cosmographie, 238, 1.Google Scholar 13-16; Mehren, , Manuel, 338.Google Scholar

102. Mehren, , Cosmographie, 268, 1. 6-7Google Scholar; Cuoq, , Recueil, 247Google Scholar reads “parmi les peuplades musulmanes assujetties;” Mehren, , Manuel, 388Google Scholar, reads “parmi les tribus des Nègres assujétis et musulmans.” John Hunwick, commenting on an early draft of this paper, remarked that al-khadamīn, lexically strange, might mean those from whom slaves, khadam, are taken.

103. Mehren, Cosmographie, 268n. Curiously, the same two manuscripts read Kān.m, almost certainly a more accurate rendering, in place of Ghān.m.

104. Corpus, 213.

105. Recueil, 249-50.

106. Ibid, 251, citing 1:209-10.

107. Remiro, G., Historia de los musulmanes de España y Africa por en-Nuguairi (Granada, 1917).Google Scholar This is labelled vol. 1. The old Encyclopaedia of Islam, sub “al-Nuwayrī,” mentioned a second published in 1919 but we have not seen this.

108. ”Histoire de la province d'Afrique et du Maghrib, traduite de l'arabe d'En-Noweiri, par le baron Mac Guckin de Slane,” Journal asiatique, 3d ser., 11 (1841), 97-135, 557–83Google Scholar; 12 (1841), 441-83; 13 (1842), 49-64.

109. Cuoq, , Recueil, 252Google Scholar, mentions ʿAbd al-mu'min, but does not quote from him. We have consulted the printed Arabic of Juynboll, T.W.J., ed., Lexicon geographicum (6 vols.: Leiden, 1850–1864), Fasc.b, page 300.Google Scholar

110. Cuoq, , Recueil, 254ffGoogle Scholar, Hopkins, and Levtzion, , Corpus, 252ff.Google Scholar

111. Cuoq, , Recueil, 257.Google Scholar

112. Ibid, 263. Another possible mention of Zāfūn occurs in the passage from Ibn Zunbul translated by Fagnan, , Extraits, 178–79Google Scholar where, among places to the south which are reached via Sijilmasa, there is Rājūn, or, according to a variant reading, Zāfūr. The difference between ‘r’ and ‘n’ in Arabic, coming at the end of a word in manuscript, may be slight.

113. Cuoq, , Recueil, 265Google Scholar, or see Gaudefroy-Demombynes, M.Masālik el abṣār fi mamālik el amṣār: Ibn Faṣl Allāh al-ʿOmarī: I: L'Afrique, moins l'Egypte (Paris, 1927), 59Google Scholar, or Hopkins, and Levtzion, , Corpus, 262.Google Scholar

114. Cuoq, , Recueil, 286.Google Scholar The booklet has been published as Al-taʿrīf bil-muṣṭalah al-sharīf, p. 27; 1894 it is also partially reproduced in al-Qalqashandī's Ṣubḥ al-aʿshā, for which see below, but this particular reference to Ghana is omitted there (viii.9). Or see Hopkins, and Levtzion, , Corpus, 276.Google Scholar

115. Cuoq, , Recueil, 269Google Scholar; Gaudefroy-Demombynes, , Masālik, 64.Google Scholar

116. Lisān ad-Dīn ibn al-Khaṭīb: Histoire de l'Espagne musulmane, extraite du Kitāb aʿmāl al-aʿlām, Arabic text edited by E. Lévi-Provençal (Rabat, 1934), 278, lines 3-4 from bottom. A German translation gives, with perfect accuracy “Aufbruch aus der Sahara” and “Besetzung Marokkos;” Hoenerbach, W., Islamische Geschichte Spaniens: Ubersetzung der aʿmāl al-aʿlām und ergänzender Texte (Zurich, 1970), 436.Google Scholar

117. Cuoq, , Recueil, 329.Google Scholar

118. The Arabic is taken from the Kitāb tārīkh al-duwal al-islāmiyya bi'l-maghrib (2 vols.: Algiers, 1263–7/1847–1851)Google Scholar a partial edition by de Slane of the Kitāb al-ʿibar, 1:263. We have also consulted the Beirut edition of the Kitāb al-ʿibar (1956–1961) 6:413Google Scholar, where the text appears to be not quite so reliable. There is also a Cairo edition of 1867, for which see 6:200. Cuoq, , Recueil, 343Google Scholar, gives a French translation, as does de Slane, , Histoire des Berbères (4 vols.: Paris, 1925 56), 2:110.Google Scholar We have been helped in our English by Hopkins, and Levtzion, , Corpus, 333.Google Scholar

119. See above, p.; Hopkins, and Levtzion, , Corpus, 330Google Scholar; Kitāb tārīkh, 1:240Google Scholar; or, in the Cairo edition of 1867, 6:184, line 9.

120. Referring to Sanhaja activity before the Almoravids Ibn Khaldūn spoke elsewhere in the Kitāb al-ʿibar of jihād waged against the peoples of the sūdān, bringing them to Islam, which many accepted while others paid the jizya or tax (Beirut, 1956–1961, 6:371–72Google Scholar; Cuoq, , Recueil, 333–34Google Scholar; Hopkins, and Levtzion, , Corpus, 328Google Scholar; Histoire des Berbères, 2:65Google Scholar). The wording is very similar to that in the passage quoted on above. Cuoq comments that these military operations refer “sans doute” to a Sanhaja expedition against Takrur, but such an identification seems in fact rather doubtful. A source for Ibn Khaldūn may have been the anonymous compilation of 712/1312, Kitāb mafākhir al-barbar, published by E. Lévi-Provençal under the title Fragments historiques sur les Berbères au moyen-âge (Rabat, 1934)Google Scholar, which discusses pre-Almoravid period on p. 57. The Almoravids are mentioned on p. 69, where it is said that their emergence, khurūj, from the desert was by order of Abū ʿImrān al-Fāsī. Abū Bakr and Ghana are not mentioned there. See also Hopkins, and Levtzion, , Corpus, 233.Google Scholar

121. This is again from the Kitāb al-ʿibar (Beirut, 1956–1961, 5:930–31Google Scholar; Cairo, 1867, 5:433); translations are in Hopkins, and Levtzion, , Corpus, 322Google Scholar, and Cuoq, , Recueil, 355.Google Scholar The Arabic is as follows:

122. Cuoq, , Recueil, 351Google Scholar; Muqaddimah Ibn Khaldūn: Prolégomènes d'Ebn-Khaldoun, ed. Quatremère, M. (3 vols.: Paris, 1858) 1:96Google Scholar; Hopkins, and Levtzion, , Corpus, 320.Google Scholar

123. Hopkins and Levtzion, Corpus, 421n1, while very sympathetic to Ibn Khaldūn, have obvious difficulty in fashioning a snug fit for this passage. It should, they say, “be compared with the more detailed (and accurate) account of Ibn Khaldūn. There he locates the Ṣūṣū east of Ghāna, whereas here he places them nearest the Atlantic, i.e., the most western people of the Sudan. (The Soso, in fact, lived south of the Soninke of Ghana.) The conquest referred to here is the Almoravid conquest, which according to Ibn Khaldūn brought about the conversion of many of the Sūdān. (Those who were converted, however, were the northern Soninke and not the Soso.…)” The frequent qualifications and corrections merely confirm the fundamental uncertainty of the account; and there is, in our view, no reason to assume that fatḥ means an Almoravid conquest, that term being nowhere else in the sources applied to that alleged event, whereas it is the standard designation for the Muslim conquest of North Africa in the seventh and eighth centuries. Cuoq, , Recueil, 355Google Scholar, despite his profession of the Almoravid conquest hypothesis, renders fatḥ here as “la conquête (du Maghrib par les Arabes).” Lange, D, “Un texte de Maqrīzī sur ‘Les races des sūdān’,” Annales Islamologiques, 15(1979), 207n2Google Scholar, gives further examples of errors which Ibn Khaldūn allowed to creep into material he derived from other sources.

124. Cuoq, , Recueil, 364Google Scholar; also Allouche, I.S., ed., Al-Hulal al-mawchiyya: chronique anonyme des dynasties almoravide et almohade: texte arabe publié d'après de nouveaux manuscrits (Rabat, 1936) iiiGoogle Scholar; and Hopkins, and Levtzion, , Corpus, 309.Google Scholar

125. Cuoq, , Recueil, 364Google Scholar; contrast Hopkins, and Levtzion, , Corpus, 310.Google Scholar

126. Cuoq, Recueil, 364n3. The original passage in al-Zuhrī about the Christians is on p. 120 in Cuoq, , Recueil, and in para. 339Google Scholar of Hadj-Sadok, “Kitāb.” This error by al-Ḥulal, the Christianization of the people of Ghana, has found its way into the secondary sources. Cornevin, , Histoire, 244Google Scholar, reports it as fact, relying upon Desplagnes, Louis, Le plateau central nigérien (Paris, 1907), 114Google Scholar, who quotes al-Zuhrī.

127. Allouche, Al-Hulal, esp. 7. There is also an Arabic edition published in Tunis under the title Al-Ḥulal al-mawshiya fī dhikr al-akhbār al-marrākushīya, which on p. 7 gives almost the same text, the differences being, we suspect, the result of typographical errors in Tunis. Only the opening fragment of this publication is available at the School of Oriental and African Studies (S.G. VII 141.5 76285); interestingly, the table of contents on p. 15 of the introduction speaks of the khurūj of the Almoravids from the Sahara, and their dukhūl into the Maghrib. The date of the publication is uncertain, and the name of the author is variously given; compare the SOAS library catalog, and Tarkhan, I.A., Imbarāturīyah Ghānat al-islāmīydh (Cairo, 1970), 95.Google Scholar

128. Or, according to one manuscript, Amīr Abū Yaḥyā b. Abū Zakariyyā b. ʿUmar. See Miranda, A. Huici, ‘Al-Ḥulal al mawsiyya’: cronica arabe de las dinastias Almoravide, Almohade y Benimerin: traduccion española (Tetuan, 1951), 24n1.Google Scholar

129. Ibid, 23-24.

130. Cuoq, , Recueil, 368.Google Scholar See Allouche, , Al-Hulal, 17Google Scholar, for the Arabic text, and Miranda, Huici, Al-Hulal, 41Google Scholar, for a Spanish translation, both these adding that Abū Bakr's death occurred in one of the wars which arose between him and the Sūdān. Hopkins, and Levtzion, , Corpus, 316, 420n23Google Scholar, point out that the whole section about Ibn Tāshfīn is taken from Ibn ʿIdhārī.

131. Ṣubḥ al-aʿshā (Cairo, 1913–1920), 5:284Google Scholar; the Arabic reads:

Or see Cuoq, , Recueil, 372, 372n.Google Scholar Cuoq attaches al-Qalqashandi's reference to that passage in the ʿIbar quoted on p. above (6:413 in the Beirut edition of 1956-61), but it seems more likely to derive from the passage discussed on p. above (5:930-31 in the Beirut edition).

132. Ibn Khaldūn speaks simply of al-fatḥ, “the conquest;” al-Qalqashandī of awwal al-fatḥ al-islāmī, “the beginning of the Islamic conquest.” Both forms of words appear from context and usage to refer to the Muslim conquest of North Africa. See also note 123 above.

133. Ṣubḥ, 8:115Google Scholar; Cuoq, , Recueil, 376Google Scholar; Hopkins, and Levtzion, , Corpus, 346.Google Scholar

134. The Arabic is given by Hamakar, H.A., Specimen catalogi codicum mss. orientalium bibliothecae Academiae Lugduno-Batavae (Leiden, 1820), 207.Google Scholar See note 141.

135. Cuoq, , Recueil, 388, 388nGoogle Scholar; compare Hopkins, and Levtzion, , Corpus, 355.Google Scholar

136. Lange, , “Un texte,” 189–90.Google Scholar

137. Cuoq, , Recueil, 395–97.Google Scholar

138. Fagnan, , Extraits, 285, 354–56.Google Scholar

139. Hazard, , Numismatic History, 61.Google Scholar

140. A brief but intriguing article by Levtzion, N. is relevant here (“Was Royal Succession in Ancient Ghana Matrilineal?,” IJAHS, 5[1972], 91–93).Google Scholar Drawing on al-Bakrī, Levtzion points out that Bāsi, king of Ghana and friend of the Muslims, who died in 1063, may have been the father of Qanmar son of Bāsi, who was the king of Alukan (or Alwakan) and a secret Muslim. Perhaps Qanmar, the natural heir to the throne, had been sent away to a lesser post because his Islamic loyalties were too strong, leaving Bāsi's sister's son to inherit (as he did). If this interpretation is correct, it would be a fairly strong indication of a developing internal loyalty to Islam within Ghana.

141. See n. 134 above. Lange, , “Un texte,” 209Google Scholar, gives the same rendering in French, but his Arabic, 195, seems to read in the final phrase, “Then the people of Mali and their kings became strong” - mulūkuhum rather than malakūhum.