1 The use of these sources for the non-Arabicist historian has been greatly facilitated by the recent publication of edited French and English translations of accounts dealing with West Africa. See Cuoq, Joseph M., Recueil des sources arabes concernant l'Afrique Occidentale du viii^e au xvi^e siècle (Paris, 1975), and Levtzion, Nehemia and Hopkins, J. F. P., Corpus of Early Arabic Sources for West African History (Cambridge, 1981).Google Scholar Most of the translations used in this paper are from Levtzion and Hopkins, hereafter referred to as Corpus.

2 There seem to be many accepted spellings for this oasis. I have adopted the most recent English transliteration for this (and most other place and tribe names used throughout) from Levtzion and Hopkins, Corpus. (The town's name Awdaghust is distinct from the name presently given to the archaeological site, Tegdaoust.)

3 The major work is a series of publications (some still to appear) based on the archaeological findings at Tegdaoust between 1960–1 and 1976. Under the direction of Devisse, Jean, Robert-Chaleix, Denise and Robert, Serge, three volumes in the collection Tegdaoust: recherches sur AoudaghostGoogle Scholar have appeared: Robert, D. And S., Devisse, J., Tegdaoust I. Recherches sur Aoudaghost (Paris, Arts et Métiers Graphiques, 1970)Google Scholar; Vanacker, Cl., Tegdaoust II. Recherches sur Aoudaghost. Fouille d'un quartier artisanal. Mémoires de l'I.M.R.S. n°2 (Paris, Arts et Métiers Graphiques, 1979)Google Scholar; Devisse, J., Robert-Chaleix, D. et al. , Tegdaoust III. Recherches sur Aoudaghost (Paris, A.D.P.F., 1983).Google Scholar Awaiting publication are: Polet, J., ‘Tegdaoust IV. Recherches sur Aoudaghost. Fouille d'un quartier: urbanisation, architecture, utilisation de l'espace construit’ (Thèse soutenue à Paris I, 1980)Google Scholar; Robert-Chaleix, D., ‘Tegdaoust V. Recherches sur Aoudaghost. Une concession médiévale, implantation et évolution d'une unité d'habitation’ (Thèse soutenue à Paris I, 1981)Google Scholar; Saison, B., ‘Tegdaoust VI. Fouille d'un quartier artisanal’ (Thèse soutenue à Paris I, 1979).Google Scholar Of particular interest here are two contributions to Tegdaoust I: Devisse, J., ‘La question d'Awdaghust’, 109–56Google Scholar, and El-Chennafi, Mohammed, ‘Sur les traces d'Awdaghust’, 97–107;Google Scholar nd the following in Tegdaoust III: Devisse, J., ‘En guise de conclusion: histoire et évolution de l'environnement’, 387–98Google Scholar; idem, ‘La question d'Awdaghust (II)’, 533–538; idem, ‘Conclusion générale’, 539–556; Meillassoux, Claude, ‘A propos de deux groupes Azer: les Giriganko-Tegdawest et les Maxanbinnu’, 525–32Google Scholar; Robert-Chaleix, D., ‘Fusaïoles décorées du site de Tegdaoust’, 447–513.Google Scholar For other publications concerning Tegdaoust/Awdaghust and additional work on other medieval sites (Kumbeh Saleh, Azuggi, Région du Fleuve Sénégal, villes anciennes) see Robert-Chaleix, D., ‘Recherches archéologiques en République Islamique de la Mauritanie’, Journal des Africanistes, LIV (forthcoming, 1985).Google Scholar Among the most useful for historians, Robert-Chaleix notes the following: Robert, D., ‘Les fouilles de Tegdaoust’, Journal of African History, XI (1970), 471–493CrossRefGoogle Scholar; Vanacker, Cl., ‘Cuivre et métallurgie du cuivre à Tegdaoust (Mauritanie orientale). Découvertes et problèmes’, in Echard, N. (ed.), Métallurgies africaines. Nouvelles contributions, Mémoires de la Société des Africanistes (Paris, 1983)Google Scholar; Robert-Chaleix, D., ‘Lampes à huile importées découvertes à Tegdaoust. Premier essai de classification’, Journal des Africanistes, LIII (forthcoming, 1984)Google Scholar; Robert, S., ‘Archéologie des sites urbains du Hodh et problèmes de la désertification saharienne au Moyen-Age’, in Colloque sur les problèmes de la désertification au Sud du Sahara: le cas de la Mauritanie (Nouakchott, 1973; Abidjan-Dakar, 1975)Google Scholar; Berthier, S., ‘Une maison du Quartier de la Mosquée à Koumbi Saleh (R.I.M.)’, Mémoire de maîtrise présenté à l'U.E.R. de Sciences de l'Homme et de l'Environnement, Section Histoire (Lyon, 1978)Google Scholar; Robert, S., ‘Rapport des fouilles archéologiques sur le site de Koumbi Saleh’ (Institut Mauritanien de la Recherche Scientifique [I.M.R.S.], Nouakchott, 1980)Google Scholar; idem, ‘Rapport de prospection régionale et de fouilles archéologiques sur le site de Koumbi Saleh’ (I.M.R.S., Nouakchott, 1980); Evin, J. and Robert, S., ‘Etudes des datations 14C d'époque médiévale: site de Koumbi Saleh’, in Records on datations 14C (New York, 1982)Google Scholar; Berthier, S., ‘Etude archéologique d'un secteur d'habitat à Koumbi Saleh (Mauritanie)’, Thèse de Doctorat de 3^e Cycle, Lyon II (Sept. 1983)Google Scholar; Saison, B., ‘Azuggi: archéologie et histoire en Adrar mauritanien’, Recherche, Pédagogic et Culture, IX, Iv (Sept.–Dec. 1981), 66–74Google Scholar; Wedoud, AbdelCheikh, Ould and Saison, B., ‘Vie(s) et mort(s) de al-Imam al Hadrami. Autour de la postérité saharienne du mouvement almoravide (XI^e–XII^e siècles)’, (I.M.R.S., Nouakchott, 1983)Google Scholar; Tandia, B., ‘Sites d'habitats anciens sur la rive mauritanienne due Fleuve Sénégal’, Mémoire de fin d'études, Ecole Normale Supérieure (Nouakchott, 1983)Google Scholar; Sognane, M., ‘La métallurgie ancienne du fer dans la Vallée du Sénégal’, Mémoire de fin d'études, Ecole Normale Supérieure (Nouakchott, 1983)Google Scholar; Robert-Chaleix, D. and Sognane, M., ‘Une mélallurgique ancienne sur la rive mauritanienne du Fleuve Sénégal’ in Métallurgies AfricainesGoogle Scholar; Vernet, R., ‘La préhistoire de la Mauritanie. Etat de la question’, Thèse de Doctorat de 3^e Cycle, Paris I (1983)Google Scholar; Holl, A., ‘Essai sur l'économie néolithique du dhar Tichitt (Mauriianie)’, Thèse de Doctorat de 3^e Cycle, Paris I (1983)Google Scholar; Lambert, N., ‘Les industries du cuivre dans l'Ouest africain’, West African Journal of Archaeology, I (1971), 9–21Google Scholar; idem, ‘La parure préhistorique en Europe méridionale et en Afrique du Nord-Ouest’, in Mélanges Balout (Paris, 1981), 361–366Google Scholar; idem, ‘Nouvelle contribution à l'étude du Chalcolithique de Mauritanie’, in Métallurgies AfricainesGoogle Scholar; Ould, BiddihSalem, Ahmed, ‘Essai de répartition spatio-temporelle des populalions néolilhiques el prolohisloriques de la région de Nouakchott’, Mémoire de fin d'études, Ecole Normale Supérieure (Nouakchott, 1983).Google Scholar

4 For a general background to this early period see Timingham, J. Spencer, A History of Islam in West Africa (Oxford, 1962, reprint 1970), 1–33Google Scholar; Levtzion, N., Ancient Ghana and Mali (London, 1973; 2nd ed., 1980), 3–42, 136–152Google Scholar; idem, ‘The Sahara and Sudan from the Arab conquest of the Maghrib to the rise of the Almohads’, in Fage, J. D. (ed.), Cambridge History of Africa, II (Cambridge, 1978), 637–684Google Scholar, including a useful bibliography; De La Chapelle, F., ‘Esquisse d'une histoire du Sahara occidental’, Hespéris, XI (1930), 35–95.Google Scholar For a fascinating critical discussion of the sources used to reconstruct this period Norris, H. T.Saharan Myth and Saga (Oxford, 1972)Google Scholarcannot be too highly recommended. On the Almoravids in particular: Levtzion, Ghana and Mali, 29–42Google Scholar; de Moraes Farias, Paulo F., ‘The Almoravids: some questions concerning the character of the movement during its period of closest contact with the western Sudan’, Bulletin de l'I.F.A.N., sér. B, XXIX (1967), 794–878Google Scholar; Semonin, Paul, ‘The Almoravid movement in the Western Sudan: a review of the evidence’, Transactions of the Historical Society of Ghana, VII (1964), 43–59Google Scholar; Norris, H. T., Myth and Saga, especially 90–217Google Scholar; idem, ‘New evidence on the life of ‘Abdullah b. Yāsīn and the origins of the Almoravid movement’, J. Afr. Hist., XII, ii (1971), 255–268Google Scholar; McCall, Daniel F., ‘Islamization of the Western and Central Sudan in the eleventh century’, in McCall, Daniel F. and Bennett, Norman, Aspects of West African Islam, Boston University Papers on Africa, V (Boston University, 1971), 20–33Google Scholar; Conrad, David and Fisher, Humphrey, ‘The conquest that never was: Ghana and the Almoravids, 1076. I. The external Arabic sources’, History in Africa, IX (1982), 21–59CrossRefGoogle Scholar; Fisher, H. J., ‘Early Arabic sources and the Almoravid conquest of Ghana’, J. Afr.Hist., XXIII, iv (1982), 549–560.CrossRefGoogle Scholar Jean-Louis Triaud discusses the nature of Islam during this period, including the Almoravids, in Islam et sociétés soudanaises au Moyen-Age: etude historique, Recherches Voltaïques, XVI (Paris–Ouagadougou, 1973), 17–54.

5 Toupet, Charles, La sédentarisation des nomades – Mauritanie centrale sahélienne (Paris, 1977). 150–152Google Scholar; on the Adrar, Modat, Lt. Col., ‘Les populations primitives de l'Adrar mauritanien’, Bulletin d'Etudes Historiques et Scientifiques de l'Afrique Occidentals Française (1919), 379–392Google Scholar; on the Tagant (and Hodh) Munson, Patrick J., ‘Archaeology and the prehistoric origins of the Ghana empire’, J. Afr. Hist., XXI, iv (1980), 457–466.CrossRefGoogle Scholar See also Amblard, S., ‘La civilisation néolithique du dhar Tichitt-Walata à travers son matériel lithique’, Thèse de Doctorat de 3^e Cycle, Paris I (1981)Google Scholar (not consulted, soon to be published), and further references in Robert-Chaleix, ‘Recherches archéologiques’: above, n. 3.

6 Levtzion, , Ghana and Mali, 3Google Scholar; Toupet, , Sédentarisation des nomades, 157–161,Google Scholar and map of location of Neolithic, Bafur and Gangara sites, 156; idem, ‘La vallée de la Tamourt en-Naaj (Tagant)’, Bulletin de l'I.F.A.N., sér. B, XX (1958), 77.Google Scholar Traditions are agreed that these ‘post-neolithic, pre-Islamic’ villages belonged to black people called ‘Gangara’, ancestors of the present-day Soninke (Sarracolet).

7 No one (yet) seems able to provide dates for the various aspects of Gangara civilization. Toupet argues on the basis of al-Bakrī's reference to ‘black brigands and robbers’ infesting the Azgûnan mountains that the Gangara had established themselves in the Assaba by the tenth/eleventh centuries. These ‘upland locations’ may reflect a search for a more defensive position, or as Toupet points out, a search for better-quality soils and water supplies: Toupet, , Sédentarisation des nomades, 157–160Google Scholar; see also Daveau, S. and Toupet, Ch., ‘Anciens terroirs Gangara’, Bulletin de l'I.F.A.N., sér. B, XXV (1963), 193–214Google Scholar; Munier, Paul A. P., ‘Ruines Gangara dans l'Assaba (Mauritanie)’, Notes Africaines, LXXIV (avril 1957), 34–35.Google Scholar

8 Toupet, , Sédentarisation des nomades, 159–160Google Scholar; quotation Daveau and Toupet, ‘Anciens terroirs Gangara’, 211.

9 Daveau and Toupet raised the question some years ago: ‘Anciens terroirs Gangara’, 210, 211. Recently, Munson, ‘Archaeology’, 462–463, has made an interesting argument in favour of this continuity with reference to the Soninke kingdom of Ghana, but the hiatus in the evidence (from c. 300 B.C. to c. A.D. 500–600) still speaks louder than superficial resemblances in architecture and ceramics. His suggestion that Azer, a Soninke/Berber dialect spoken in the commercial oases of the Adrar–Tagant–Hodh in the Middle Ages, had its origins in ‘the final phase of the Tichit Tradition’ (c. 600–300 B.C.) is not totally convincing. From what we know of this ‘trade language’ its origins were more recent, and its use fairly closely confined to groups involved in central Mauritanian commerce. Munson's theory that the Soninke speakers of the Dhar Tishit were ‘encapsulated’ by Libyo-Berbers and that this gave rise to Azer does not really explain why it was not more widespread, especially after several centuries, why it would be adopted by selected Berber tribes, and why it would also take root in the Adrar (Shinqiti, Wadan) long after Soninke cultivators had moved south: see below, nn. 119–121.

10 Modat, , ‘Populations primitives’, 379–386.Google Scholar The traditions he collected suggest they arrived during the first centuries of our era and co-existed with black (Gangara?) cultivators. In the eighth century, Berbers in the Draa area were converted to Islam and during the second half of the century would have begun their infiltration into the Adrar–Tagant. From time to time they were able to challenge the supremacy of the local ‘black race’; the Bāfūr remained cultivators. See also Huguet, Capt., ‘Les populations primitives de l'Adrar mauritanien: les gîtes bafor de la subdivision d'Atar et quelques autres’, Bulletin du Comité de l'Afrique Française, Renseignements Coloniaux, III (Mar. 1927), 119–124Google Scholar; Lucas, A. J., ‘Considérations sur l'ethnique Maure et en particulier sur une race ancienne: les Bafours’, Journal de la Societé des Africanistes, I, ii (1931), 151–194CrossRefGoogle Scholar; Norris, , Saharan Myth and Saga, 65, 80, 127.Google Scholar

11 According to Norris, ‘Béfúr (Ba thawr = ‘Bull father’) [is] a vague and general term applied to pre-Ṣanhāja [Znāga] inhabitants of the Western Sahara, part Berber, part negro, part Semite’: Myth and Saga, 139. Toupet is convinced of their white Berber origin: ‘Sedentarisation des nomades’, 155, 157. Thomas Whitcomb's work on the powerful Kunta group has disclosed evidence pointing strongly to Znāga origins, irrespective of their occupation. ‘It would seem likely that they [the Bāfūr] were simply an old Znāga tribe which may have been of considerable importance, but which was defeated, possibly disgraced, and absorbed by other tribes.… The original form of the name of the region of the Adrār may have been Adrār an Bāfūr, “the mountain of the Bāfūr”, as suggested in the accounts of ninth/fifteenth-century Portuguese explorations’: Whitcomb, Thomas, ‘New evidence on the origins of the Kunta’, 2 parts, Bulletin of S.O.A.S., XXXVIII, i and ii (1975), 103–123, 403–417Google Scholar: I, 120, n. 85.

12 Toupet, , Sédentarisation des nomades, 155–157.Google Scholar

13 Huguet, ‘Populations primitives...gîtes bafor’, 119–24, passim. Norris presents an intriguing account of the Adrar conquest by the Almoravids in which wealthy but ‘non-numerous’ Christians were said to be in control of the plateau oasis, and populous, powerful but ‘to no extent wealthy’ herders occupied the valleys. While the description of the economic development of the Adrar at the time is probably reflective of reality, the role of various groups of Christians, ‘huge dogs trained to fight’, and pagan Negro Bāfūrs evokes the ‘strange ideas’ of the medieval Arab geographers Norris speaks of earlier: see his discussion of ‘Yemenite “two-horned” conquerors and the people of the veil’ in Myth and Saga, 26–93, esp. 65. In fact, this semi-legendary account may represent a real insight into the evolution of the Bāfūr from early Znāga goat and sheep herders who became increasingly involved in developing the Adrar's potential for date-palm cultivation. This is consistent with Huguet's contention about the region's ability to support a privileged few in the rare mountain oases and the growth of a differentiated social constitution, as well as Whitcomb's theory about Bāfūr origins: Huguet, loc. cit.; Whitcomb, ‘New evidence’ I, 120, n. 85.

14 This belief and the historical evidence which has nurtured its growth in the context of the western Sahara is discussed in my ‘The Sahara reconsidered: pastoralism, politics and salt’, African Economic History, XII (1983), 265–288. Since writing this article in which the term Ṣanhāja is employed, Whitcomb's argument has convinced me that Znāga is more appropriate. My thanks to Michael Brett and Humphrey Fisher for drawing Whitcomb's discussions to my attention.

15 Whitcomb, ‘New evidence’, I, 114, n. 1. Norris retains the use of Ṣanhāja, distinguishing ‘Znāga’ as ‘a name of a Berber dialect still spoken by some inhabitants of south-west Mauritania. The word is related to Ṣanhāja and Zanziga which may represent dialectical variations. Znāga (Ṣanhāja) is also a pejorative term in Mauritania to describe the laḥma or vassal class’: Myth and Saga, 228–230, n. 36, quotation n. 45. The ambiguity of the term and its meaning is underscored by the fact that in some parts of the western Sahara, 'it refers to a specific warrior tribe, the Idaw ‘Ish, whose chiefs claim descent from the Almoravid chiefs’: Whitcomb, loc. cit.

16 Most of the studies on pastoralism in recent years comment on these issues in passing; some give them more direct attention: Nicolaisen, J., Ecology and Culture of the Pastoral Tuareg (Copenhagen, 1963), 209–213Google Scholar; Baier, Stephen and Lovejoy, Paul E., ‘The desert-side economy of the Central Sudan’, Int. J. Afr. Hist. Studies, VIII (1975), 551–571Google Scholar; Smith, Susan E., ‘The environmental adaptation of nomads in the west African Sahel’, in Wolfgang, Weissleder (ed.), The Nomadic Alternative: modes and modelsof interaction in the African-Asian deserts and steppes (Paris, 1978), 75–96 (compare her chart of milk, meat and grain consumption, p. 80, with Nicolaisen's, p. 216).Google Scholar For the most complete study of the needs of pastoralists and their herds, as well as pastoral production in the real sense of the food value of herds, see Dahl, G. and Hjort, A., Having Herds: Pastoral Herd Growth and Household Economy (University of Stockholm, Stockholm series in Social Anthropology No. 2, 1976).Google Scholar Their findings show that there are substantial differences between the quality and quantity of milk produced by different animals and that seasonal variations and periodic droughts are not felt equally in the milk production of each.

17 Smith, , ‘Environmental Adaptation’, 76–80Google Scholar; Rogon, P., ‘Problèmes des Touaregs du Hoggar’, in UNESCO, Nomades et nomadisme au Sahara (Recherches sur la zone aride No. 9, 1963), 61–63Google Scholar; Capot-Rey, R., ‘Le nomadisme des Toubous’Google Scholar, ibid., 91; Hardy, G. and Richet, Ch., L'alimentation indigène dans les colonies françaises (Paris, 1933), introduction and 142–5Google Scholar; Sweet, L. E., ‘Camel raiding of North Arabian Bedouin: a mechanism of ecological adaptation’, American Anthropologist, LXVII (1965), 1132–1150, esp. 1138.CrossRefGoogle Scholar According to Sweet, ‘access to a market, to water supplies in summer, and to cultivators' surpluses of dates and wheat are essential ecological relations of camel pastoralists…’.

18 Leriche, A., ‘Phytothérapie maure: de quelques plantes et produits végétaux utilisés en thérapeutique’, in Mélanges Ethnologiques, Mémoires de l'I.F.A.N. no. 23 (Dakar, 1953), 268–306.Google Scholar There is some very useful information in this under-utilized study. The ‘date cure’ for nomads (like the ‘salt cure’ for their animals) is an important aspect of general health care. Dates and milk are taken as aphrodisiacs, and dates (preferably black dates) are used in a number of ways to treat biliousness and acne. In addition to its use in tanning and dyeing skins, henna is applied in ointments to wounds and sores, and drunk in an infusion for treatment of ‘inner organs’. Grains, especially millets, are used in many ways to cure eye illnesses, fevers, poisoning, anaemia, asthma, dysentery and whooping cough, among others. Leriche's research was carried out in the twentieth century, but it is highly likely that plants as medicines had similar, if not identical, predecessors and should be given more attention (along with disease itself) in historical studies. Humphrey Fisher points out that Gustav Nachtigal's description of his Saharan experiences includes a great deal of medical information which historians would do well to make use of: Nachtigal, Gustav, Sahara and Sudan, Eng. trans, with notes by Allen, G. B. and Fisher, Humphrey J., vol. IV, London (1971), vol. I, London (1974).Google Scholar

19 Too little attention has been given to the relation between oases and nomads in a historical context. Some of the issues involved are discussed in: Capot-Rey, R., Le Sahara François, I, II (Paris, 1953), 271–380, 303–367Google Scholar; Toupet, , ‘Sédentarisation des nomades’, 212–216Google Scholar; Dubié, Paul, ‘La vie matérielle des Maures’, in Mélanges Ethnologiques, Mémoires de l'I.F.A.N. no. 23 (Dakar, 1953), 111–251.Google Scholar

20 Corpus (al-Ya'qūbī), 22.Google Scholar

21 Ibid. (al-Bakrī), 68, 73–4. Al-Bakrī compiled his work in Spain from a variety of sources, some of which preceded by at least a century the eleventh-century manuscript itself.

22 Ibid. (Ibn Ḥawqal), 48, 49.

23 Ibid. (al-Ya'qūbī), 22; (Ibn Ḥawqal), 45–9; (al-Muhallabī), 197.

24 Ibid. (al-Bakrī), 73–4.

25 Ibid. (al-Bakrī), 73, 4. Though these statements were once thought to provide conclusive proof of Ghana's rule over Awdaghust from the tenth century (see for example Levtzion, Ghana and Mali, 28), the most recent interpretation of al-Bakrī's text is much more cautious. Hopkins and Levtzion note that ‘before the Arabs entered Ghana' could refer to the period before the first Arab traders reached Ghana (the eighth or ninth century) and that the reference to Ghana's authority could mean any time before the rise of the Almoravids, not necessarily that this was still the case in 1056–7: Corpus, 385, nn. 25, 26. There is also nothing to suggest that both references were to the same time period. It is quite feasible that Ghana's influence in the region ebbed and flowed according to other factors like the strength of the Znāga.

26 Ibid. (al-Idrīsī), 118. This appears to be inconsistent with archaeological evidence (see below).

27 Ibid. (al-'Umarī), 274; (al-Dimashqī), 209. He adds that this ‘sandy town with palms’ is a very unhealthy place.

28 The Tegdaoust ruins are situated at 17° 25′ latitude north and 10° 25′ longitude west in the region of southern Mauritania which has a Sahelian climate. The ruins are nestled in the small sandstone outcropping of the Rkiz, an ‘annex’ of the Affolé, located in the western part of the great depression known as the Hodh. (See Map 1; also Daveau, Suzanne, ‘Etude géographique de la région de Tegdaoust’, in Tegdaoust I, 39–61.)Google Scholar The main site is about 700 × 400 metres in surface, with cemeteries lying beyond, and considerable evidence of habitation outside this zone.

29 The earlier identification of four strata ( Tegdaoust I, 9, and Robert, D., ‘Fouilles de Tegdaoust’)Google Scholar has been considerably refined in the light of more recent analyses. There is as yet no agreement on the precise dating of these occupations. In Tegdaoust III, 554–6, Devisse presents the following hypotheses:

He emphsizes that the datings are provisional and that the work of Robert-Chaleix, Vanacker, Saison and Polet offers modifications which will have to be taken into account in later estimations. I have chosen to use the more detailed analysis of Robert-Chaleix, whose dates differ slightly for Occupations III to VI (see below).

30 Robert, D., ‘Fouilles de Tegdaoust’, 480Google Scholar; Tegdaoust I, 9; Devisse, , ‘Conclusion générale’, 554–556Google Scholar; and Robert-Chaleix, ‘Tegdaoust V: une concession médiévale’, whose chapters deal with each Occupation individually. Of particular interest is the chart she has compiled comparing each of the Occupations under the headings: ‘Carbon 14 datings’, ‘Characteristic elements’, ‘Most common objects unearthed’, ‘Pottery industry’, ‘Metalworking’, ‘Daily life’, ‘Imported goods’, ‘Relations North/South’, and ‘Relations East/West’. Drawing on the research of Vanacker, Saison and Polet (in addition to her own) she provides, at least for the present, the most comprehensive view of the evolution of Awdaghust's material culture: see esp. pp. 194–208.)

31 Devisse, , ‘Conclusion générate’, 554–556Google Scholar; Robert-Chaleix, , ‘Tegdaoust V’, 24, 57–58, 115–118, 167, 175–192, chart noted n. 39.Google Scholar

32 Robert-Chaleix, , ‘Tegdaoust V’, 190–192.Google Scholar

33 Vanacker's work especially discusses in detail the changes which occurred in the production of specific ceramics, jewellery and metal in the course of the occupations and is summarized in her conclusions: Tegdaoust II, 171–176. Of particular importance was the virtual disappearance of copper working in the twelfth century. Both Vanacker and Robert-Chaleix attach considerable significance to this development. Vanacker notes that we may have been overlooking an important industry whose techniques were diffused as far west as the lower Niger, and whose decline was largely due to changes in international trade (173–175). Robert-Chaleix postulates the interesting hypothesis that Awdaghust's processing of copper into brass may have been providing Ghana with a means of exchange (‘money’) and that destruction of this industry was the real target of eleventh-century Almoravid raids: ‘Tegdaoust V’, 281–4.

34 Robert, , ‘Fouilles de Tegdaoust’, 479–483Google Scholar; Vanacker, , Tegdaoust II, 171–175.Google Scholar Vanacker argues that the level of artisanal activity at Tegdaoust during the ninth, tenth and eleventh centuries was such that we need to re-evaluate the role of the town, no longer to see it purely as a centre of salt-gold exchange and an intermediary in international trade, but as an ‘active city’ participating in exchange based on its own artisanal production (p. 173).

35 Robert-Chaleix, , ‘Tegdaoust V’, 61–70.Google Scholar

36 Ibid., passim; conclusions, 281–283; Vanacker, , Tegdaoust II, 173–175Google Scholar; Robert, D., ‘Fouilles de Tegdaoust’, 483.Google Scholar

37 Corpus (al-Bakrī), 68, 74; Mauny, Raymond, Tableau géographique de l'ouest africain au moyen âge (Dakar, 1961), 341Google Scholar: ‘Although the texts do not speak specifically of them, the principal occupation assigned to the slaves was cultivation of the earth and the guarding of herds’. Mohammed el-Chennafi also suggests the intriguing possibility that slaves were used in Awdaghust as warriors: ‘Traces d'Awdaghust’, Tegdaoust I, 103.

38 Mauny, , Tableau géographique, 479.Google Scholar

39 Devisse, ‘Question of Awdaghust’, 109–56, takes into account the role of locally produced and traded salts and makes the much-needed link between the regional and international economy. But there remains a twofold problem: his assumptions that the salt trade acquired its importance to Awdaghust in the context of the trans-Saharan trade, rather than the regional economy (as will be discussed below); and that there was only one source of salt supplying the Awdaghust market in the tenth century (see my discussion of this in ‘The Sahara reconsidered’. Devisse's discussion of the role of the salt in the medieval political economy of the region is best developed in the important article ‘Routes de commerce et échanges en Afrique occidental en relation avec la Méditerranée’, Revue d'Histoire Economique et Sociale, I. (1972): I, 42–73; II, 357–397.

40 Idem, ‘En guise de conclusion’, Tegdaoust III, 393–395.

41 Not only the working of metal but the mining of copper may have been important in the region's early history: Vanacker, , Tegdaoust II, 173–175Google Scholar; see above, n. 33.

42 The most recent summary of Mauritania's history is in Gerteiny, Alfred G., Historical Dictionary of Mauritania (London and New Jersey, 1981).Google Scholar But the most balanced and thoughtful account of the medieval period is still La Chapelle, ‘Sahara occidental’, 57–70. See also Stewart, C.C. with Stewart, E.K., Islam and Social Order in Mauritania (Oxford, 1973), 12–16,Google Scholar and his footnotes for further references. On Shur Bubba see, in addition, Curtin, Philip D., ‘Jihad in West Africa: early phases and interrelations in Mauritania and Senegal’, J. Afr. Hist., XII, i (1971), 11–24CrossRefGoogle Scholar; Norris, H.T., ‘Znaga Islam during the seventeenth and eighteenth centuries’, Bulletin of S.O.A.S., XXXII (1969), 496–526Google Scholar; Desiré-Villumen, G. M., ‘La Guerre de Char Boubha’, Le Saharien, LVI (1960), 20–25.Google Scholar

43 The question of conflict and its role in nomadic societies is addressed in a general sense in Pierre Bonte's stimulating article ‘La Guerre dans les sociétés d'éleveurs nomades’, Etudes sur les Sociétés de Pasteurs Nomades no. 133 (Les Cahiers du Centre d'êtudes et de recherches marxistes, 1977), 42–67 (there are some problems, however, with his interpretation of the early Mauritanian historical data, pp. 58–60, in contrast to my analysis below). Of similar interest are the ideas of Sweet, L. E. in ‘Camel raiding of North Arabian Bedouin’, where raiding is seen as performing several functions (social, political, economic) which tend to equalize resource distribution and are critical to the reproduction of the tribe.Google Scholar For examples in the context of the western and southern Sahara, see Corpus (Ibn Ḥawqal), 48, 71–73; (al-Bakrī), 73; (Ibn al-Athīr), 160; (Ibn 'Idhārī), 222, and Ibn Khaldūn, cited in La Chapelle, ‘Sahara occidental’, 47, 58, 60–62; Norris, , Myth and Saga, 77–80, 92Google Scholar; Fernandes, Valentim, Description de la Côte d'Afrique de Ceuta au Sénégal (1506–1507), ed. De Cenival, P. and Th., Monod (Paris, 1938), 69–73, 91–93.Google Scholar Fernandes offers us a glimpse of medieval transhumance wherein 300–500 men annually ‘followed the rains’, sending two men ahead to look for promising pasture, leaving one to ‘take possession’ of places where it had recently rained, while the other returned to fetch the tribe. ‘Sometimes’, he noted ‘[two tribes] approach the same place where it has just rained from different sides; and ultimately enter into battle [over it]’. On the impact of drought on tribal and herd structure: Sweet, ‘Camel raiding”, 1135–1139; Dahl and Hjort, Having Herds, 114–129. Conflict generated by seasonal moves to pasture, water and to date and grain harvests was frequently commented upon in early colonial reports on the region, for example, Rapports Politiques et Militaires, Archives Nationales, R.I.M. (Nouakchott), E2–103, 110. On the role of drought, famine and other ecological and climatic problems in tribal divisions, see Norris, , Myth and Saga, 198–199, also his The Tuareg (Westminster, 1975), 76, 84, 100, 105.Google Scholar A case in point is the history of the multi-branched Kunta tribe found from the Mauritanian Adrar–Tagant to the Malien Azawad to Southern Algeria: see Whitcomb, ‘New evidence’.

44 Chapelle, La, ‘Sahara occidental’, 58–60.Google Scholar

45 Ibid., 60; Bonte, ‘La guerre’, 42, 50. La Chapelle points out that in the fourteenth and fifteenth centuries the Ṣanhāja were blocked from northern pastures by the Ma'qil Arabs (‘Sahara occidental’, 67, 68).

47 Ibid., 92, 117, n.C; Corpus (al-Tādilt), 179, as cited by Conrad and Fisher, ‘The conquest that never was’, 31. This relationship between those seeking relief from drought and a willingness to follow an Islamic teacher is paralleled in al-Bakī's account of the introduction of Islam to the Sudanese kingdom of ‘Malaal': Corpus, 82–3; repeated in a later account, pp. 368, 369).

48 Corpus (Ibn al-Athīr), 160; La Chapelle, ‘Sahara occidental’, 62–64. He calls references to the Almoravid movement into the Adrar as a jihād a ‘deformation of history’, suggesting that the first raids were little different from those the Znāga periodically made into the ‘pre-desert’ zone. This is perhaps an excessively simplistic explanation which goes too far in its attempt to counterbalance the received wisdom. Terrasse has suggested that population growth resulted in over-population: Histoire du Maroc, I (Paris, 1949), 217–18. Each uses Ibn al-Athīr's thirteenth-century account compiled from unidentified sources in which the dates appear to be too late in every case. He has the movement leaving the desert in 1058/9 instead of 1054/5: Corpus, notes, 397.

49 Semonin, , ‘Almoravid Movement”, 59.Google Scholar Al-Bakrī mentions the Almoravids taking their herds for their annual ‘salt-cure’ at the salt-water wells of Agharaf, south of the Adrar, and to their summer pasture in the Amatlūs: Corpus, 67, 70. And Norris comments that ‘the Saharan milieu and the habits of its nomads could in themselves have compelled him to modify and adjust the ideals and practices he had learnt…’: ‘New evidence’, 263.

50 Amilhat, , ‘Petite chronique des Id ou Aich’, Revue des Etudes Islamiques, II (1937), 45.Google Scholar On Gao see Fage, J. D., A History of Africa (London, 1978), 180Google Scholar; and for more detail, Bovill, E. W., ‘The Moorish invasion of the Sudan’, Part I, J. Afr. Soc., XXVI (Apr. 1927), 255–256.Google Scholar These considerations tend to support recent scepticism about the protracted jihād carried out against Ghana in the 1070s and explain why the Almoravids continued to prefer their desert habitat. See Conrad and Fisher, ‘The conquest that never was’ for discussion and further references on the question of the Almoravids’ role in the history of the Ghana empire.

51 b, Abū Bakr. ‘Umar al-Lamtūnī was the successor to Ibn Yāsīn as leader of the Almoravids; he died in 1087.Google Scholar For an account of his role in the history of the Almoravids and the Sudan, see Levtzion, , Ghana and Mali, 30–46.Google Scholar

52 Norris, , Myth and Saga, 153–154Google Scholar; Mauny, R., ‘Notes d'histoire et d'archéologie sur Azougui, Chinguetti et Ouadane”, Bulletin de l'I.F.A.N., sér B, XVII (1955), 142–162, esp. 148Google Scholar; Amilhat, , ‘Petite chronique’, 45–48.Google Scholar

53 Norris, , ‘New evidence’, 263Google Scholar; Semonin, , ‘The Almoravid Movement’, 45.Google Scholar The mujāhidūn were those who fought in the holy war or jihād against non-Muslims or ‘bad‘ Muslims.

54 Norris, , Myth and Saga, 85–86.Google Scholar

55 Ibid., 80. Some traditions argue that the Bāfūr were chased from the Adrar; others that they were made tributaries. Given that the Almoravids needed a workforce, and that several tributary tribes of the Adrar which later emerged trace their origins to the Bāfūr, the later seems more likely. See Norris, Myth and Saga, 152–6, and Huguet, ‘Populations primitives… gîtes bafor’. Many of the tributaries or freed slaves reputedly of Bāfūr origin were ‘very rich’ in colonial times: pp. 124–125.

56 Cited in Norris, Myth and Saga, 85–86.Google Scholar

57 Corpus (al-Idrīsī), 127; (Yāqūt), 173; Norris, Myth and Saga, 80. Their use in Saharan warfare is illustrated in Ibn Ḥawqal's accounts of Znāga warfare, Corpus, 48, and those of al-Bakrī, Ibid., 68–69, 72–73. See also Farias, , ‘The Almoravids’, 810–813Google Scholar; Norris, , Myth and Saga, 105.Google Scholar For scattered references to Saharan medieval warfare see Corpus, 22, 28, 95, 98, 129, 134, 164, 165, 173, 179, 180, 185, 220–2. Unfortunately, we know nothing of the identity of these town-dwelling artisans, nor of their relations with the ‘men of the tents’ they supplied. In the nineteenth century such artisans seem to have been few in number and limited in talent. Al-Wasīṭ reports that they formed a sort of caste who paid tribute to warrior tribes. The women worked in leather and the caste was endogamous: Ahmed Lamine ech Chenquiti, El Wasīt, Etudes Mauritaniennes No. 5 (Sénégal, 1953), 117–118.

58 References to these extraordinary shields, some four-and-a-half to five-and-a-half feet long and seven-and-a-half feet in diameter, are numerous; see especially Corpus, 127, 134, 173; on the lamṭ near Awdaghust, Ibid., 69.

59 Amilhat, , ‘Petite chronique’, 46–47.Google Scholar

60 Ibid., 47. For a discussion of zawāyā; or Zwāya, see below, pp. 18–19, also footnote 61.

61 Ibid., 49; Levtzion, , Cambridge History of Africa, II, 654, 662.Google Scholar Moorish tradition has the Almoravid community organized into al-mujāhidūn, warriors of Islam; al-zwāya or zawāyā;, teachers and jurists, and al-laḥma, herders, cultivators and traders: see Norris, H. T., Shinqītī folk literature and song (Oxford, 1968), 21.Google Scholar

62 See Bonte, ‘La Guerre’, his footnotes and especially those to his studies of the Kel Gress. Studies of the Tuareg in general are revealing of this pattern. See for example Nicholaisen, ‘The Pastoral Tuareg’, 393–446, and Norris, The Tuaregs; cf. also Sweet, ‘Camel raiding’, and Equipe écologie et anthropologie des sociétés pastorales, Pastoral Production and Society (Cambridge, 1979), esp. Bonte, , ‘Segmentarity, social classes and power’, 171–199.Google Scholar

63 I would agree in large part with Levtzion's argument that the Almoravids ‘did not reconstruct a tribal confederation which had existed in the past but introduced a new element into the segmentary politics of the Sahara. This politico-religious movement under the combined leadership of the spiritual authority and the tribal amir mobilized the resources of the Sanhāja [Znāga] and made them rulers of the Maghrib and Spain’: Cambridge History of Africa, II, 654. But I would take issue with his suggestion that this took place because of the introduction of a new element into the Saharan system. It seems to me that it was the exigencies of existing Saharan politics and kinship structures, and the nature and distribution of Znāga resources which shaped the initial movement. Norris's account of the early history of the movement and the restraints placed on it by a leader conscious of the need to promote common interests and suppress dissension among the local Znāga also tends to support this analysis: Norris, , Myth and Saga, 92–94.Google Scholar

64 Sweet, , ‘Camel raiding’, 1135Google Scholar; Bonte, , ‘La Guerre’, 49, 50.Google Scholar In other words, from a situation in which the art of war was merely one aspect of pastoralism, Saharan society was slowly moving towards one in which war was necessary for its very survival and reproduction. Hence the needs of war began to shape the character of pastoralism. For an insightful analysis of this process in the eighteenth-century Sahelian society of the Segu Bambara see Roberts, Richard, ‘The Maraka and the economy of the Middle Niger Valley: 1790–1908’ (Ph.D. thesis, University of Toronto, 1978), 28–38.Google Scholar

65 Bonte, , ‘La Guerre’, 45–49.Google Scholar

66 Corpus (al-Bakrī), 71–72; Levtzion, , Cambridge History of Africa, II, 662Google Scholar; Norris, , Myth and Saga, 152–154.Google Scholar

67 Bonte, , ‘La Guerre’, 49, 50.Google Scholar

68 Whitcomb notes that the Portuguese rendering of the name, Azenèque, was actually closer to the indigenous pronounciation than the Arabs' Ṣanhāja: ‘New evidence’, I, 114, n. I.

69 Fernandes described these nomad hunters who lived among the Arabs as ‘a sort of caste, in a sense a profession and not a race’. De Cenival and Monod discuss the identities of Azeneques (whom they consider part of the great Ṣanhājan movement), Barbaros and Ziguis in the context of other fifteenth- and sixteenth-century accounts: Fernandes, Description, 73, 150–2, n. 140.

70 Fernandes, , Description, 53–79, 91–9, 117, 121.Google Scholar See especially the notes which are not only explanatory but introduce comparative material which confirms the highly stratified and specialized nature of this society: pp. 142–156 passim.

71 Ibid., 71, 95.

72 Mosto, Ca Da, The Voyages of Cadamosto, trans. Crone, G. R. (London, 1937), 19, 26.Google Scholar

73 Fernandes, , Description, 81, 121Google Scholar: see also notes, 150, 1.

74 Amilhat, , ‘Petite chronique’, 48.Google Scholar This phrase succinctly captures the transitional mood of the period and a very important stage in the evolution of ḥassānīya (warriors) and zawāyā; (clerics).

76 A qāḍī; is ‘a judge who, according to Muslim Law, has to decide all cases involving questions of criminal and civil law. In the Sahara the title has occasionally carried with it certain other administrative and sometimes military duties’ (as in this case): Norris, , Myth and Saga, 228, n. 33.Google Scholar

76 Norris, , Myth and Saga, 190–191 and n. 4.Google Scholar

77 For a realistic discussion of the socio-economic structure of the Mauritanian zawāyā;, see Stewart, , Islam and Social Order, 58–65,Google Scholar and Miské, Ahmed-Bâba, Al-Wasīt: Tableau de la Mauritanie au début du xx^e siècle (Paris, 1970), 93–104.Google Scholar

78 Fernandes, , Description, 71.Google Scholar

79 This is not inconsistent with findings elsewhere. See for example Sweet, ‘Camel raiding’, 1134–1138, and Fredrik Earth's interesting discussion of pastoral economic growth, ‘A general perspective on nomad-sedentary relations in the Middle East’, in Nelson, C. (ed.) The Desert and the Sown (Berkeley, 1973), 11–22.Google Scholar Barth's well-known study of the Basseri in South Persia best illustrates the kind of limitations pastoralists can face without the means to find sources of labour and forms of investment outside the pastoral sector narrowly defined: Barth, F., Nomads of South Persia (Boston, 1961), esp. ch. VIII, ‘Economic processes’.Google Scholar

80 For Sijilmāsa see Corpus, 65, 66; for Hunwick, Timbuktu J. O., ‘Gao and the Almoravids: a hypothesis’, in Swarz, B. K. and Dumett, R. E. (eds.), West African Culture Dynamics (The Hague, 1980), 426Google Scholar, and Herbert, Eugenia, ‘Timbuktu: a case study of the role of legend in history’Google Scholar, Ibid., 431–422; for Azuggi, Norris, Myth and Saga, 85, Corpus (Ibn Sa'īd), 191.Google Scholar Walata developed on the site of ‘Biru’, a Soninke word meaning ‘temporary’ or ‘provisional place’, undoubtedly a seasonal camp for pastoralists before it became a major market. (Personal communication, Abdoulaye Bathily.)

81 I think further reflexion and research on Tegdaoust findings, especially metal working, will soon require that this statement be qualified and that inter-regional copper and brass trade will also prove to be exceptions to the rule.

82 For a discussion of the different salts available and the needs of the Sudan see my ‘The Ijil Salt Industry: its role in the precolonial history of the Western Sudan’ (Ph.D. thesis, University of Birmingham, 1980), especially 199–225.

83 Corpus (Ibn Ḥawqal), 46–7; (al-Bakrī), 81; see McDougall, , ‘The Sahara reconsidered’.Google Scholar

84 Corpus (al-Bakrī), 57; Cuoq, , Recueil des sources (al-Muhallabī), 77–78.Google Scholar

85 El-Chennafi, Mohammed, ‘Traces d'Awdaghust’, 101–103.Google Scholar

86 Fernandes, , Description, 61, 78–85, 114–115, notes, 142–144, 166–167Google Scholar; Mosto, Ca Da, Voyages, 16–21Google Scholar; Gomes, Diogo, ‘The voyage of Diogo Gomes’, in Mosto, Ca Da, Voyages, 91–102Google Scholar; Pereira, Duarte Pacheco, Esmeraldo de Situ Orbis, ed. and trans. Kimble, G. H. T. (London, 1937), 75.Google Scholar

87 Corpus (al-Idrīsī), 106, 107. He describes Awlīl as supplying salt to ‘all the towns of the Sudan’ but by way of the Senegal River, rather than overland via Awdaghust. Ibn Sa'īd's mid-thirteenth-century mention of Awlīl appears to be nothing more than al-Idrīsī, repeated: Ibid., 184.

88 Corpus (Ibn Baṭṭūṭa), 282–285.

89 Fernandes, , Description, 76–79Google Scholar; McDougall, , ‘The Ijil Salt Industry’, 75–77.Google Scholar Two Znāga kings were said to control Ijil. Given that Wadan and Tinigi were both major centres of Ijil's trade and the Idaw al-Hajj and Tajakant were powerful Znāga tribes, the hypothesis is not unlikely. Nevertheless, we should not exclude the possibility that there were still other salt sources being exploited. There are some puzzling aspects of Fernandes' account and inconsistencies in his, Ca da Mosto's and Pacheco Pereira's descriptions of the number of salt bars a camel carried, which leaves room for question. See McDougall, , op. cit., 62–63, 67–69Google Scholar; Mosto, Ca Da, Voyages, 22Google Scholar; Levtzion, , Ghana and Mali, 172.Google Scholar

90 Fernandes, , Description, 82, 3.Google Scholar

91 Ibid., 70–73; quotations 94–95.

92 Ibid., 70–71.

93 Ibid., 94–95.

94 Roberts explores a comparable tension in the context of the Segu Bambara State in the Middle Niger Valley: ‘The Maraka’, 44–52. Meillassoux has suggested a much broader relevance in the introduction to his The Development of Indigenous Trade and Markets in West Africa (London, 1971), 54–55. It would seem a logical extrapolation from the long-recognized ambiguity of the relationship between trade and war: while war can disrupt trade patterns, as often as not it is used to protect (or simply acquire) them; while war invariably destroys much in its path which might otherwise be traded, it can also provide booty (both human and inanimate) which, in turn, becomes marketable. Meillassoux develops the ‘warrior-trader’ theme and the implications of their relationship for early Sahelian political evolution in his ‘The role of slavery in the economic and social history of Sahelo-Sudanic Africa’, in Inikori, J. E. (ed.), Forced Migration: the Impact of the Export Slave Trade on African Societies (London, 1982), 74–99.Google Scholar

95 The complex relationship between the formation of ethnic boundaries, the structuring of social (class) relations and the emergence of occupational specialization are explored by Fredrik Barth in ‘Nomad-sedentary relations’ and in Barth, F. (ed.), Ethnic Groups and Boundaries (Bergen, 1969), 9–38.Google Scholar His analysis focuses primarily on the dichotomized ethnic status of nomads and sedentaries, but much is relevant to emerging ethnic differences even among nomads. Two points of discussion are especially pertinent here. With respect to how and why people become associated with a particular group (and how and why they change ethnicities) he notes that migration and conquest play only an intermittent role, and that other processes are often more critical. (Economic factors are not infrequently involved, as his own studies have shown.) Secondly, keeping in mind ecology and available resources, he notes that when one ethnic group has control of the means of production utilized by another group, a ‘relationship of inequality and stratification obtains’. It can be inferred, then, that the stratification (as distinct from the dichotomy) at least initially has a causal base separate from the characteristics associated with the ethnicities involved, and that it can change fundamentally if the second group finds independent access to resources formally controlled by another: Ethnic groups and boundaries, 21–27.

96 Meillassoux has recently constructed a view of socio-political development for this same period which is similar in analysis and conclusions with respect to emphasis on the emergence of servile labour, a merchant class, and the professional warrior; see ‘The role of slavery’. But, pp. 76–85, he takes the existence of slavery and its requirements for reproduction as the key factor in an analysis which leads him to identify different stages of development, and a different chronology, from those suggested here. While I am indebted to Meillassoux's stimulating ideas in this important article, it seems to me that this emphasis upon slavery as the chief consideration is excessive, and leads to an interpretation which, among other things, simply does not fit the chronology of the region's economic and urban history.

97 On the importance of transferability of surplus and the inherent limitations of pastoral capital see Earth, , ‘Nomad-sedentary relations’, 11–22.Google Scholar It is likely that other commodities (above all ceramics and metals) also contributed to this evolution, though probably to a lesser extent. (See note 81, above.)

98 Corpus (Ibn Ḥawqal), 48; (al-Bakrī), 68. ‘Tents’ in this context probably refers to a social unit as in the modern usage in which it represents a man, his wife, servants and children: Stewart, , Islam and Social Order, xvi.Google Scholar

99 Robert-Chaleix, , ‘Tegdaoust V’, chaps. VII and IX on ceramics and metals respectively; Vanacker, Tegdaoust II, 171–176.Google Scholar

100 Levtzion, , Ghana and Mali, 6–7.Google Scholar

101 Devisse, , ‘Conclusion générate’, Tegdaoust III, 554.Google Scholar

102 In fact, this continuity is stressed by Devisse, Robert-Chaleix and Vanacker.

103 Robert-Chaleix, , ‘Tegdaoust V’, 25.Google Scholar

104 See above, p. 12; the population estimate is from Mauny, Tableau géographique, 482.

106 Robert-Chaleix, , ‘Tegdaoust V’, 282.Google Scholar

106 Corpus (al-Bakrī), 73–74.

107 el-Chennafi, ‘Traces d'Awdaghust’, 105; Meillassoux, ‘A propos de deux groupes Azer’, Tegdaoust III, 528, n. 30.

108 The connexion comes through a group called the ‘Tafanko’ or ‘Tafaranko’, a name sometimes given to a certain fraction of the Guiriganke, said to be former slaves (haratin) who at one time specialized in weaving and had as their slaves the maxanbinnu. (Meillassoux, ‘A propos de deux groupes Azer’, 526–527. The tradition linking the maxanbinnu to the Tafaranko (and hence, the Guiriganke/Tegdawest), comes from Monteil, Ch., ‘Notes sur le Tarikh es-Soudan’, Bulletin de l'I.F.A.N., sér. B, XXVII (1965), 492.Google Scholar Meillassoux notes that he found no evidence confirming this tradition (p. 528, n. 24), whereas el-Chennafi draws an even closer connexion by identifying the name ‘Tafaran’ with Tegdawest (‘Traces d'Awdaghust’, 105).

108 The only direct discussion of this question is presented by Robert-Chaleix, , ‘Fusaïoles’, Tegdaoust III, 510–513.Google Scholar On the basis of the diffusion of certain types of spindle-weights she postulates a migration of the Tegdawest from the Tishit to the Awdaghust region in the fifteenth century. El-Chennafi on the other hand implies a much earlier occupation, though he specifies no date. One might also argue that the evidence of ‘fusaïoles’ links the migration to the Tafaranko, rather than the Tegdawest per se; hence, it does not preclude an earlier presence of Guiriganke/Tegdawest clans in the Awdaghust area.

110 Meillassoux, , ‘A propos de deux groupes Azer’, 526, n. 8.Google Scholar

111 Robert-Chaleix, , ‘Tegdaoust V’, 281Google Scholar, discusses the significance of Tegdaoust's position in the Hodh, surrounded by natural defences in the Tagant and Dhar Tishit–Walata to the west, north and east, but noticeably open to the south.

112 Mosto, Ca Da, Voyages, 17–19.Google Scholar

113 Corpus, 49.

114 Meillassoux, , ‘A propos de deux groupes Azer’, 228–230.Google Scholar The Masna organized the exploitation of the Tishit salt sebkha which produces an inferior earth salt called amērsal; according to one tradition they used maxanbinnu labour.

115 Robert-Chaleix, , ‘Tegdaoust V’, 284.Google Scholar She argues that in the face of existing physical evidence the damage from excessive rains in the latter part of the century caused more disruption than can be attributed to the Almoravid raid.

116 Levtzion, , Ghana and Mali, 137–141Google Scholar; Devisse, , ‘Question d'Awdaghust’, 116.Google Scholar Farias suggests that the development of Azuggi (in the Adrar) and the old ‘ṬTrīq Lamtūnī’ provided the necessary foundations for this thriving commerce. Recently undertaken archaeological work at Azuggi may provide us with a better understanding of this important stage in Mauritania's economic history: see De Moraes Farias, Paulo, ‘The Trīq Lamtūnī (XIth–XIIth c.) a trade route?’, paper presented at the Centre of African Studies, School of Oriental and African Studies, London, December, 1970Google Scholar, and Robert-Chaleix's bibliography ‘Recherches Archaeologiques’ (forthcoming), references to Azuggi. This view of the Almoravids as encouraging commercial growth contrasts with a more traditional view that their movement disrupted western Saharan trade, as in La Chapelle, ‘Sahara occidental’, 70.

117 Robert-Chaleix, , ‘Tegdaoust V’, chart of Awdaghust's OccupationsGoogle Scholar; also Devisse, , ‘Conclusions générales’, Tegdaoust III, 554–556.Google Scholar

118 This argument is consistent with an interesting suggestion Devisse has made that Awdaghust probably retained an important economic position as a centre for raising transport camels: ‘Routes de commerce’, 60–61; idem, ‘Question d'Awdaghust’, 155, 156. Robert-Chaleix's analysis emphasizes Awdaghust's continuing role as a centre for the inter-regional trade in salt and cotton, though at a somewhat later date: ‘Fusaïoles’, 510, 511. This still leaves us with al-Idrīsī's puzzling information about Awdaghust's twelfth-century decline. One possibility is that his information actually dates to a period shortly after the Almoravid incursion when the departure of North African merchants may have momentarily disrupted trade. It is also possible that problems with water supplies or short-term drought periodically affected the town, and that al-Idrīsī is describing a temporary phase.

119 Azer or ‘Azayr’ is a Soninke dialect ‘more or less contaminated by Berber’ which is almost extinct in present-day Mauritania. It seems to have been closely associated with the commercial development of the desert-edge and was widely spoken in the south-western Sahara until the fifteenth and sixteenth centuries, when it gave way to the languages of the Znāga and the newly arrived Banū Ḥassān Arabs. Norris describes it more generally as a ‘Sudanic… perhaps autochthonous language spoken widely in parts of the Western Sahara when its southern regions were subject to Ghāna, Māli and Takrūr’. It survives in various place names like Shinqītī derived from Azer shi-n-gede, ‘the horses' springs’ and certain loan words. See Monteil, Ch., ‘La Langue Azer’ in Th., Monod (ed.), Contributions à l'étude du Sahara Occidental (Paris, 1938), 215–221Google Scholar; Norris, H. T., Shinqītī Folk Literature, 31.Google Scholar

120 El-Chennafi, , ‘Traces d'Awdaghust’, 101–105.Google Scholar

121 Ibid. Monteil refers to Azer as an ‘intertribal commercial language, commonly spoken in the famous empires of Awdaghust and Ghana’: ‘La Langue Azer’, 215. He notes that one source (Delafosse) says Azer was spoken more recently at Arawan and Tawdenni: Ibid., 220. Though this has not been confirmed by further research it would, if true, underscore the conclusion of this article.

122 Corpus (Ibn Ḥawqal), 48; (al-Bakrī), 69. See also Levtzion, analysis, Ghana and Mali, 31.Google Scholar Devisse's interpretation of Ibn Ḥawqal's account differs: ‘Question d'Awdaghust’, 120.

123 El-Chennafi, , ‘Traces d'Awdaghust’, 103–106.Google Scholar

124 Ibid., 105–106.

125 For the operation of the trade in 1506–1507 see Fernandes, , Description, 82–83Google Scholar; merchants from Walata could also travel to Tishit to buy Ijil salt at lower prices. Levtzion's argument that Awdaghust was replaced by Walata stems from his focus on the trans-Saharan trade, and their respective roles as termini: Ghana and Mali, 147. Those who look to Wadan, on the other hand, are referring to their roles in the regional economy. See for example Chapelle, La, ‘Sahara occidental’, 70Google Scholar; Amilhat, , ‘Petite chronique’, 59.Google Scholar Fernandes' and Ca da Mosto's descriptions of Wadan make it clear that the Adrar centre had usurped Awdaghust's earlier role in the salt, slave and gold trade: Fernandes, , Description, 82–85Google Scholar; Mosto, Ca Da, Voyages, 17–25.Google Scholar El-Chennafi attributes the declining fortunes of Tegdaoust's merchants and their market to the twelfth-century development of the new caravan cities in general (including Wadan): ‘Traces d'Awdaghust’, 107.

126 Toupet, , ‘Sédentarisation des nomades’, 162Google Scholar; Robert, , ‘Fouilles de Tegdaoust’, 480Google Scholar; Devisse, , ‘En guise de conclusion’, Tegdaoust III, 394–395.Google Scholar It is significant that the most recent analysis of the decline of Ghana using local oral traditions arrives at a similar conclusion about the impact of long-term drying conditions in the region. See part II of Conrad, and Fisher, ‘The conquest that never was:…the local oral sources’, History in Africa, X (1983), 53–78.CrossRefGoogle Scholar

127 See above, n. 27. Robert-Chaleix comments that the wells must have been a tremendous attraction for neighbouring nomads. But as they began to dry up, claims to property rights (based on control over water points) may have been eroded, which would help explain the complete disappearance of enclosures by the fifteenth century: ‘Tegdaoust V’, 192.

128 Forgue, Pierre La, ‘Notes sur Aoudaghost’, Bulletin de la Société de Géographic et d'Archéologie de la Province d'Oran, LXIV (1943), 32–33.Google Scholar

129 El-Chennafi, , ‘Traces d'Awdaghust’, 102.Google Scholar

130 Their inability to move in search of supplies left them prey to hunger and thirst, and above all to disease. Exactions by masters, also ‘in need’, often became excessive: McDougall, , ‘The Ijil Salt Industry’, 321.Google Scholar In the central Sudan, among the Tuareg, the phenomenon has been observed in several manifestations. See Baier, Stephen, An Economic History of Central Niger (Oxford, 1980), 32–35.Google Scholar

131 El-Chennafi, , ‘Traces d'Awdaghust’, 102.Google Scholar

132 Ibid.