TRANSLATOR’S INTRODUCTION

SONGHAY: AN INTERPRETIVE ESSAY

The Physical Milieu

The history of Timbuktu and the Songhay empire is embedded in thebroader history of the Middle Niger, and is intimately linked to theexistence of that river and its regime. The Middle Niger may beroughly defined as that segment of the river which begins when theNiger takes a sharp northward turn around the present city of Segu,and ends approximately where the river enters the modern state ofNigeria. This great arc is often referred to as the Niger Bend, or inFrench la boucle du Niger. For most of this distance—close to onethousand miles—it is navigable by shallow-draft boats and canoes,except during the low-water season at Labbezenga on the easternreaches of the great bend in the river at 15° N, where rapids impedeprogress. Throughout this long stretch the river is flowing througharid or semi-arid lands, where average annual rainfall at the twentiethcentury has varied between 750 mm. (30 in.) in the southern reaches,and less than 250 mm. (10 in.) in the northerly areas, mostly fallingbetween June and September.1 Timbuktu averages 9.1 in. (231 mm.)of rainfall, mainly in July and August, the three preceding monthsbeing the hottest season of the year, with daily high temperaturesaveraging between 41° C. (105° F.) and 43.5° C. (110° F.). Gaoscarcely differs.The vegetation produced by such a rainfall and temperature regime iseither of the savannah or semi-desert type. In the semi-desert region,where the soil is lateritic, often overlaid with sand, trees are few andfar between, except along the dry river-beds. Elsewhere some thornyacacias and fan palms (Songhay: kangow, Latin: Hyphæne thebaica)may be found in favoured spots; otherwise the vegetation consistsprincipally of short spreading bushes, suitable for goats and camels tobrowse on. During the short rainy season grasses rapidly spring up,and some of these provide seeds that are eaten in a lean season (e.g.cramcram—Cenchrus biflorus, and askanit or haskanit—Cenchrus

1 It must be remembered that these are only averages. The annual variation is considerable,and as Rouch (1953b), 135-9, points out, ‘The Songhay region suffers from a very unstablerainfall pattern, pitilessly alternating years of famine with years of abundance’.

echinatus). In the savannah lands a wider variety of trees may befound, including the baobab (Adansonia digitata), and the shea-butter tree (Butyrospermum parkii; Songhay: karite).2 Bushes arelarger and may form thickets, while grasses are more varied and moreabundant. In the savannah regions lion, hyena, giraffe and gazelle areto be found, and migratory elephants still exist in the lacustrineregion, moving to south of Hombori during the driest season of theyear. In the south-east of the region (Dendi) vegetation is the mostluxuriant, and the humidity allows the tsetse fly to breed in someareas, creating a danger to horses and cattle. It is also in this regionthat the caïlcédrat tree (Khaya senegalensis Juss.) grows, providingwood for the building of boats.3 Most of the inner lands of the NigerBend experience a temporary respite during and after the rainyseason when streams or small rivers are briefly in spate, and pools fillup. Some of these pools last through the dry season, and water maybe obtained by shallow digging in dry wadis or from regular wells.In such an inhospitable climatic zone, the river is truly a lifeline,providing year-round drinking water for humans and their domesticanimals, and pasture, not only at the water’s edge, but in the veryriver itself where a plant known as burgu (Echinochloa stagnina)provides fodder, especially for horses living along the Niger’s banks,but also for the ‘river horse’—the hippopotamus—living in theriver.4 The bulk of the water that flows through the bend in the Nigeroriginates in the highlands of Guinea, the Futa Jallon, where the riverhas its source, and is augmented by other rivers such as the Sankaraniand the Bani that flow into it. In the Futa Jallon, rains begin in lateMarch and continue until late October. By July, the rain that fell inFuta Jallon in March and April is beginning to reach the Segu area.As it flows past Segu it enters a broad flat plain where the river has agradient of only 6 cm. per km. (3.5 in. per mile).5 As a result of thecontinual flood of water over several months, both rivers, the Nigerand the Bani, overflow their banks and a huge Inland Delta isformed, stretching from a little downstream from Segu to close toTimbuktu, a distance of some 300 miles, with a width of up to 150miles. Much of the volume of water the Niger brings down isdeposited in this area, but a still considerable flood pushes on

2 Barth (1965), iii, Ch. LXIV, gives an incidental account of the flora of the lacustrine

region during his journey narrative.3 The caïlcédrat is an evergreen tree, growing to as high as 30 m (90’) under favourable

conditions. It flourishes in areas with 650-1300 mm. of rainfall over seven months, and produces afine-grained wood, easily worked; see Giffard (1974), 260-2.

4 Burgu has a syrupy sap and rhizomes rich in sugar; its seeds, which resemble wild rice,are edible. Burkhill notes: ‘The plant is deemed to be the most useful of all wild plants in theTimbuktu area of Mali, providing as well as food and drink, fodder, thatch, caulking for boats,vegetable salt after calcinating which is used to make soap and indigo dye’; see Burkhill (1985-95), ii, 236-8.

5 See Morgan & Pugh (1969), 252.

eastwards past Timbuktu and through the gorge of Tosay beforeturning abruptly to the south around the present town of Bourem.Downstream from there the Niger does not flood extensively, butduring the high water season from January to March, the river isoften one kilometre (1,000 yards) wide, and there are numerousinlets and creeks.As the floodwaters abate in the Inland Delta, a network of creeks andnarrower waterways reveals itself, and in the northern half of thedelta a large number of lakes, especially to the north-east of the delta,but also many of lesser size to the north-west, and in the far north thegreat Lake Faguibine bordering the desert. This flood regime createsa moist agricultural terrain where grains and vegetables can be grownand cattle pastured, forming a breadbasket for Timbuktu and itsdesert and riverine hinterland. The river Niger and the lacustrine zoneboth support an abundant aquatic life—hippopotami, manatees andmany species of fish. Agriculture is also possible, but on a muchsmaller scale, along the banks of the eastern sector of the MiddleNiger as the water recedes, and in the creek areas. South of latitude15° N, where annual rainfall excedes 500 mm. (20 in.), rainfedagriculture is possible during the short rainy season from July toSeptember, and this, of course, becomes more viable farther south.

The Human Milieu

This huge swathe of river and floodplain was a natural refuge forpopulations responding to the increasing dessication of the Saharaafter 5,000 BC. At first moving into seasonally flooded north-southvalleys, they eventually moved farther south as the Inland Deltaslowly transformed itself from a vast permanent swamp into the moreaccessible floodplain it now is. By 250 BC we have the earliestevidence of urbanisation in the region. Archaeological work in the1970s and 1980s at Jenne-jeno has revealed the site of a complexsociety that developed into a considerable regional centre over thenext millenium, and was ultimately ‘captured’ by the nearby risingtown of Jenne around 1200 AD.6 The first millenium AD wasclimatically a relatively benign period for this region with increasedrainfall allowing for sufficient tree growth in Mema, to the immediatewest of the Inland Delta, for example, to support an iron industrytowards its end.7 This was also the period of the development of thestate of Ancient Ghana, and the period during which the Zuwdynasty of ancient Songhay emerged at Kukiya, later relocating toGao. Since about 1100 AD reduced rainfall and increasing dessi-cation have been the general pattern for the Middle Niger region, and

6 See McIntosh, R.J. & McIntosh, S.K. (1988), for an overview of settlement patterns andmaterial culture in the first millenium AD.

7 Håland (1980).

for West Africa as a whole, with significant long and short termdeviations from the pattern both in terms of increased and decreasedhumidity. The sixteenth and early seventeenth centuries, for example,appear to have been an era of relatively favourable climaticconditions, whereas the period c. 1650-1850 was generally morearid.Our knowledge of the history of the various ethnic groups of theMiddle Niger region is slight, though it is evident that there has beena good deal of stability as regards named ethnic units between thesixteenth century and the present.8 Rather than attempt any historicalreconstruction of specific groups, the following discussion will lookat ethnicity through the lens of environment and occupation. Thepopulations of this region fall into three broad categories, accordingto their economic base, which is in turn determined by the ecologicalniche they inhabit. Along the desert fringes are pastoral nomads;along the river, sedentary hunting and agricultural populations,segments of which engage in inter-regional trade; and on and alongthe river Niger and the lakes, populations that engage in fishing andhippopotamus hunting, and supply water-borne transportation forpersons and goods.Around the year 1000, the pastoral nomads, with an economy basedon camel and sheep rearing, long-distance transportation, and raidingand extortion, were mainly ∑anhja and other Berber groups. By1600 these had been largely overtaken in the west by ˘assniyyaArabs migrating through the western Sahara from southern Morocco,and in the east by Tuareg groups migrating in from the north-east.9

Some clans of the Massüfa ∑anhja, however, became urbanised. IbnBa††ü†a found them in both Walta and Timbuktu in 1352-3, and inthe sixteenth century the two leading scholarly families of Timbuktu,

8 The ethnic make-up of this Middle Niger region is extremely complex, each group having

many sub-groups, and some of these having intermarried with other, different ethnic groups, whilesome may have abandoned their earlier language and most cultural practices yet still regardthemselves as being part of the original group; see Diakhite (1984); Olivier de Sardan (1984). Thefollowing discussion paints in only the broadest of strokes.

9 The actual social processes involved in these transformations are little understood. In thewestern Sahara the Arabic language and some Arab cultural traits were adopted, though Berber(Znaga) survived into the twentieth century. Allowing for many exceptions and anomalies, we cansay that in general the Arab groups regarded themselves as warrior clans to whom the existingBerber groups were subservient. Some of these latter became religious specialists. The situation inthe Azawd, north of the Middle Niger, was different. Both ∑anhja and Tuareg are camelnomads who wear the face veil, and both speak dialects of Berber (Znga and Tamasheqrespectively), and hence the cultural metamorphoses cannot have been very great. As will be seenfrom Ch. 8 below, al-Safidı, no doubt for this very reason, conflates the ∑anhja and Tuareg. Weshould also recall that early Arab writers such as Ibn ˘awqal and al-Bakrı mention Berber groupsalready in the area of the northern Niger bend, the names of some of which appear to be cognatewith those of modern Tuareg groups (e.g. Saghmra / Issekomaren, Yantasir / [Kel] Antasar).

the Aqıt and the Anda Ag-Mu˛ammad were both of ∑anhja origin,and there were ∑anhja scholars at Tagidda (Takedda).10

The other pastoral nomadic people were the Fulani,11 possibly themost discussed ethnic group in West Africa, and now certainly themost dispersed. This is not the place to enter into a discussion of thevarious theories concerning their origin, or the complex pattern oftheir migrations.12 Those whose history impinges most directly on theMiddle Niger during the period under consideration are the Jallobewho migrated from Termes in northern Kaniaga in the late fourteenthcentury, and the Sangare who followed them. They settled in Msina,and the Jallobe established their authority, being termed sultans ofMsina or Msina-koi by al-Safidı.13 Some of these Fulani migratedand established themselves in other areas of the Middle Niger, whileothers became urbanised religious specialists. TS also has a number ofreferences to people it calls Zaghrnı or Zughrnı. These are to beidentified with the Joghorani (a Soninke term) or Zooran (theSonghay term). They are a client group associated with the Fulani,who call them Diawam∫e.There were, of course, numerous settled agricultural and urbanmercantile or scholarly groups in the Middle Niger in the fifteenthand sixteenth centuries. In the Inland Delta area these were mainly ofMande origin; in the eastern area of the Niger bend they wereSonghay (and perhaps Zerma),14 and neighbouring them in thesouth-east, Hausa; in the lands of the interior of the Niger Bend(broadly what is called gurma in Songhay) were the hill Dogon(Tombo) along the Bandiagara uplands and the plains Dogon(Kumbebe) to the east of them; various groups of Mossi, one ofwhich ruled over a powerful state based on Yatenga; and the Gur-mantche towards the south-eastern reaches of the Middle Niger. Weknow, too, from the writings of A˛mad Bb that various otherethnic entities still existent in the twentieth century had a corporateidentity (though we cannot be entirely sure of their areas of habi-

10 See Mu˛ammad Bello, Infq al-maysür fı ta√rıkh bild al-Takrür, ed. Bahıja al-Shdhilı,

Rabat: Mafihad al-Dirst al-Ifrıqiyya [Jmifiat Mu˛ammad al-Khmis], 63 ff. Bello says that Aïrwas inhabited by Tuareg and remnants of the ∑anhja and südn.

11 This is the term used in Arabic and Songhay, and also the term most familiar in the Englishlanguage. They call themselves Pul or Ful (sing. Pulo, pl. Ful∫e).

12 See Tauxier (1937); Labouret (1955); summarised in R. Cornevin, art. ‘Fulbe’, EI(2), ii,939-43.

13 See below, Ch. 26.14 The Zerma (or Jerma) speak a dialect of Songhay and inhabit a territory between

approximately 12° 30’ and 15° N, mainly on the left bank of the eastern reaches of the Niger andits hinterland; see map in Prost (1956), 15. Their traditions point to their having migrated therefrom the northern part of the Inland Delta, perhaps during the early sixteenth century, but this isonly a small part of the story of the amalgam of populations that have gone to make up the Zerma;see Rouch (1953), 187-8, 208; Hama (1967); Gado (1985); Streicker (1980), 105-26; Olivier deSardan (1982), 406-12.

tation) in the sixteenth century: Bobo, Dagomba, Borgu (i.e. Bariba)and Yoruba.15

The Mande groups in the area consisted of Soninke, Malinke,Bambara (or Bamana) and Dyula. The rulers of Ancient Ghana wereSoninke, and certain Soninke clans were among the earliest peoplesin West Africa to have converted to Islam following contact withBerber and Arab traders from North Africa doing business within theterritories of Ancient Ghana. Soninke appear, in fact, to haveinhabited a broad swathe of territory stretching from the banks of theriver Senegal, through Diara and the Hodh to as far as the banks ofthe Niger.16 There are echoes of a Soninke presence on the banks ofthe Niger in the chiefly greeting tunkara that was still offered in thesixteenth century to the Balmafia (an official at Kabara) and theKurmina-fari (the viceroy in Tindirma), and in the Soninke languagestill used even today by some of the Songhay griots. The chief griotin the days of the askiyas bore the title gesere-dunka, the first elementof which is a Soninke word.17 The first askiya, al-˛jj Mu˛ammad issaid to have been of Soninke stock (of the Sila or Ture clan), whileAskiya Is˛q II had a Soninke nickname, kede bini, meaning ‘blackstone’.The Dyula, or Wangara as they are known in the Middle Niger andHausaland, were described by al-Bakrı in the eleventh century as‘non-Arab südn who conduct the commerce in gold dust betweenthe lands’. They brought gold up from the banks of the Senegalriver, opposite where it was mined in Bambuhu, to metropolitanGhana to be exchanged for North African goods.18 As AncientGhana slowly collapsed and rising Mali established itself as theleading power of the region in the thirteenth century, these Wangaramerchants dispersed too, some among them by now havingapparently become religious specialists.19 The latter seem to have

15 The list comes from his Mifirj al-ßufiüd written in 1613. For an edition and translation of

this work, see John Hunwick & Fatima Harrak (2000). For earlier translations of the Mifirj, seeZeys (1905); Barbour & Jacobs (1985).On ethnicities see also Boulègue & Dramani-Issifou(1989); Devisse (1989).

16 Soninke-speaking populations once lived even farther north to judge from the presence ofa Soninke-based creole called Azayr (or Azer) in towns such as Shinqı† and Tishıt.

17 Tamari (1997), 84, suggests in addition that dunka may be the Soninke wordtunka—‘ruler’. Gesere, written gisiri in Arabic, is pronounced jèsérè in modern Songhay.

18 For the identification of al-Bakrı’s Banü Naghmarta with Wangara, see Hunwick (1981),420-1. These Wangara were undoubtedly Soninke. Later, during the heyday of the Mali empire,as Wangara established themselves as the principal merchants of the empire, they became‘Malianized’, speaking a dialect of Malinke, which, because of their calling as merchants (dyula),became known as Dyula, a name that was applied broadly to the people themselves. These seemto be the broad lines of what was undoubtedly a highly complex cultural transformation.

19 Wilks (1968), has presented the most detailed study of Dyula traditions of Islamiclearning, but he is unable to trace these back effectively before the legendary al-˛jj SlimSuwari, whom he believes lived around the middle of the fifteenth century. Al-˛jj Slim Suwaricertainly seems to have been a seminal figure in the teaching of Mlikı fiqh, to whom mostteaching genealogies revert. But the ultimate source of such a tradition is likely to have beenamong ∑anhja scholars of the Almoravid period, for whom the Mlikı madh’hab was the focal

settled in two towns in Msina: Diakha and Kbara, both mentionedas well-established centres of Islam by Ibn Ba††ü†a. Kbara scholarswere to make a name for themselves in Timbuktu, according to al-Safidı, while Diakha became a point of dispersion for teachers andpreachers of Islam who became known as Diakhanke—those fromDiakha.20 Some also settled in Jenne; the q∂ıs of Jenne mentionedby al-Safidı nearly all have Soninke names, while Mu˛ammadBaghayogho, one of Timbuktu’s greatest scholars, and the shaykh ofA˛mad Bb, was a Wangarı from Jenne. It was also from a base inJenne that Dyula merchants pioneered a trade route down to Beghojust north of the Akan forests of present-day Ghana, where goldbegan to be mined in the fourteenth century.21 Wangara merchantslater pushed into Hausaland, probably during the period of Maliancontrol of the northern segment of the Middle Niger, and by the latefifteenth century had opened up a transverse route from the VoltaBasin to Hausaland, supplying the Hausa states with both gold dustand the much sought-after cola nut.Other Mande groups inhabited large territories to the west and south-west of the Niger Bend and had an important impact on its history.The great Mande empire of Mali, founded in the mid-thirteenthcentury, had conquered the Middle Niger region round to as far asGao by the end of that century, and the Malians were to be thedominant power in the area until early in the fifteenth century.Although by 1438 Mali had abandoned this far-flung province of itsempire, it remained a threat to the southern Inland Delta until the endof the sixteenth century.22 During the sixteenth century, however,when Songhay power was at its height, three major expeditions weremade into Mali, during the second of which the Songhay armyhumiliated the Malians in their royal residence.23 On the other hand,Songhay owed some debt to Mali in the field of imperial admin-istration. We do not know how deep this debt may have been, butSonghay’s retention of Mande titles such as farma, fari, and possiblymondyo, is a sure sign of it.24 The governor of Bara province in the

point of their religious life. If this is so, then the origins of Dyula scholarly activity should be soughtin the twelfth century. Indeed, some sources give the date of al-˛jj Slim’s death as 542/1147-8,which Wilks rejects as too early, based on the teaching genealogies that go back to him. While hemay be formally correct in this, a mid-twelfth century date for the beginnings of Dyula Mlikıscholarship is not inherently unlikely. See further below.

20 See Sanneh (1989).21 See Wilks (1982).22 See below, Ch. 25.23 See below, p.140. It is clear that this royal residence cannot have been as distant as the

alleged ‘capital’ of Mali at Niani-on-Sankarani, just within the borders of the present-dayRepublic of Guinea. The area I have argued for as the royal city visited by Ibn Ba††ü†a, midwaybetween Bamako and Segu would be a location well within the reach of a Songhay expedition; seeHunwick (1973).

24 Balmafia (or perhaps Bal-magha) may perhaps be another such title.

northern Inland Delta, the Bara-koi, continued to use the Manderoyal title mansa.The last Mande element to be considered is the group that al-Safidıcalls the Bambara. This term has been subject to abuse in variousliteratures, including al-Safidı’s work, and it should not always beassumed that it refers to a single readily identifiable group.25 Itsprimary meaning in the writings of Muslims seems to be ‘pagan’, andby extension ‘enslaveable barbarian’; on the other hand, they areseen more threateningly as the archetypal barbarian hordes at thegates of civilization. Thus, following the Safidian invasion ofSonghay in 1591 and the breakdown of central authority, we are toldthat Bambara ravaged the area of Jenne, though those whom al-Safidıenumerated in the barbarian hordes included Fulani, and some part-Songhay warriors. Nevertheless, the chiefs of Kala and Bendugu,former provinces of Mali taken over by Songhay, were probably ofBambara origin, and later, in the eighteenth century, the risingBambara power of Segu was to make inroads into the Inland Delta upto the gates of Timbuktu.Before coming to the Songhay themselves, let us refer briefly to theriverine peoples, two of whom were Manding-speaking, and theother Songhay-speaking. In terms of their occupations and life-style,the Somonou, who live in the southern Inland Delta,26 the Bozo,whose habitat is the northern sector of this region, and the Sorko,whose territory stretches across the northern delta and downstream toas far as Yawri in present-day Nigeria, are very similar. TheSomonou and Bozo are fisherfolk who speak a Manding language;according to Rouch, they fish at different seasons anywhere betweenthe rapids of Kolikoro and those of Labbezenga.27 The Sorko areSonghay-speaking masters of the water, monopolising river trans-portation, and living from hippopotamus hunting and spear fishingover the entire northern and eastern sectors of the Middle Niger.28

They are doubtless amongst the earliest inhabitants of the MiddleNiger and originate in the far south-east; their place of retreat is alonga tributary of the river Niger, the Gulbin Kebbi, where they wouldhave been able to find trees sufficiently large to build boats from.29

During the period of the Songhay empire, the Sorko were considereda socially inferior group, who were ‘owned’ by the askiyas; that is to

25 See Bazin (1985).26 On the Somonou, see Tamari (1997), 216-9.

27 See Rouch (1950), 127.28 See Rouch (1951 & 1954).29 In this Hausa-speaking environment they are known as Sorkawa. Niger rivercraft are

made in two ways: by hollowing out a tree trunk, a method only useable for canoes; and sewingwooden planks together and caulking them, which permits much larger boats to be constructed. Onthe Sorko of Kebbi, see Harris (1942).

say they had the absolute obligation of service to them at all times.The TF describes them, along with another group known as the Arbi,as mamlüks (‘possessions, slaves’) of the askiya.30 For how long thishad been their status we do not know, but control of the Sorko wasnecessary for control over the river Niger, which was the economicand strategic lifeline of Songhay.31 On the one hand Sorko boatswould have served to transport rice to Gao from the royal plantationson the river Niger; on the other hand, their large boats called kantawere ready in the harbour at Gao to transport the askiya and hishousehold to the other bank of the Niger after their defeat byJawdr’s forces in 1591. They were no doubt also called into servicefor ferrying foot-soldiers and transporting royal messengers from onepart of the empire to the other. The kanta boats were probably namedafter the kingdom of Kebbi (known to the Songhay as Kanta after thetitle of its ruler), homeland of the Sorko where the boats were built.Apparent confirmation of this comes from the letter of Müly A˛madal-Manßür of Morocco addressed to Kanta Dwüd of Kebbi, in whichhe asks him ‘to hand over [to us] the whole of the annual quota ofboats which you used to give to Askiya and to continue to performtheir necessary duties’, i.e. to provide their crews, who would havebeen Sorko.32

The Sorko were not the only group that had open-ended obligationstowards the rulers. Although we should be cautious about acceptingthe full list of servile peoples (‘groups in thrall to the ruler, not freepersons enslaved’—qab√il ariqq√ lahu l a˛rr usturiqqü) listed in(pseudo-) TF,33—one group, the Arbi mentioned above, ismentioned in what is evidently part of the original text of TF. Boththey and the Sorko may have been those referred to as ‘slaves of theruling institution’, or fiabıd al-sal†ana, in the phraseology of al-Maghılı’s replies to the questions of Askiya al-˛jj Mu˛ammad.34

The askiya had asked whether it was lawful for him to retain suchgroups in a sort of perpetual slavery, and he was told that suchpersons might be thought of as an endowment (˛ubus) which it waspermissible to hold and inherit.35 The Arbi are described in TF aspalace slaves, bodyguards and grain producers for the askiyas. Theterm ar bi simply means ‘black man’ in Songhay, and it may be that

30 See Hunwick (1968 & 1996b). The Sorko are also called ‘Zanj’ elsewhere in TF. On the

term Arbi, meaning ‘black man’ (ar bi) in Songhay, see below.31 Seku A˛madu, founder of the Islamic state of ˘amdallhi in Msina in the early

nineteenth century tried to revive such claims. As late as 1873, ‘Zanj’ (i.e. Sorko) were still beingsold as slaves; see Hunwick (1996b).

32 See below, ‘Other Contemporary Documents’, 2 (v).33 That is those mentioned only in MS C, which is a nineteenth-century re-writing of the TF;

see Levtzion (1971). They may, however, in part reflect earlier social realities. On the servilegroups, see especially TF, 57 (trans. 111).

34 Hunwick (1985a), 86, 88.35 See Hunwick (1985a), 104-6, & Hunwick (1996).

the Arbi were descendants of some authochtonous group living alongthe banks of the Niger whom the Songhay had reduced to servilestatus, just as other groups had been so reduced (the ‘casted’craftspersons—potters, weavers, woodworkers, smiths, etc., andmusicians) among the Mande and the Fulani. 36

Songhay Origins and early History

The Songhay, who are the principal actors in al-Safidı’s history, are inmany ways the most elusive of the groups under discussion. In thefirst place the classification of the Songhay language and its severaldialects within a larger family of African languages remainsproblematic.37 For long unclassified, Songhay was eventuallyassigned by Greenberg to the Nilo-Saharan group.38 More recently,Nicolai has proposed first that it belongs in the Manding family, andlater that it may be a Tamasheq creole.39 The term Songhay was usedto refer to the language as early as the beginning of the sixteenthcentury when Leo Africanus noted Songhay as the language ofWalta, Timbuktu, Jenne and Gao (and, implausibly, Mali).40 Exceptin Gao, Songhay was, at that time, little more than a lingua francawhose currency resulted from the incorporation of the citiesmentioned into an expanded Songhay state from the 1460s. Mothertongues spoken in these areas were dialects of Soninke and otherManding languages, Berber, and Arabic. Prior to the expansion of theSonghay state, Songhay appears to have been mainly spoken alongthe banks of the Niger and its hinterland from Gao southwards, butfor how long we cannot be sure. This distribution pattern centred onthe Niger would seem to indicate that the language originally spreadinto the area through riverine expansion of its speakers.41

This expansion was apparently from south to north. Such evidence asthere is points to a southerly homeland for the Songhay (includingthe Songhay-speaking Sorko), in the area known as Dendi (meaning,essentially, ‘south, downstream’) in the south of what is now the

36 See Tamari (1997). There is a parallelism, both semantically and functionally between

these Arbi (‘black man’) and the present-day Gaabi or Gaabibi (‘black body’), a term used torefer to Songhay-speaking agricultural populations and slaves in the fifth-generation; see Olivierde Sardan (1982), 144.

37 On the Songhay language and its dialects, see Rouch (1954), 12-16; Prost (1956).38 Joseph H. Greenberg, The Languages of Africa, The Hague: Mouton, 1966, 130.39 R. Nicolaï, Préliminaires à une étude sur l’origine du Songhay: matériaux, problématiques

et hypothèses, Berlin: Reimer, 1984; Idem, Parentés linguistiques (à propos du Songhay), Paris:Éditions du CNRS, 1990.

40 Leo Africanus (1956), i, 16. He spells it quite clearly ‘Songai’.41 The origins of the isolated pockets of Songhay patois in Agades (down to the nineteenth

century) and the Tegidda-In Gall area are a matter of conjecture, but a plausible hypothesis is thatthey resulted from the establishment of Songhay-speaking merchant communities. An alternativehypothesis would be that they are relict communities of a wider Songhay-speaking communitywhich was at some stage overrun by Tamasheq-speaking Tuareg. On the In Gall-Tegidda region,see Bernus & Bernus (1972).

Republic of Niger and the extreme north of the Republic of Bénin.Dendi was a province of the Songhay empire, whose governor, theDendi-fari, was one of the highest ranking state officials. It was toDendi that the askiyas retired after defeat at the hands of Moroccanforces in 1591, and from there that they organized resistance. Dendiwas an ancient source of warriors for Songhay. A century earlierSunni fiAlı had gone to Lolo in Dendi to raise a large army which hehad placed under the command of the Dendi-fari.42

Given the mobility of the Sorko, it is likely that they were the firstSonghay speakers (or speakers of proto-Songhay)43 to move up-stream from Dendi and to establish small settlements on the banks ofthe Niger.44 One such settlement may have been at Kukiya, justabove the rapids of Fafa and Labbezenga where canoes would havehad to be unloaded and carried some distance during the low-waterseason. At some stage they may have been followed on land bySonghay-speakers mounted on small local horses who subdued theexisting agricultural populations—perhaps of Voltaic origin—as theywent.45

Again, Kukiya would have attracted settlement as a natural way-station on the river route, and by the fact that it was close to thenorthern limit of rainfed agriculture, that is, if rainfall at that time(somewhere in the first millenium) was roughly the same as at thepresent day. These incoming Songhay horsemen would have estab-lished control over the Sorko there and, while forming a symbioticrelationship with them, would have made them the inferior partner.This process may have been mythologized in the legend of thealleged Yemeni brothers who arrived at Kukiya, one of whom killedthe river god (symbolized as a fish), and assumed that god’s positionof authority over the local folk.46

Later, when North African traders arrived on the banks of the Nigerat the mouth of the Tilemsi valley—probably in the early ninthcentury—they began by doing business with Sorko encamped on the

42 TF, 46 (trans. 89).43 In point of fact, we know nothing about the origins of the Sorko, and they may have

adopted Songhay only after they came under the hegemony of land-based Songhay speakers.44 The arguments made in this paragraph and the following two paragraphs represent a

hypothetical reconstruction of events, though Arabic sources help to corroborate the account sofar as the growth of a small kingdom at Gao (Kawkaw) is concerned; see Hunwick (1985a), 3-9for an elaboration.

45 See further below, p. xlv.46 See below Ch. 2. The ‘Yemeni’ element in this legend would be a later Islamic

embellishment designed to give a prestigious origin to the Songhay, just as Zb˙r/Jb˙r istransmuted into Jbir [al-Anßrı], a Companion of the Prophet Mu˛ammad; see TF, 29-30 (MS Conly). See also MS 2410/173, Bibliothèque de l’Institut de France, Paris, for an account of Songhayorigins that would make Jbir one of four men who deserted from an army of the Caliph fiUmar b.fiAbd al-fiAzız in the Yemen, and fled to West Africa.

opposite bank.47 As this developed into more regular trading,per–haps involving grain transported from south of Kukiya, theSonghay chiefs at Kukiya were encouraged to move north todominate this trade, and settle themselves on the left bank at whatbecame Gao (or Kawkaw in the traders’ parlance). The settlementflourished, and North African traders established a permanentsettlement for them–selves at Sane, some five miles up the Tilemsivalley on the right bank of a wadi. By the tenth century the Songhaysettlement at Gao had developed into a small kingdom that hadestablished its hege–mony over the peoples living along the traderoutes that radiated out from Gao: northwards towards Tdmakkat,eastwards towards Aïr, and westwards towards Ancient Ghana. Thisfirst Songhay kingdom could flourish because it lay at a crossroads oftrade routes leading on the one hand to North Africa, and on theother to Egypt. The raison d’être of both routes was the gold-dustthat these Mediterranean-based merchants obtained from AncientGhana. The merchants could also bring southwards that preciouscommodity, salt, on which the Gao rulers seem to have maintained amonopoly;48 they may also have brought the larger Barbary horseswhich could be cross-bred with local horses to produce a breed thatwas both stronger and better adapted to local conditions, thusfacilitating domination of neighbouring peoples. One by-product ofsuch domination would have been slaves, which could be barteredwith the North African merchants for more horses or for othergoods.49

The TS presents us with a list of thirty-two rulers of the Zuwdynasty, though it is not clear from al-Safidı’s account where theylived, except for the first, Alayaman, who was the ‘Yemeni’ whoarrived in Kukiya. According to al-Safidı, the fifteenth of these rulersconverted to Islam, allegedly in the year 400/1009-10. In fact, it must

47 Again, this may be what is mythologized in another ‘stranger from the east’ legend. The

‘Notice Historique’, in TF (trans.), 329-31, tells of a giant from the Yemen who arrived on the leftbank of the Niger and established relations with the people of Gao, who then lived on the rightbank, eventually becoming their chief. It is possible that the Yemeni element in these legends wasintroduced through contact with the ∑anhja, who claim a remote Yemeni ancestry.

48 See al-Muhallabı, Kitb al-fiAzızı, quoted in Yqüt, Mufijam al-buldn; see Corpus, 174.49 Law (1980), 121, argues that the paucity of evidence for the military importance of horses

before the thirteenth and fourteenth centuries suggests that before this time horses were only ofmarginal military importance. Yet it is difficult to see how an empire as large as Ancient Ghana,for example, stretching between the Middle Senegal river and the Inland Delta of the Niger, couldhave been established and maintained without a cavalry base, and the same may be said of thestate based on Kawkaw (Gao), described by al-Yafiqübı in the ninth century; see Corpus, 21. Al-Muhallabı (loc. cit. supra) says that the ruler of Gao and his court rode horses without saddles, andit is hard to believe that they did this merely for ceremonial purposes. Law also suggests (op. cit.,28) that the Barbary horse was not introduced into West Africa until the time of Mansa Müs,since it is only at this time that we have an explicit reference to trans-Saharan trade in horses, andearlier references often refer to the West African horses as being small. Formally, he is correct,yet it is difficult to imagine that in the preceding half millenium of contact no Barbary horses werebrought across the Sahara for trade, and no sub-Saharan African ruler sought better breedingstock.

have been somewhat earlier than this, since al-Muhallabı, writingbefore 985 A.D., reported that the ruler at Gao was a Muslim.50 Al-Bakrı, writing in 1068, reported that none but a Muslim might rulethere, indicating that the dynasty was solidly attached to Islam by thistime.51 The militant Islamic movement of the Almoravids in thewestern Sahara, which was to conquer and unify Morocco and mostof Andalusia during the second half of the eleventh century, hadsome repercussions south of the Sahara. In the west it seems to havedestabilized Ancient Ghana, while ∑anhja Massüfa nomadsassociated with the Almoravids apparently pushed eastwards alongthe Sahelian corridor, establishing a short-lived dynasty at Sane thathad commercial contact with its clansmen in Almeria.52 Other∑anhja nomads at this time probably settled in Walta, some createda settlement that grew to become the city of Timbuktu, while othersmoved on towards Aïr.53

Malian Hegemony

The history of Gao and the Zuw dynasty in the thirteenth century isobscure, though some of the Zuw rulers were apparently buried atSane, further strengthening the argument for the integration of the∑anhja of Sane with the Zuw dynasty of Gao.54 Towards the end ofthe thirteenth century the newly established regional power of Malimoved to extend its hegemony over the Middle Niger, and soon alsopushed the borders of its authority to the Atlantic coast in the west.This was the same strategy that Songhay was to attempt two centurieslater: that is to control all the active trading entrepots along theSahelian corridor in a bid to monopolise the gold trade in particular,and other exchanges between the lands of the Mediterranean and theWest African forests in general.What became of the Zuw rulers in the face of Malian hegemony isnot clear, but they may have survived as tributary rulers, just as underthe askiyas many conquered territories were left under a form ofindirect rule, with Songhay officials installed to see that tribute waspaid and loyalty maintained. TS passes rapidly on to the Sunni

50 See Corpus, 174.51 See Corpus, 87. It is clear, however, that the rulers still observed certain indigenous rites,

a pattern which was to be familiar for centuries to come.52 Much mystery remains surrounding these ∑anhja rulers, who perhaps intermarried with

the Zuw rulers’ daughters; see Lange (1991), 261-2. For interpretations of the inscribedtombstones associated with these rulers, see Sauvaget (1949); Hunwick (1974, 1979 & 1994), andLange (1991 & 1994). Paulo de Moraes Farias has prepared a corpus of these inscriptions andmany others from Es-Souk (Tdmakkat) and Bentia (Kukiya), which will be published shortly; seeMoraes Farias (forthcoming). The ∑anhja association with the Zuw rulers may possibly be whatlies behind Leo Africanus’s statement that the Sunnis were of ‘Libyan’ (i.e. Saharan) origin, if, asTS claims, the Sunnis were of the same family as the Zuws; see below, p. 7.

53 The probable descendants of these latter ∑anhja are the Inusufen. They, like all otherswho moved east became culturally and linguistically completely Tuareg.

54 See Hunwick (1974), Appendix A.

dynasty, but with no chronology, though it later mentions the Malianconquest of the Middle Niger, again without chronology. TF is evenless helpful. It devotes a chapter to Mansa Müs’s pilgrimage, butmakes no overt mention of the Malian conquest of the Middle Niger.A later chapter gives the names of the last five Sunnis, while the‘Notice Historique’ gives a full list. The latter source, like the TS,links the first Sunni ruler, fiAlı Kulun (or Golom, as in the ‘NoticeHistorique’), to Mali, saying that he was born there and served theruler, but broke with him for reasons it would take too long torecount. TS makes fiAlı Kulun (and his successor Silmn Nri) a sonof Zuw Ysiboy, and says he was taken away as a hostage prince toserve at the Malian court,55 later finding his way back to Songhayand freeing Songhay from the Malian yoke.If we are right, as argued in Appendix 2 below, that the title Sunni is,in fact, Manding sõ-„nyi, meaning ‘subordinate or confidant of theruler’, then the founder of this dynasty would have been a Malianofficial who revolted. But this does not necessarily mean he was aMalian (i.e. a Malinke) himself. TS clearly sees fiAlı Kulun as a sort ofnationalist hero, and in order for him to have led a breakawaySonghay he would have to have been himself Songhay. A possibleexplanation is that he was one of the Zuw princes who had been leftin place to govern his territory under the eye of a Malian official, andthat he subsequently revolted and sought independence. In thisinter–pretation the Sunnis would have been in some sense acontinuation of the Zuw dynasty. Some support for this may be seenin the fact that, in the traditions of Songhay magicians, Sunni fiAlı isknown as za beri wandu—‘the great and dangerous Z’.56

Since Malian hegemony apparently continued in the Gao region untilwell into the fifteenth century, and such a revolt probably took placein the early fourteenth century, it is proposed that this Songhay‘loyalist’ dynasty had its seat at Kukiya rather than Gao. Kukiya wasthe ancient seat of Songhay rulers, but was of no particularimportance to the Malians, for whom the commercial centre of Gaowith its links to North Africa, to the copper mines of Tagidda, and toEgypt was the prize to be retained. They could afford to let Sunnirulers control Kukiya and southwards, just as the Arma by around1610 realised that they needed control of nothing beyond Gao, andthat it was preferable to let the askiyas have their way in the

55 There may have been a similar practice of holding the sons of rulers of tributary

territories hostage at court in Ancient Ghana; see al-Bakrı in Corpus, 80.56 Rouch (1953), 186. This also provides a better explanation for the identification of the

group known as the Zuw (or Z) b˙r banda—‘posterity of the great Zuw’. Instead of trying tolink them to the Zuw dynasty which, under that name, had disappeared from history in the earlyfourteenth century (see Lange (1994), 299), we may now postulate that they were survivingmembers of Sunni fiAlı’s clan whom Askiya al-˛jj Mu˛ammad had every reason to wish to be ridof; see below, p. 109.

southlands rather than squander scarce resources attempting to wipethem out. Beyond Gao the Malians, like the Arma later on, werelogistically just too far extended.When internal troubles weakened the Malian empire towards the endof the fourteenth century, their grip on the Middle Niger loosened.57

By 1438 they had withdrawn from Timbuktu, and it seems logical toassume that by or before that date they had withdrawn from Gao.Once the Malians had withdrawn from Gao, the Sunnis wouldpresumably have moved back to claim the city. The antepenultimateruler of the dynasty, Sunni Sulaymn Dma, is said to have laid wastethe territory of Mema, to the west of the Inland Delta, which hadalready established its independence from Mali. This event is notdated, but Sulaymn’s reign ended in 1464, and we may view hisaction as a harbinger of the expansion of Songhay, which took placeunder his successor Sunni fiAlı B˙r (1464-92). It certainly establishedthat some of the lands of the old Malian empire were ripe for thetaking.

The Creation of an Empire

From the accession of Sunni fiAlı onwards down to the end of theperiod covered in this translation of the TS , we are much betterinformed by our local sources, although they often leave tantalizinggaps in their narrative and provide us only minimally with the sort ofsocial and economic data so valuable to historians. Rather than dealwith the political history in detail here, a summary of the politicaldevelopments of the period will be offered, followed by an attempt tobuild a picture of the internal organisation and administrativestructures of Songhay, and the intellectual life of Timbuktu.Sunni fiAlı (reg. 1464-92) is portrayed in the chronicles as a man ofrestless energy and ruthless disposition, who rapidly conquered anextensive swathe of territory around the Niger Bend to as far as Jennein the west. His principal residence—his ‘capital’—was Gao, but hehad royal residences also in Kukiya, Kabara, and in Tindirma, re-flecting the mobile nature of his kingship in a period of conquest.The heart of the empire he created was the Songhay-speakingterritory from the region of Gao southwards to Dendi. What he addedto this—the Timbuktu region, the Inland Delta and the northernlacustrine region, the Jenne region, and the Bandiagarauplands—were all non-Songhay territories, and except for theBandiagara uplands, all inhabited by predominantly Muslimpopulations, often of long standing. His aim was clearly to control theentire stretch of river constituting the Middle Niger, and the tradepassing through its western reaches—specifically, the gold trade

57 On the history of Mali and its decline, see Levtzion (1973).

passing through Jenne and Timbuktu. He had a sizeable andevidently well disciplined army at his disposal, led by cavalry, partlybuilt up by his predecessors, and partly raised by him for thepurpose. He was, of course, opposed in these ventures, but suchopposition as there was, was local. The Malians had withdrawn fromthe region some thirty years earlier, and the ∑anhja/Tuareg whocontrolled Timbuktu connived at his con–quest. Jenne withstood asiege, but in the end an honourable settle–ment was reached. Theonly competition was from the Mossi of Yatenga, who attackedWalta in 1480, but were defeated by Sunni fiAlı as they withdrew.58

Songhay conquest, exacerbated by Sunni fiAlı’s harsh rule, wascertainly not welcome in these westerly regions. It was alien, and wasprobably looked on as ‘pagan’ or near-pagan domination, Sunni fiAlıhimself being not a scrupulous practitioner of the faith, and the forceshe recruited in Dendi, for example, being most probably littleinfluenced by Islam.59 The dislocation caused by his constantcam–paigning may also have disrupted trade, and I have suggestedelsewhere that Dyula gold traders may have avoided the MiddleNiger at this time, and used a direct route from the Middle Volta toHausaland.60 When Sunni fiAlı died in November 1492, his son SunniBru (Abü Bakr) was immediately challenged by a man who hadbeen governor of the Bandiagara uplands under his father, with thetitle Tondi-farma . With the help of the governor of at least onewestern province, the Bara-koi (though one may suspect that he hadmore support), he was able to defeat Sunni Bru in battle five monthslater, and take power as Askiya Mu˛ammad.61

Much is unclear about this process of transition. To begin with, thevery origins of Askiya Mu˛ammad are debated. The two chroniclesindicate that he (or his father) belonged to the Silla or Ture clan, bothof which are Soninke. The TS is silent on his mother’s origin, but TFmakes her a daughter of the chief of Kur, a large island in the Nigerupstream from Timbuktu. Oral traditions, perhaps wanting to linkhim to their great hero Sunni fiAlı, make Askiya Mu˛ammad a son ofa sister of his. Whatever the case may be, Askiya Mu˛ammad waspart of the Songhay governing elite, and seems to have thought ofhimself as being part of a long line of Songhay rulers stretching backinto the past, since in his questions to al-Maghılı he refers to hisforefathers having conquered the land [of Songhay] and parcelled it

58 For a detailed analysis of Sunni fiAlı’s conquests, see Ba (1977).59 This argument is based on the fact that south of Kukiya there were no major trade routes

along which Islamic influence might have passed, and the evidence presented in al-Maghılı’s‘Replies’ regarding Sunni fiAlı and his matriclan (see Hunwick (1985), 69-70), admittedly partisan,but not to be dismissed out of hand. It is also clear from the work of Rouch, Stoller and others thatlocal non-Islamic religious cults still flourish in this area in the twentieth century.

60 See Hunwick (1985b), 334-5.61 It was only after his pilgrimage of 1497-8 that he could be called al-˛jj Mu˛ammad.

out among themselves.62 The fact that he had to fight for the throneshould not surprise us. From the questions to al-Maghılı, it wouldseem that such a practice was common, and there is evidence ofsuccession struggles during the Askiya dynasty too. AskiyaMu˛ammad’s struggle, however, had a different complexion. Nomatter how Songhay he had become culturally, patrilineally he wasan outsider. His only recorded support in his struggle came from aformer province of Mali. In the light of his subsequent ideologicalstance as the pilgrim king, it seems likely that he was also seen in thewestern lands at any rate, as a Muslim warrior who would remove thetaint of semi-pagan rule.63 His success in taking power over theempire created by Sunni fiAlı enabled these former provinces of Malimore easily to accept Songhay overlordship.Under Askiya al-˛jj Mu˛ammad, the Songhay empire expandedbeyond the core riverine territories established by Sunni fiAlı toinclude tributary lands to the east and west, as well as desert lands toas far north as the salt pans of Taghza. In the east the Tuaregsultanate of Agades was made tributary, and although the evidence isthin, it seems that attempts were made to bring some of the Hausastates within the political orbit of Songhay, though Kebbi, which hadpreviously been under fairly tight Songhay control, asserted itsindependence in 1517. Attempts to control Borgu by Sunni fiAlı,Askiya Mu˛ammad, and later Askiya Dwüd, were unsuccessful. Inthe west, former territories of the Malian empire such as Kala andBendugu in the south-west, and Bghana between the Inland Deltaand Futa Kingui were made tributary, and more ephemeralexpeditions were launched against Galam in the upper Senegal valleyand central Mali. In the south, the province of Hombori remained theborder of Songhay territory, though a jihd was made against theMossi of Yatenga without a lasting result.Thus, by around 1520 a large, somewhat loosely knit empire hadbeen established, which controlled the Middle Niger very firmly, andsomewhat more tenuously a large swathe of territory stretching fromthe river Senegal in the west to the Aïr massif in the east, from theborders of Borgu in the south to the salt pans of Taghza in the north.In the tributary territories former rulers retained political control andsimply paid tribute to the ruler at Gao. In the core territories somelocal rulers were left in political control, while a parallel Songhayofficial looked after Songhay interests and ensured loyalty. At Jennethe Jenne-koi remained traditional ruler of the city and its hinterland,

62 See Hunwick (1985a), 72.63 He performed the pilgrimage to Mecca in 1497-8, and returned with the authority of the

fiAbbsid caliph of Cairo to rule in his name; see Hunwick (1966a, 1990). On the attitude of theTimbuktu fiulam√ to the legitimacy of Askiya al-˛jj Mu˛ammad, and to his successors, seeHunwick (1995b), 299-303.

while the Jenne-mondyo looked after Songhay interests. In Msinathe Msina-koi was a Jallo∫e Fulani, but there was also a SonghayMsina-mondyo. In the desert north of Timbuktu the Maghsharen-koi was a ∑anhja/Tuareg chief, but Songhay also had a Koira-banda-mondyo and a Barbüshi-mondyo in Timbuktu and an importantmilitary commander, the Balmafia in Kabara. The key revenue-producing salt pans of Taghza—also a strategic northernoutpost—were commanded by a Taghza-mondyo, an officialappointed by the Songhay ruler, though belonging to one of thedesert peoples.The strategy of expansion beyond the Niger valley seems to havebeen one of controlling the axes of north-south trade by establishingdominance over the gold-salt trade. Thus it was not sufficient tocontrol Timbuktu and Jenne, but it was also necessary to controlWalta, and as far as possible all the territory westwards to the riverSenegal. In the east control over Agades was essential to this strategy,and because there was a trade in gold from the Middle Volta regionto Hausaland it would have been desirable to control Kano andKatsina, the principal mercantile cities, or failing that to controlBorgu, through which territory the trade route from the river Voltapassed. Neither area was successfully controlled, making Agades themore important for Songhay to dominate. In the north, control overTaghza was essential, since salt was a major exchange commodity inthe gold trade. Control of Taghza was challenged by the Safidians ofMorocco as early as the reign of Askiya Is˛q I (1539-49), while in1590 Songhay rejection of such a claim was to be used as a casusbelli by the Moroccans.There was no further expansion after the reign of Askiya al-˛jjMu˛ammad. On the other hand, there were campaigns into severaltributary territories during the long reign of Askiya Dwüd (1549-82), suggesting that their allegiance had fallen away. Whilst therewere no serious local challenges to Songhay in the sixteenth century,the challenge from Morocco began even before the Safidian dynastyhad completely ousted the former Wa††sid dynasty. The challengetook the form of claims to the taxation on salt extracted from the saltpan of Taghza, though it was not until after the accession of MülyA˛mad al-Manßür in 1578 that serious consideration was given toconquering Songhay.The Safidians claimed descent from the Prophet Mu˛ammad, and inthe face of the final extinction of an fiAbbsid caliphal line in 1517,they began to assume claims to the title of imam and amır al-mu√minın.64 Al-Manßür in particular, stressed these claims based ondescent from the Prophet. Threatened by the Ottomans in the east,

64 See below, pp. 299-301.

and with the Portuguese still occupying forts on his Atlantic coast, al-Manßür conceived of a way of enriching his state by controlling thegold trade at (as he thought) its source, and at the same timeenhancing his caliphal status by gaining the allegiance of a range ofWest African and Saharan rulers. In rapid succession in the 1580s hesecured a diplomatic coup with Bornu’s act of allegiance to him asamır al-mu√minın, and the submission of a number of nomadic chiefsin the western Sahara.65 This was quickly followed by the conquestof the oases of Gurra and Tuwt, keys to trans-Saharan trade, andlastly by the occupation of the salt pan of Taghza. Finally in 1590al-Manßür dispatched a well-trained and highly disciplined force ofsome 4,000 men, many of whom were of Christian Andalusianorigin, armed with muskets. They arrived in Songhay just as Songhaywas recovering from the effects of a disastrous civil war. Despite theirsuperior numbers, Askiya Is˛q II’s forces, fighting with swords,spears, and arrows were no match for the technological superiority ofMoroccan firearms. Nor yet could they rally in defeat and drive theMoroccans out. Old divisions plagued them, and the Moroccans wereable to divide and rule, restricting their own territory to a thin linearound the Niger from Jenne to Gao, and for the first twenty yearsconstantly reinforced from Morocco. In the end, however, there wasan implicit truce. The Moroccans were left in control of their swatheof the river Niger, while the Songhay who rallied to the successors ofAskiya Is˛q II were left to run their own affairs in Dendi.

Songhay Administration

Askiya al-˛jj Mu˛ammad inherited some sort of state administrativestructure from his predecessor. Among the offices that certainlyexisted in Sunni fiAlı’s reign were those of Balmafia, Bara-koi(governor of Bara province), Benga-farma (governor of thelacustrine region), Hi-koi (commander of boats), and Tondi-farma(governor of the Bandiagara uplands), and we may presume that heintroduced offices such as Jenne-mondyo, Msina-mondyo, andTimbuktu-mondyo when he established Songhay hegemony in thoseplaces. To these offices we know for certain that Askiya al-˛jjMu˛ammad added two important posts: those of Bghana-fari, orgovernor of the province of Bghana, and Kurmina-fari, a powerfulregional governor who resided in Tindirma, and who later becameviceroy for all the western provinces.Several other important offices are mentioned during the reign ofAskiya al-˛jj Mu˛ammad; some may have been created by him,while others may have existed from the days of Sunni fiAlı. Such

65 See al-Fishtlı (n.d.), 65-71, 78-81; al-Nßirı (1955), v, 104-11; al-Hajj (1983); Hunwick

(1985b).

offices included the Dendi-fari (governor of the southlands), one ofthe highest-ranking officials in Songhay, whose occupant had theprivilege of wearing special clothing and of speaking his mind to theaskiya; the Dirma-koi (governor of Dirma province), who had theprivilege of entering the askiya’s palace on horseback, and ofbuilding a two-storey house; the Fari-mondyo, a senior official, oftena son of the ruling askiya, who probably oversaw the collection ofland taxes, and perhaps administered the royal estates; the Hugu-koray-koi, ‘Chief of the interior of the palace’, a close confidant ofthe askiya, and probably a eunuch;66 the Kabara-farma, harbourmaster and collector of custom dues at the port of Kabara (an officeapparently first mentioned by Leo Africanus); and the Shfi-farma,who may have been the governor of the important river port of Sahon the Niger a little downstream from Lake Debo. We may alsoassume that the office of Taghza-mondyo existed at least fromAskiya al-˛jj Mu˛ammad’s day. We know of the titles of manyother offices, but the function of many of them is unclear, and it isnot possible to say with any accuracy when they were introduced.67

There was, of course, a hierarchy to these offices, some glimpses ofwhich we are given by al-Safidı. The Kurmina-fari was the mostsenior, the first holder of this office having been delegated by Askiyaal-˛jj Mu˛ammad to run the affairs of state during his absence onpilgrimage. Later, towards the end of Askiya Dwüd’s reign, theKurmina-fari was made viceroy for all the western provinces.Never–theless, he could still be subordinate to the Hugu-koray-koi,the vizier-like official and close confidant of the askiya, during amilitary expedition, and when Kurmina-fari Dwüd over-hastilyreadied himself to succeed his brother Askiya Is˛q I, he was rebukedby the loyal Hi-koi. The Hi-koi was clearly in the top echelon ofSonghay officials, and we learn that he took precedence over theHombori-koi; however, the Balmafia took precedence over themboth.

The Askiyate

As for the supreme office, that of askiya, it was held throughout thesixteenth century by sons and grandsons of Askiya al-˛jjMu˛ammad, except for the short reign (1531-7) of his nephewMu˛ammad Bonkana. After the Moroccan conquest of Songhay asplit in the family, which probably had been in formation for someyears before the conquest, manifested itself: some sons of Askiya

66 We find, for example, a Hugu-koray-koi commanding a eunuch corps; see below. p. 171.

The title Hugu-koray-koi could also be read Hu-kokorey-koi—‘chief of the palace eunuchs’, andit is probable that only a eunuch could enjoy the intimacy and confidence of the askiya such as theHugu-koray-koi evidently enjoyed.

67 For a full list of Songhay offices, see Appendix 4.

Dwüd (and hence grandsons of Askiya al-˛jj Mu˛ammad) led amajority of Songhay who fell back upon Dendi and resisted theMoroccans; others gradually joined the Moroccans, being appointedby them as leaders of the minority of Songhay who chose tocollaborate with the new regime. This raises the question of whatexactly is meant by the term ‘Songhay’ as used in al-Safidı’s work.The most frequent use of the term occurs in the phrase ‘the Songhayfolk’ (ahl Sughay), or occasionally just ‘Songhay’.68 In such usagethe term evidently means the ruling elite, primarily the descendants ofAskiya al-˛jj Mu˛ammad, but also clearly includes a much widergroup.69 These Songhay would have been the descendants of thoseclans that, at a much earlier date, had ‘conquered the land andparcelled it out among themselves’,70 reducing the local peoples tosubordinate or servile status. We may thus best understand al-Safidı’sterm ‘Songhay folk’ by thinking of it as representing a cavalry-basedgroup of clans constituting a ruling class superimposed onagricultural, hunting and fishing populations and, despite taboos onintermarriage with them,71 no doubt genetically intermingled withthem, and also with captives from other neighbouring groups. Themodel would thus be one analagous to that of the Mande horsemenwho conquered Gonja and remained as a ruling elite.72

We know little of the rules of succession to the supreme office beforethe period of the askiya, but the little we do know would suggest thatit was patrilineal, with power passing through a number of brothers inevery generation before descending to the succeeding generation.The TS, for example, tells us that Silmn Nri succeeded his brotherSunni fiAlı Kulun, the first Sunni ruler. At the other end of thedynasty, Sunni fiAlı’s father had also been a ruler, but there wereeither four or seven rulers between them. During the askiya periodthe pattern is particularly clear. After Askiya al-˛jj Mu˛ammad,four of his sons ruled, followed by nine of his grandsons (includingsix as askiyas of Dendi). There was probably an inherent rule ofseniority; we are told that Askiya al-˘jj succeeded his father AskiyaDwüd as the most senior of his sons, though in other cases the

68 More often he uses ‘Songhay’, without qualification, to refer to Gao, or the lands of the

western Middle Niger where the ‘Songhay folk’ lived; this area is also sometimes called ar∂Sughay—‘the land of [the] Songhay’. The term dawlat Sughay also occurs, evidently meaning therule of the Songhay folk (i.e. before the Moroccan conquest).

69 The term ahl in Arabic signifies any coherent group, but is used primarily for a group thathas genealogical links, e.g. 'family', 'clan'. In a broader sense it may be used in phrases such as ahlMlı—'the Malians', or ahl al-kitb—'people of the scripture', i.e. Jews and Christians.

70 See above, p. xxxiii.71 See Hunwick (1992).72 See Jack Goody, ‘The over-kingdom of Gonja’, in Daryll Forde & P.M. Kaberry (eds),

West African Kingdoms in the Nineteenth Century, Oxford: Oxford University Press for theInternational African Institute, 1967, 179-205; Ivor Wilks, Nehemiah Levtzion & Bruce Haight,Chronicles from Gonja: a Tradition of West African Muslim Historiography, Cambridge: CambridgeUniversity Press, 1986.

determinant was the support given by the brothers to one amongthem. To secure the succession, it was necessary to be close to Gao atthe crucial moment to forestall the challenge of rivals. In a bid tosecure the succession, Askiya Dwüd (then Kurmina-fari) rushed toKukiya before his brother Is˛q had died, and was brusquely orderedback to his post by the Hi-koi. Often the moment of transfer of powerwas determined by the successor, sometimes violently. Müs deposedhis father Askiya al-˛jj Mu˛ammad and seized power, and he inturn was assassinated by his uncle Mu˛ammad Bonkana. The latterwas ousted in a coup d’état by his nephew Askiya Ismfiıl. Later inthe dynasty Mu˛ammad Bni deposed his brother al-˘jj, but soonhad to ward off an attempted coup by his brother Nü˛, and then facea major revolt led by two other brothers.The power of the askiya was largely determined by his own force ofcharacter and the degree to which he could rein in his fraternal rivalsand secure their cooperation.73 The most dangerous were simplyeliminated; Askiya Müs, who may have been paranoid, broughtabout his own demise through excessive fratricide. On the otherhand, all of the askiyas appointed brothers of theirs to high office, aredistribution of such offices often taking place on their accession,thus eliminating hereditary governorships. Since there was ahierarchy of offices (the Kurmina-fari being the most senior),brothers could work their way up and thus be groomed for thehighest office.Although all of the provincial governors would have had fightingmen under their command, the askiyas themselves had a central armyunder their control, which they led on campaigns. Under Sunni fiAlı itseems that all males could be drafted into service, but Askiya al-˛jjMu˛ammad made a distinction between ordinary subjects and thearmy.74 Under his second successor Mu˛ammad Bonkana, the armywas increased by 1,700 men, suggesting a clearly defined militaryforce. The ‘Songhay folk’, as a cavalry aristocracy, would haveprovided the military leadership, while the bulk of the forces, and inparticular the foot soldiers, would have been made up of the sociallyinferior indigenous populations. Askiya Dwüd apparently felt free totake the daughters of his soldiers as concubines, surely indicating thatthese soldiers were of slave-like status, and perhaps taken from theArbi/Gabibi whom the askiyas treated as ‘possessions’.75 Some troops

73 Al-Safidı delivered his own judgment on the askiyas, declaring that only two of them were

greater than the office they held, two were equal to it, and the remainder were not up to it; seebelow, p. 160

74 See below, p. 103,75 See further Hunwick (1985d).

were unambiguously slaves, inasmuch as there was a considerablecore of eunuch cavalry.76

Since the rank and file of the central army was composed mainly ofsuch servile people, it was necessary for a new askiya to have onlythe allegiance of his brothers and other leaders of the ‘Songhay folk’,since the bulk of the army would serve whomever was in command.When the askiya went on campaign there was a definite marchingorder for the ‘Songhay folk’, and a special commander for the footsoldiers. Both the askiyas and their senior commanders probably tooka significant part of their households into the field with them. AskiyaDwüd, for example, once took with him a wife or concubine whobecame pregnant with a future askiya;77 on another expedition hetook his mother with him.78 When Balmafia al-∑diq came out inrevolt he took his concubines with him.79 Certainly the askiya and hiscommanders would have needed a considerable retinue of persons tolook after their physical comforts: food, clothing, shelter, etc., and tocare for their horses. Leo Africanus estimated that the ‘king ofTimbuktu’ (by whom he evidently meant the askiya) had some 3,000cavalry in his army and a huge number of foot soldiers.The askiya’s palace at Gao was quite an elaborate affair, despite thedisparaging remarks of Pasha Jawdr, who was simply looking for anexcuse to move out of Gao, and if possible to return to Morocco.80

Leo Africanus describes it as follows:

The king has a special palace set aside for a huge number of wives, concu-bines, slaves, and eunuchs assigned to watch over these women. He also has asizeable guard of horsemen and foot soldiers armed with bows. Between thepublic and private gates of his palace there is a large courtyard surrounded by awall. On each side of this courtyard a loggia serves as audience chamber. Al-though the king personally handles all his affairs, he is assisted by numerousfunctionaries, such as secretaries, counsellors, captains and stewards.

The askiya had some sort of formal place where he sat in statepartially enclosed by a pallisade. When people addressed him theyhad to pour dust on their bare heads first, except for the Jenne-koiwho could pour flour, and the Kurmina-fari who could leave hisheadgear on whilst pouring dust.81 When subjects addressed theaskiya formally in court, or he replied, communication passed

76 See below, p. 171.77 See below, p. 15478 See below, p. 155.79 See TF, 139.80 He described it to Müly A˛mad al-Manßür as inferior to the house of the head of the

donkey-drivers in Marrakesh; see below, (text, 142)81 TF, 11. He also had the privilege of using ceremonial horns called kakaki, otherwise a

prerogative solely of the askiya.

through an intermediary, the Wanadu, who conveyed his words tothe assembly.82

Certain persons could meet with him informally, such as the Askiya-Alfa, who, as a scholar, could even call on Askiya Is˛q II at siestatime;83 a Fari-mondyo could gain an emergency audience late atnight.84 At his court the askiya had a royal griot—the gesere-dunka—just as in the Malian court witnessed by Ibn Ba††ü†a, and healone could address the askiya by his personal name.85 AskiyaMu˛ammad Bonkana is credited with enlarging and embellishing hiscourt, providing splendid garments for his courtiers and introducingvarious musical instruments and male and female singers.86 At acertain point near the palace those high officials who were permittedto have official drums had to silence them; only one official, theDirma-koi, is said to have been allowed to enter the palace onhorseback.87

The reigning askiya was the guardian of certain symbols of kingship.There were twelve royal standards which would have been carriedinto battle by his bodyguard, a royal drum, and an object called dintüri, which means kindling wood, and may have been a symbol ofthe first fire lit by their ancestors in the land, symbolising theirownership of it.88 The askiyas subsequent to Askiya al-˛jjMu˛ammad inherited his signet ring, sword and turban, and theseand the other objects mentioned above had to be handed over if theaskiya were deposed or abdicated.89 In addition there was a specialstable of horses reserved for the askiya’s use.The askiyas all made Gao their base, and we may rightly refer to it asthe capital of the Songhay empire during this period. UnlikeTimbuktu, it had direct access to the river Niger. It was laid out alongthe waterfront, and even if in the low-water season islands formed,restricting the flow to a narrow channel, as is now the case, it wasaccessible twelve months of the year to rivercraft. We do not know

82 It was a common practice among West African rulers to have a spokesperson whom the

ruler would inform of what he wanted to say, and who would then announce this to the assembledcourt. They also acted acted in a similar capacity vis-à-vis those who wished to address the ruler,and are referred to in Arabic accounts as turjumn (i.e. dragoman); see Hopkins & Levtzion(1982), 80 (Ancient Ghana), 289 (Mali). In Asante in modern Ghana this type of official is stillcalled okyeame, and in English, ‘linguist’.

83 See below, p. 173.84 See below, p. 145.85 For a discussion of griots in both Mali and Songhay, see Tamari (1997), 79-86.86 See below, p. 126,87 TF, 11. On another occasion shortly before battle, however, we are told that one of the

Songhay elite rode to the askiya’s palace and performed an advancing and retiring manoeuvrewith his horse before addressing the askiya; see below, p. 174.. This sort of manoeuvre is stillperformed as an act of allegiance by groups of horsemen in the Hausa states, and in Morocco.

88 See note to TF, trans., 274.89 The detailed list comes from TF, 153-4, and constitutes the catalogue of state objects that

Askiya Is˛q II tried to take with him when he was deposed by Mu˛ammad Gao and fled toDendi. TS only mentions ‘insignia and [ceremonial] objects of the sultanate’; see below, pp. 198-9.

exactly how large Gao was in the sixteenth century, either in physicalextent or in population, but it must have been considerable. Theinformation reaching an anonymous intelligence-gathering Spaniard,writing in Marrakesh in 1591, was that Gao had a large population,but the city had no surrounding wall.90 A count of compound houses(qußür) in Gao in the reign of Askiya al-˘jj a few years earlier,revealed a total of 7,626 such structures, and an uncounted numberof straw huts.91 Considering that one such compound wouldprobably have housed from five to ten people (including childrenand, in some cases, slaves), we would seem to have a population ofbetween 38,000 and 76,000 in the late sixteenth century, notcounting the ‘squatting’ population living in huts, no doubt on theoutskirts, as in modern Timbuktu. Barth estimated the city’scircumference at its height to have been about six miles.92 Such a sizeand population may sound exaggerated, and well they may be. Butwe must remember that Gao was the epicentre of an empire thatextended over 1,400,000 sq. km. (500,000 square miles), or roughlythe same size as the modern Republic of Mali, and was home to aroyal court of some splendour, an administrative hierarchy, and aconsiderable number of fighting men, both mounted and infantry. Itwas also an important trade emporium. When it was no more than anoutpost of the Malian empire in the mid-fourteenth century it wasdescribed by Ibn Ba††ü†a as ‘one of the finest and biggest cities of theBlacks, and best supplied with provisions’.93

It is regrettable that we have no information on how the city wasorganised or governed in the sixteenth century. A city of theproportions implied in the sixteenth-century ‘census’, and capital of ahuge empire, must have faced considerable problems of waterdistribution, sanitation and waste disposal, and above all of foodsupply. Much of its food had to come from some distance away.While some agriculture was possible on the land exposed after theNiger flood waters had abated, the crops grown were probablyvegetables and melons for the most part, and fodder for horses. LeoAfricanus remarked on the abundance of melons, cucumbers andsquash.94 Meat and dairy products would have been obtained from

90 See below, p. 283.91 TF, 146.92 Barth (1865), iii, 482.93 Ibn Ba††ü†a (1854), iv, 435 / (1994), iv, 971.94 Earlier Ibn Ba††ü†a (loc. cit.) had noted ‘incomparable fiinnı faqqüß’ at Gao. Defréméry

& Sanguinetti (Ibn Ba††ü†a, 1969) translate this as ‘concombre surnommé fiinnı’, and Beckingham(Ibn Ba††ü†a, 1994) follows them, but suggests reading fiattbı (‘striped’) in place of fiinnı, whichhas no known meaning. Mauny et al. (1966), however, translate this phrase as ‘melons appelésfiinnı’. There is certainly some semantic confusion. Al-Zabıdı (1888-9) defines faqqüß as anunripe water-melon (bi††ıkha). Lane (1863-93) follows this, but also notes that in Egypt the term isnow applied to the common cucumber. In a botanical lexicon based on al-Zabıdı’s Tj al-fiarüs, al-Dimy†ı (1965), who spells it faqqüs, is clear that it is a melon but allows two possibilities: cucumis

Fulani and Tuareg nomads, and fish directly from the Niger.95 Butthe bulk of the grain had to come from farther away. The staplegrains of this area of the Sahel are bulrush millet (Pennisetum),guinea corn (Sorghum), and, because of the river, rice (Oryzaglaberrima).96 Millet and guinea corn are chiefly grown as rainfedcrops, and hence would have come from lands at least a hundredmiles south of Gao.97 Rice would have been grown along the banksof the Niger when receding waters created favourable conditions, andin inlets that flooded during high water. From the time of AskiyaDwüd, if not before, the askiyas had their own estates where rice wascultivated along the Niger from Dendi to Lake Debo. The estateswere worked by slaves and supervised by slave officials, some ofwhom evidently grew quite rich and powerful. These estatesproduced large quantities of rice, which was sent to Gao in largeboats capable of carrying loads of some twenty tons (20,000 kg.).98

At times the total sent amounted to 4,000 sunnu, or between 600 and750 tons. At a rate of two-thirds of a pound per day, this would havebeen sufficient to feed 5-7,000 persons for a year, which may besome indication of the number of persons the askiya had to providefor directly, and hence a measure of the size of his palaceestablishment, his bureaucracy, his army and support personnel.Leo Africanus remarked that the Songhay kingdom’s revenues wereconsiderable, though its expenses were even greater. Although thiscannot have been true literally, it would seem to indicate that therewas a flourishing state-controlled economy. Details are sketchy, butwe can deduce what the askiya’s main sources of income were. First,there were direct taxes on the agricultural populations, on merchants,and on salt production at Taghza. The Fari-mondyo, overseer ofcrop fields, has already been mentioned. The question of land tax(Ar. kharj) had been discussed in al-Maghılı’s replies to Askiya al-˛jj Mu˛ammad, and al-Maghılı had advised that the ruler shouldimpose such tax ‘as will bring benefit to the Muslims anddevelopment to the land without hardship’.99 Either perceptions ofhardship differed, or Leo Africanus was exaggerating when heasserted that the askiya taxed his rural populations so heavily thatthey were scarcely left with enough to subsist on. There were alsostate obligations that at least some communities had to fulfill, and

melo (sweet melon) or Cucumis citrullus (watermelon). A common variety of watermelon is,indeed, striped, and grows easily in semi-desert conditions.

95 Ibn Ba††ü†a (loc. cit.) mentions poultry as well.96 On agriculture in the Songhay lands, see Morgan & Pugh (1969), 352-6.97 At present rainfed agriculture is possible no farther north than the Tillabery region, about

150 miles south of Gao. In the sixteenth century this limit may have been somewhat farther north.98 See TF, 94 ff.99 See Hunwick (1985a), 85, 107. See also the account of al-Ifrnı below, p. 312, quoting al-

Imm al-Takrürı, who claimed Askiya al-˛jj Mu˛ammad’s taxation was light.

could be excused from by royal privilege. The descendants of MoriHawgru were released from ‘all obligations (wa÷√if) [owed to] thesultanate and of all its dues (gharma) … even hospitality’.100

Merchants would have been taxed in one way or another, andsometimes simply by extortion. Al-Safidı tells us that during the reignof Askiya Is˛q I some 70,000 mq. was extorted from Timbuktumerchants by two agents of the askiya. This information is given as ifthe practice were outside the norm, but it does not seem a very largesum over ten years. Almost certainly, the Timbuktu-mondyo and theJenne-mondyo regularly taxed goods passing through their cities; theKabara-farma taxed every cargo that passed through his harbour, andthe Taghza-mondyo taxed every slab of rock salt that leftTaghza.101

In addition to such revenues, income was generated from tributefrom conquered territories, and from the booty seized duringcampaigns. Leo Africanus mentions the sum of 150,000 ducats (i.e.mithqls)102 that was paid by the sultan of Agades; he also claimedthat the sultan of Kano had to pay one third of his revenue to theaskiya every year, though Songhay hegemony over Kano was at bestephemeral, and perhaps non-existent. As for booty, apart fromanything else, we know that captives were often taken during theaskiya’s campaigns and enslaved. When Askiya al-˛jj Mu˛ammadcampaigned against the Mossi in 1498, so many slaves were broughtback to Gao that a special quarter had to be constructed for them. In araid on Dialan near Futa Kingui in 1501-2 many captives werebrought back, including a woman who became the mother of AskiyaIsmfiıl. A similar campaign against Borgu in 1510-6 produced manycaptives including the future mother of Askiya Müs. During AskiyaIsmfiıl’s last campaign so many captives were taken that the price ofa slave in the Gao market dropped to a paltry 300 cowries.103 Tojudge by the knowledge of enslaveable non-Muslim populationsdisplayed by A˛mad Bb in his Mifirj al-ßufiüd, slaves werefunneled into the Middle Niger valley from a wide hinterlandstretching through Dogon, Bobo and Mossi territory and to as farsouth as Dagomba, Borgu, and Yorubaland.104 While some of thesewould have resulted from military campaigns, the more distant onesmust have been brought into Songhay territory by merchants who

100 See Hunwick (1992), 138, 140.101 When Müly A˛mad al-Manßür demanded that Askiya Dwüd allow him to collect the

revenue obtained from Taghza, the askiya sent him a ‘gift’ of 10,000 mq., which may, in fact,have been equivalent to one year’s revenue.

102 i.e. approximately 22,500 oz.103 See below, p. 136,104 By Yoruba here, we should probably understand the old kingdom of Oyo, considerably

farther north than present-day Yorubaland.

purchased them in distant lands. In either case, they would have beena source of revenue.105

The askiya had certain other sources of revenue. Upon the death ofhis slave officials, the askiya inherited any gains made by them. Anelaborate example is given of this in TF, concerning a slave officialwho died and left 500 slaves, 1,500 sunnu of grain, cattle, sheep,horses, and arms of various kinds.106 All of this became AskiyaDwüd’s property. It was, in general, his practice to inherit theproperty of his soldiers, who, as argued above, were mainly of servilestatus. We do not know how far the askiya was directly involved intrans-Saharan trade, but it is likely that he had first option to buygoods imported by the caravans, and he clearly sold excess slavesobtained in campaigns to North African traders.107 He probably alsocontrolled the cowry trade. According to the ‘Anonymous Spaniard’,the askiya made a decision to forbid buying cowries from Moroccansources, and to obtain these monetary shells from Egypt or Mecca,where they were cheaper, instead.108 This suggests that the askiya hada hand in the importation of cowries, and may well have fixed theirinitial exchange value against the mithql of gold.That the askiya also had heavy expenses cannot be doubted.Although it appears from one instance that the askiya had funds ofhis own different from the state treasury,109 in practice there wasprobably not a very fine distinction. In any case he had theresponsibility for making sure that there were sufficient fundsavailable to pay the army, and to feed and provide for his largehousehold. Leo Africanus claims that the religious officials ofTimbuktu, the q∂ı, and the major scholars, were supported by theaskiya. The askiya was certainly responsible for appointing the q∂ı,upon advice, though it seems doubtful that the independent-mindedAqıt scholars, for instance, would have allowed themselves to be inthe askiya’s pocket. But if he did not pay such persons directly, hewas expected to make generous gifts to the scholars and to manyother persons of civil rank. Ma˛müd Kafiti, author of TF, was therecipient of several gifts.110 When the Great Mosque, Jingere B˙r, wasbeing renovated in Timbuktu, Askiya Dwüd sent gifts in cash andkind as his personal contribution towards the project.111 Even hischief slave on a rice plantation in Dendi was sent gifts at harvest time:a slab of rock salt, a thousand cola nuts, a black tunic and red cap,

105 See further Hunwick (1985d).106 TF, 102-3.107 See TF, 104.108 See below, p. 324. See further Hunwick (1998).109 See below, p. 154, concerning the rebuilding of the Great Mosque of Timbuktu.110 See TF, 108; Hunwick (1996), 186.111 See below, p. 154. TF, 107-8 lists further gifts of slaves sent for the mosque and for Q∂ı

al-fi◊qib’s personal use.

and a piece of black broadcloth.112 In this manner the askiya servedto redistribute wealth within his kingdom, not of course on anentitlement or need basis (though the poor are said to have receivedspecial gifts), but in accordance with the needs of a patriarchalpolitical economy. An askiya could thus be seen to be ‘generous’towards his subjects, or, for example, towards foreign embassies,113

and this enhanced the respect in which he was held.Viewed from one vantage point, the askiya was apparently all-powerful. His authority in matters of life and death, for example, wasabsolute. Al-Safidı’s account is replete with instances of askiyascondemning officials of his to death, or simply ordering theirmurder. In such matters his word was final.114 Yet, at the same timehe had to temper his absoluteness with acts that would ensure that hefound continuing favour with both his officials and with thepopulation at large. He had to be on the lookout for smoulderingrebellion, or treachery from his brothers, who could, and in someinstances did, unseat their sibling; and he had to seek ways toreinforce the loyalty of local chiefs and tributaries. One of the waysin which Songhay rulers did this was by marriage alliances. SunnifiAlı married the mother of the young Sultan of Jenne. To secure thefuture Askiya Ismfiıl’s allegiance, Askiya Mu˛ammad Bonkanamarried him to one of his daughters.115 Askiya Dwüd married adaughter of the Sultan of Mali, and married one of his own daughtersto the Maghsharan-koi, the great nomadic chief, and another to theJenne-koi; yet others were married to scholars, merchants, and armycommanders.116

The Religious Estate

Another way in which the askiyas, with the apparent exception ofAskiya Müs, attempted to maintain the loyalty of the Songhayempire’s populations—especially those of its western reaches—wasthrough their relations with the religious estate. Despite the absence ofa formal clergy in Islam, it can be argued that those men with areligious training, who were public exponents of the faith—q∂ıs,imams, kha†ıbs, and all manner of other holymen (Songhay: alfa,Manding: mori , karamokho , Fulfulde: modibbo, cerno)—didconsti–tute a ‘religious estate’. Bonded together by a common faithex–pressed through the Mlikı law-school, the theology of al-Ashfiarı,

112 See TF, 97.113 See below, p. 166.114 Sunni fiAlı was notorious for condemning people to death in a fit of rage and later

regretting it. His wise courtiers would hide away favourites who were condemned in this way, andproduce them when Sunni fiAlı expressed remorse; see below, p. 96.

115 See below, p. 126.116 See below, p. 184. For a discussion of the various techniques of the askiyas for reinforcing

their legitimacy, and ensuring loyalty, see Tymowski (1992).

and a broad ∑üfı mystical understanding, their task was to uphold thefaith and to translate it into action. For some, the task was to ensurethat the rulers themselves observed the religion faithfully, and to beready to exercise their power of invoking divine wrath in the con-trary case. Sunni fiAlı’s death following a curse upon him by aholyman is a prime example.117

The holymen all had their constituencies: q∂ıs were leading citizensof their communities and dispensers of justice; imams had theirmosque congregations; teachers had their learned colleagues andstudents; rural holymen were like parish priests, involved in allaspects of their flock's life-cycles, and teachers, medicine men, andcounsellors. Sunni fiAlı anticipated the Timbuktu scholars’ oppositionto his rule, and moved against them swiftly when he conquered thecity in 1468. Many fled to Walta, and many of those who remainedbehind lost their lives. The post-mortem judgment delivered by al-Maghılı, a North African religious advisor to Askiya al-˛jjMu˛ammad, was severe: Sunni fiAlı was an unbeliever whoseremaining goods and surviving concubines could be seized for thestate treasury—a judgment he must have made on the basis of whatlocal scholars told him.118 Askiya al-˛jj Mu˛ammad adopted aconciliatory policy towards the scholars. A sincere Muslim himself,whose accession was evidently welcomed in Timbuktu and Jenne, hesoon went on pilgrimage to Mecca, and returned imbued with thebaraka of one who has been to ‘God’s House’ (bayt Allh—theKafiba). He also returned with authority to act as the deputy (khalıfa)of the fiAbbsid caliph of Cairo to rule Songhay in his name—anempty formality in real terms, but one, nevertheless, whichlegiti–mized him as an Islamic authority in the eyes of the Timbuktuscholars. This set the stage for a century of equilibrium between theruling estate and the religious estate, each respecting the domain ofthe other, and overtly or implicity giving the other support.119

Timbuktu’s role as a centre of Islamic scholarship goes back deepinto its history and is closely connected with its role as a city ofcommerce. Al-Safidı’s account of Timbuktu’s origins makes itoriginally a seasonal camp for nomads who were taking theirlivestock to the banks of the Niger in the dry season. It is notinherently unlikely that nomads pasturing their flocks and wateringtheir camels on the river banks would want to establish a semi-permanent camp a few miles away from the river, its humidity, andwater-borne diseases. These nomads were undoubtedly ∑anhja, mostprobably members of the Massüfa branch of this great confederation,and when they established themselves there around the year 1100,

117 See below, p. 100118 For this judgment, see Hunwick (1985a), 73-5.119 This argument is elaborated in Hunwick (1996).

they were part of an eastward migration of Massüfa, influenced byAlmoravid ideology, but not necessarily active propagators of it.They remained an important element in the population of Timbuktudown to the Moroccan conquest. When Ibn Ba††ü†a visited the city in1352 he found that its population was predominantly Massüfa, as wasthe case in Walta. The leading scholarly family of the sixteenthcentury, the Aqıt, were also Massüfa.It would seem that Timbuktu was little more than a semi-permanentnomadic settlement in the twelfth century, and probably through thethirteenth, though by then other elements may have drifted there toprovide various services for the nomads. The city effectively emergesinto the light of history with the visit of the Malian ruler Mansa Müson his way home from his pilgrimage of 1324. He is said to havebrought back with him a number of Muslim scholars, as well as theAndalusian poet and man of letters Abü Is˛q Ibrhım al-S˛ilı, whomade it his home. Mansa Müs is credited with having ordered theconstruction of the Great Mosque, and al-S˛ilı with having super-vised its construction, and the building of palaces for Mansa Müsboth there and in the mansa’s capital.120 By 1375 Timbuktu, togetherwith several Saharan locations, had found its place on a Europeanmap drawn by the Jewish cartographer of Majorca, AbrahamCresques, in an atlas prepared for Charles V of France. This is a suresign that Timbuktu was, by now, a commercial centre linked to NorthAfrican cities. Cresques’s information may have been derived frommembers of Mediterranean Jewish trading networks that extendedinto North Africa and as far south as Tuwt. It is certain that in thefifteenth century Jewish merchants of Tuwt were engaged in trade tothe Middle Niger, and they had presumably been doing so for sometime.121

Scholars from North Africa, and from oases such as Tuwt, Waltaand Awjila, visited or settled in Timbuktu in the period 1350-1500.Among the best known of these was Sıdı Ya˛y al-Tdallisı b. fiAbdal-Ra˛ım al-Thafilibı, a ∑üfı shaykh who claimed sharifian descent,and evidently arrived in Timbuktu some time between 1438 and1468. He was made imam of the mosque built in his honour andnamed after him by the ∑anhja governor Mu˛ammad-n-Allh, andhe held classes in its courtyard.His teacher and friend, Modibbo Mu˛ammad al-Kbarı, also settled inTimbuktu around this time, but came from a different quarter. Hisnative town of Kbara lay to the west of Jenne, not far from Diakha,in the Msina region. Both Kbara and Diakha were known to IbnBa††ü†a, and Diakha was singled out by him as a Muslim town of long

120 On the life and activities of al-S˛ilı, see Hunwick (1990b); Binsharıfa (1992).121 See Dufourcq (1966), 141-2; Hunwick (1985c, 1991c).

standing. Kbara had produced many scholars who had taken upresidence in Timbuktu, and when Modibbo Mu˛ammad died he wasburied in a plot alongside no less than thirty of his townsmen. Hisethnic origins are not known, but it is likely he was of Soninke stock.A celebrated later immigrant to Timbuktu from the Inland Deltaregion was Mu˛ammad Baghayogho al-Wangarı, who came toTimbuktu from Jenne before 1550, and whose nisba clearly indicateshis Dyula ancestry.What is interesting here is the fact that these southerly towns wereearly centres of learning that produced scholars deeply versed in theliterature of the Mlikı madh’hab, from whom Timbuktu profited.Although there is no direct evidence, we may hypothesize that thistradition ultimately stems from Almoravid scholarship, which wasimbibed by Soninke Muslims who came in contact with ∑anhjascholars in the Sahelian regions, and who subsequently migrated totowns of the Inland Delta. They would have been closely associatedwith long-distance merchants (often in fact of the same family) andtogether established centres of trade and learning in the region. This∑anhja Almoravid tradition of scholarship was later passed backagain to generations of ∑anhja living in Timbuktu. ModibboMu˛ammad al-Kbarı taught fiUmar b. Mu˛ammad Aqıt, a ∑anhjascholar of the Massüfa branch whose lineage goes back to Abü Bakrb. fiUmar—perhaps the Almoravid leader who died in 480/1087.Mu˛ammad Baghoyogho (d. 1002/1593) was one of the mostcelebrated teachers of his generation, and among his best-knownstudents was A˛mad Bb, a great-grandson of the same fiUmar. TheAqıt family had migrated to Timbuktu from Msina during theperiod of ∑anhja (often called Tuareg) rule (1434-1468), andprovided q∂ıs for the city throughout the 10th/16th century.The mosque that was chiefly associated with teaching in this periodwas the Sankore Mosque. Sankore is a quarter in the north-east ofTimbuktu, and its name means ‘white nobles’, the term ‘white’ herereferring to the light-skinned ∑anhja, and corresponding to theArabic term bı∂n. The term ‘University of Sankore’,122 or‘University of Timbuktu’ has been applied to this teaching complex..While there is no evidence of any such institution, and the Islamictradition of learning (especially in Africa) is individualized ratherthan intitutionalized, the advanced nature of Timbuktu scholarshipand teaching is in some sense a reflection of such a term.Nevertheless, what was taking place in Timbuktu should be viewedwithin the cultural context of Islamic civilisation, rather than beingassociated conceptually with a European-style institution. The core ofthe Islamic teaching tradition has been the receiving (akhdh) of a text

122 The first writer to use this term was Dubois (1897).

(matn), which was handed down through a chain of transmitters(isnd) from the author to the student, preferably through the shortestand most prestigious set of intermediaries (isnd filı). The studentmade his own copy from his teacher’s dictation and then read it backto him, or listened whilst another student read his. When he had acorrect copy—and some shorter works might at the same time bememorized—he could then study the meaning of the text and itstechnical intricacies from lectures delivered by his teacher, and at ahigher level, by question and answer. Many texts would be studiedalong with commentaries written in other times in other parts of theMuslim world.123

The Sankore Mosque was not the only locus for teaching. The GreatMosque and the Sıdı Ya˛y Mosque were evidently also used on anad hoc basis as locations for classes. But much of the day-to-dayteaching process took place in scholars’ houses, probably in specialrooms set apart, where the scholar had his own private library whichhe could consult when knotty points arose. There is no evidence of acentralized teaching institution such as the term university implies.124

Teachers issued their individual licences (ijza) to teach particulartexts; indeed the value of the licence lay wholly in the reputation ofthe teacher. Even in North Africa and the Middle East, where colleges(madris, sing. madrasa) existed, they were not corporate bodiessince Islamic law recognizes only individuals as legal entities. Theretoo, individual licences were the norm. The institutional aspect ofsuch madrasas lay in the buildings themselves, salaries for certainteachers, and sometimes bursaries for students, all provided out ofpious endowments (awqf, sing. waqf) established by earlierbenefactors.The endowment of colleges appears not to have been practised inWest Africa, and the term madrasa, as used in TS seems to mean a‘school’ organised by an individual teacher;125 had it been a collegeof the North African or Middle Eastern type, it would have beennamed after its original benefactor. This is somewhat odd, especiallysince, during his pilgrimage, Askiya al-˛jj Mu˛ammad is said tohave established an endowment of some gardens in Medina for theuse of West African pilgrims, indicating that this ruler, at any rate,was familiar with the waqf institution.126 Two of the mosques of

123 An idea of the teaching process and curriculum in Timbuktu may be gained from A˛mad

Bb’s biography of his shaykh, Mu˛ammad Baghoyogho; see Hunwick (1990c), and below, pp.62-8.

124 The same may be said of other centres of learning in the region such as Jenne, Walta,Shinqı†, and Tishıt.

125 The madrasa of fiAbd al-Ra˛mn b. Ma˛müd b. fiUmar was held at the Sankore Mosque;that of Mu˛ammad al-Kbarı outside the Mosque of Sıdı Ya˛y, in the shadow of the tower-minaret. The term majlis is also used, meaning teaching circle.

126 In North Africa the term for pious endowment is generally ˛ubs.

Timbuktu, the Sankore Mosque and the Sıdı Ya˛y Mosque werebuilt through the generosity of individual patrons: Sankore by thegenerosity of a woman of the Aghll tribe,127 and Sıdı Ya˛y by the∑anhja governor Mu˛ammad-n-Allh to honour the scholar andmystic Sıdı Ya˛y al-Tadallisı.128 But there is no indication that therewas an endowment to pay for their upkeep and the salaries of theirimams.129 If we seek for a reason for the lack of endowed buildingsin the Middle Niger it may lie in the scarcity of productive land thatcould be made into an inalienable endowment, whose revenues couldbe regularly devoted to the upkeep of mosques and colleges.130

There was certainly no ‘official’ curriculum in Timbuktu, but certaintexts were fundamental across generations. These included theQur√n, which was memorized in childhood, though its exegesis doesnot seem to have had a prominent place in the curriculum, though itwas implicit in the study of other texts; most Islamic texts makeconstant reference to the Word of God (qawl Allh), and thus itsexegesis is studied in the context of its use as a proof-text. Alsoregularly studied were the two ‘authentic’ collections of ˛adıth, the∑a˛ı˛ of al-Bukhrı, and the ∑a˛ı˛ of Muslim, as well as the Kitb al-shif of the Almoravid Q∂ı fiIy∂, a work of piety centred on theProphet, his qualities and the reverence due to him. Next inimport–ance were the fundamental works of Mlikı law: theMuwa††a√ of Mlik b. Anas, the Mudawwana of Sa˛nün, reportingrulings of Mlik, the Risla of Ibn Abı Zayd, a conspectus of Mlikılaw written in Qayrawn in the tenth century, and the more wide-ranging, but much terser, Mukhtaßar of Khalıl b. Is˛q, a fourteenth-century Egyptian scholar, a work so terse that it is almostincomprehensible without a detailed commentary. A˛mad Bba saidhe read it with his shaykh Mu˛ammad Baghayogho no less than eighttimes. By the late sixteenth century scholars had available to them thegreat twelve-volume collection of Mlikı fatws of North Africa andAndalusia, al-Mifiyr al-mughrib by A˛mad al-Wansharısı, who haddied only in 1508. Theology was little studied, the three creeds ofMu˛ammad b. Yüsuf al-Sanüsı, a fifteenth century scholar ofTlemcen, being the principal texts. From A˛mad Bb’s account of

127 The word ‘tribe’, having had so much opprobrium heaped upon it by Africanist scholars,

deserves a word of comment. It is used here as descriptive of a particular type of socialorganisation marked by patrilineal descent from an eponymous ancestor. It is the normativecharter of identity among Arab pastoral nomads of the Sahara (and elsewhere). While this descentmay often be fictive, having one’s claim to it recognised by others in the group validates it.

128 On these mosques, see below, pp. 32, 88.129 We should note, however, the endowment of a complete set of sixty parts (juz√) of the

Qur√n by Askiya al-˛jj Mu˛ammad to the Great Mosque of Timbuktu, an endowment that wasrenewed a century later by a North African merchant (or so we may deduce from his nisba al-Misrtı); see below, p. 83.

130 For the Sudan, O’Fahey (1997), 338, suggests the opposite, i.e. land was abundant andbuilding costs low.

his own education we can see that beyond the Qur√n, ˘adıth, lawand theology, certain other disciplines were available for study insixteenth-century Timbuktu: rhetoric, logic, prosody, astronomy, andof course Arabic grammar and syntax. A single ∑üfı work ismentioned, the ˘ikam (‘aphorisms’) of Ibn fiA†√ Allh al-Iskandarı,but it is clear that ∑üfı concepts were well understood, and an asceticmysticism was practised by some.Beyond the teaching curriculum the scholars built up their own, oftenextensive, libraries. When A˛mad Bb was arrested in 1593, hislibrary of 1,600 volumes was seized, and his collection, he claimed,was the smallest among those of his contemporaries. He probablyreplenished it during his enforced residence in Marrakesh. In histreatise on slavery, the Mifirj al-ßufiüd, written in 1615, he was ableto make quotations from Ibn Khaldün’s great history, Kitb al-fiibar,and a book about Ethiopians entitled Raffi sha√n al-˛ubshn by al-Suyü†ı. He also made his own copy of Ibn Khallikn’s famousbiographical dictionary Wafayt al-afiyn, a part of which is stillpreserved in Timbuktu. Another Timbuktu scholar had a copy madeof a dictionary in thirty parts, the Mu˛kam of Ibn Sıdah,131 andMa˛müd Kafiti, author of TF, purchased a copy of al-Fırüzbdı’sgreat lexicon al-Qmüs al-mu˛ı†, through the generosity of AskiyaDwüd.132 Taken together, the presence of these major works ofArabic-Islamic learning indicates that scholarship in Timbuktu hadreached a very sophisticated level by the sixteenth century. In theteaching curriculum many original works of Islamic scholarship werebeing taught as well as some later teaching texts, while the fewglimpses we have of private collections indicates that these wereresearch libraries. The information which the copyists of the Mu˛kamof Ibn Sıdah give at the end of each volume show that there was aprofessional copying industry in Timbuktu, each volume of this workbeing the subject of a contract with the scholar for whom it was beingcopied, and the copyist and the proof reader being paid in gold.133

History was never part of the teaching curriculum in Timbuktu, noranywhere else in the Islamic world. Apart from ‘sacred history’—thelife of the Prophet and his Companions and the ‘Rightly-Guided’caliphs—it was a ‘worldly’ subject, one which a well-educatedscholar might be expected to have some knowledge of, but it was notseen as knowledge leading to salvation. Nevertheless, it is clear thatTimbuktu scholars took a keen interest in history. Although thechronicles we have were not put together until well into the seven-

131 See Hunwick (1984-85).132 See TF, 108.133 See Appendix 6 below. Askiya Dwüd is said to have established libraries, and employed

scribes to copy manuscripts for him, some of which he gave to scholars. He is also said to havememorized the Qur√n and to have studied the Risla of Ibn Abı Zayd al-Qayrawnı; see TF, 94.

teenth century, both Ibn al-Mukhtr and al-Safidı based theirnarratives of fifteenth and sixteenth-century history on accounts(akhbr) that were handed down orally or in written form from theseperiods. Ibn al-Mukhtr enlarges on the history written by hismaternal grandfather Ma˛müd Kafitı around 1593, and addsinfor–mation from some of his maternal uncles. He also quotes fromthe now lost Durar al-˛isn fı akhbr mulük al-südn. Al-Safidı,though he rarely alludes to his sources, must have gathered thedetailed information that went into his history from an existingTimbuktu tradition of historiography, again partly oral and partlywritten, which included material going back to the early fourteenthcentury, and even, though sketchily, before.The burgeoning of a tradition of learning in Timbuktu in thesixteenth century is certainly related to that city’s growth as acommercial centre, and to the relative political stability that the askiyadynasty brought to the Middle Niger. Timbuktu’s prosperity,grounded in the gold and salt trades and enhanced by its overall roleas gateway to the Middle Niger, meant that scholars could afford theleisure to study and teach. Such conditions encouraged other scholarsto settle there and students to gather there for instruction. It alsomeant that books could be imported from North Africa (at greatprofit for the merchants),134 and scribal services paid for. TheSafidian conquest of Songhay in 1591 and the establishment ofTimbuktu as the headquarters of the Arma administration135

disrupted the city’s equilibrium, and, insofar as the purpose of theconquest was to lay hands on the region’s wealth, it had deleteriouseffects on the city’s economy. The scholars, whom Pasha Ma˛müd b.Zarqün thought had been behind an uprising in the city, weresystematically plundered, and those of the Aqıt family were exiled toMorocco.136 The effect on Islamic scholarship was immediate.Although A˛mad Bb eventually returned to Timbuktu in 1608, allhis exiled relatives died in Morocco. After his own death in 1627there was scarcely another scholar of note in the city until thenineteenth century.137 The tradition of teaching continued, but it isclear that standards fell. A glance at the text of TS shows howinadequate its author’s Arabic language training had been, to thepoint where his meaning is sometimes obscured.

Al-Safidı and his History

134 See below, 281.135 Arma is the name given to the Moroccan occupational force that ruled the Middle Niger

region after 1591. The title is derived from the Arabic term ar-rumåh, meaning “archers” or“marksmen”.

136 See below, 218-20.137 Al-Safidı, in fact, describes Mu˛ammad b. A˛mad Baghayogho al-Wangarı, d. 1066/1655

as the last of the great shaykhs of Timbuktu; see TS, 322.

fiAbd al-Ra˛mn b. fiAbd Allh b. fiImrn al-Safidı, author of theTa√rıkh al-südn was born on 30 Rama∂n 1004/28 May 1594, anddied at an unknown date, sometime after 1065/1655-56, the last dateto be mentioned in his chronicle. His father’s male line was traced tothe Arab tribe of the Banü Safid, though the family had been settled inTimbuktu for several generations. Unlike the rulers of Morocco, whowere also of the Banü Safid, he did not claim descent from theProphet. Nothing is known of his youth, but at some time heevidently took up residence in Jenne, since in 1036/1626-7 hebecame imam of the Sankore mosque of Jenne. In mid-life he wasemployed by the Arma bureaucracy,138 initially in the administrationof Jenne and the Msina region of the Inland Delta. In 1056/1646 hebecame chief secretary to the Arma administration in Timbuktu.His chief claim to fame is his history of Timbuktu and the MiddleNiger, simply entitled Ta√rıkh al-südn. This work, in thirty-eightchapters, is mainly concerned with the history of the Songhay empirefrom the mid-fifteenth century until 1591 and the history of theArma of Timbuktu from that date down to 1655. The latter periodoccupies about half of the work. The early chapters are devoted tobrief histories of earlier Songhay dynasties, of imperial Mali and ofthe Tuareg, and to biographies of the scholars and saints of bothTimbuktu and Jenne. His acknowledged sources are few. For theseventeenth century he relies mainly on personal knowledge,evidently supported by notes (there are several chapters of obituariesand noteworthy events), and on records of the Arma administration;for earlier periods he rarely mentions his sources, other than‘trustworthy persons’ or ‘one of my colleagues’. He does, however,cite Ibn Ba††ü†a, the anonymous al-˘ulal al-mawshiya and, for someof the biographies of Timbuktu scholars, the biographical dictionaryof A˛mad Bb, Kifyat al-mu˛tj. For the period after about 1610we may assume that much of his information was gained first-hand.The long chapter 35 is largely a first-hand account, and is sometimesfrankly autobiographical.As noted above, al-Safidı’s command of Arabic was sometimesinadequate for the task before him. Certain phrases that are calques ofSonghay phrases suggest that he thought in Songhay while writing inArabic. Although he claimed Arab descent, Songhay was the linguafranca of the Middle Niger region, and it is likely that he grew upspeaking Songhay as his first language. His Arabic prose aims atbeing formally correct ‘classical’ (fu߲) Arabic, and he oftenborrows phrasing from the Qur√n or pious usage. Nevertheless, hisstyle lacks elegance and is often grammatically faulty, sometimesincluding Moroccan colloquialisms. Like many writers of Arabic of

138

the period he overuses the third person masculine singular pronoun,so that it is sometimes difficult to know who did what to whom. Hiswork, nevertheless, has one great merit; unlike contemporaryMoroccan chronicles, it is quite uncluttered by rhetorical excess.Apart from occasional moralising reflections, al-Safidı pursues hisnarrative history with dedication, and by and large with rationalorder. From the reign of Sunni fiAlı onwards, his chronicle is purehistoire événementielle, with the addition after the year 1612 ofperiodic chapters recapping major events and recording obituarynotices of scholars and important personalities.139 It is in thesechapters that we learn a few details of al-Safidı’s siblings and parents-in-law.Al-Safidı's approach to history reflects his training as a member of thescholarly class. He views events through the eyes of this group. Whatbenefits them is the subject of his approval; what is to their detrimentis to be deplored. This is most clearly seen in his denigration (albeitwith some hidden undertones of admiration) of Sunni fiAlı, and hishighly flattering portrait of Askiya al-˛jj Mu˛ammad. While helooks back with some nostalgia to the era of the Songhay empire, heis careful not to present any outright criticism of the Moroccaninvasion, or of the Arma regime by which he was employed. Ratherthan condemning the Moroccans for their invasion of Songhay andthe overthrow of the Askiya dynasty, he attributes this change offortune to the decadence of the ‘Songhay folk’, viewing it as justdivine retribution. With the exception of the largely borrowedbiographical sketches of Timbuktu scholars, Al-Safidı's style is in themain annalistic, after the manner of Arab chroniclers such as al-‡abarı, Ibn al-Athır, and (at least in his general history) Ibn Khaldün,though it is sometimes leavened by anecdotes and conversationaldramatizations that have their roots in an oral story-telling art.Although the assumptions al-Safidı makes of his audience create somedifficulties for the lpresent day reader, his work is a rich resource tobe mined, and each re-reading of the text offers new insights.Altogether, despite its stylistic infelicity, the Ta√rıkh al-südndeserves to be ranked as one of the great African chronicles. Withoutit, our knowledge of the workings of one of Africa’s greatest pre-modern empires would be considerably diminished, and ourunderstanding of a notable Islamic civilization much impoverished.Indeed, the existence of his work helps Timbuktu to cease to be seenas just a legendary fantasy, and helps it to be recognized for what itreally was - a spiritual and intellectual jewel inspired by the Islamicfaith.

139 This feature was to develop into a genre in its own right, especially in Mauritania.