1Located at the confluence of the Niger River and the Tilemsi Valley below the Ahaggar Mountains via the Adrar of Iforas, Gao was one of the oldest and most important commercial centers in West Africa. Since the beginning of the first millennium bc, Gao was linked with North African coastal cities by the “route des chariots” that Henri Lhote (1958: 24-25) reconstituted from the rock paintings dispersed in the mountains of the central Sahara.
- 1 In the “medieval” Arabic text between the 8th and the 16th centuries, Gao was called Kawkaw or Kūkū
2The strategic importance of Gao did not diminish during the historical period. Early Arabic texts concerning Bilād al-Sūdān never fail to mention it. The text of al-Khuwārizmī written before his death in 846-847, which is the oldest extant document referring to West Africa, cites Gao1 as one of the mains towns of West Africa, together with Ghāna and Zaghawa (Levtzion & Hopkins 1981: 8). According to al-Ya’qūbī who wrote in 872-873, Gao was “the greatest of the realms of the Sūdān, the most important and powerful. All the kingdoms obey its king. Al-Kawkaw is the name of the town” (ibid.: 21).
3We are informed also that Gao was, like the capital of ancient Ghāna, the twin cities composed of a royal residence and a commercial town inhabited by Muslim merchants. Based on the account of al-Warrāq who died in 973, Al-Bakrī gives us a decent report of these cities. This passage merits citation as it touches the heart of our article.
- 2 Tādmakka which means “Mecca-like” is identified with the actual archeological site of Es-Souk locat (...)
- 3 “Nile” in the “medieval” Arabic text concerning Bilād al-Sūdān refers to the Niger River.
“Between Tādamakka2 and the town of Kawkaw is a distance of nine stages. The Arabs call the inhabitants of the latter the BZRKĀNYYN. This town consists of two towns, one being the residence of the king and the other inhabited by the Muslims. The king is called Qandā. […] They worship idols as do the other Sūdān. When their king sits down a drum is beaten, the Sudanese women dance with their thick hair flowing, and nobody in the town goes about his business until he has finished his repast, the remnants of which are thrown into the Nile3. […] When a king ascends the throne he is handed a signet ring, a sword, and a copy of the Koran which, as they assert, were sent to them by the Commander of the Faithful. Their king is a Muslim, for they entrust the kingship only to Muslims.”(Levtzion & Hopkins 1981: 87)
4This passage states that the king of Gao was a Sudanese Muslim, but that the people nevertheless maintained their old religious practices attached to the Niger River. Al-Bakrī also asserts that the royal residence of Gao is situated “on the opposite bank” of the “Nile” (ibid.: 85). The location of early Gao on the west bank of the “Nile” was also affirmed by al-Muhallabī in the late tenth century, cited by Yāqūt. According to him, the king of Gao had two towns, one called Sarnāh dedicated to markets and trading houses, and another for him and his trusted associates which was sited “to the west of the Nile” (ibid.: 174).
- 4 Ibn Hawqal declares: “The route from Egypt to Ghāna went over them but the winds blew continually u (...)
5These descriptions are as important as they are the only available information on the early days of Gao. After this, unfortunately, Arabic writers lost interest in Gao and turned to Ghāna, where gold was so abundant that it was said that gold “grows as carrots do” (ibid.: 28). This change must have been the result of the westward shift of the main trade route for gold which linked Ghāna directly to Sijilmāsa, as described by the late tenth-century writer Ibn Hawqal who worked for the Fatimid caliphate that occupied the latter city4. After the writings of al-Bakrī in 1068, no valuable information was recorded on Gao until Ibn Baţţūţa visited it in the mid-fourteenth century.
6We thus lack sufficient documentation on Gao from the eleventh to the mid-fourteenth centuries. It is this scarcity of information that has provoked so many controversies on the early days of Gao. The location of the royal residence, its weight in the trans-Saharan trade, and the list of kings of the early Gao kingdom are among the subjects at issue.
7The identification of “Sarnāh” of al-Muhallabī with the actual archaeological site of Saney situated seven kilometers east of Gao is unanimously recognized (Fig. 1).
Fig. 1. Gao, Gao Saney and Waji Gangaber
- 5 Lange (1994: 284) cites a letter from Raymond Mauny who had visited Gao in the 1950’s to carry out (...)
8But this is not the case for the royal residence of the twin cities of Gao. Some say that it should have been located on the west bank of the Niger, following al-Bakrī and al-Muhallabī (Hunwick 1985: 5, 1994: 267). For others, “the west of Nile” of these writers should not be interpreted as the west bank of the Niger, but as the west bank of the Wadi Gangaber that flowed down in the past from the Tillemsi valley to separate Gao from Gao Saney (Lange 1991: 258, 1994: 284)5.
- 6 Based on the mintage of the gold coins in North Africa which continued during the fourth and fifth (...)
9There are also diverse opinions on the beginning of the trans-Saharan trade in which Gao was involved (Fig. 2). Given the scarce information provided by the early Arab writers, some consider that the trans-Saharan trade cannot have intensified until the tenth century (Devisse 1972: 49-50). Others say that this view should be revised, to take account of the Ibādī documents that affirm the existence of communication between North Africa and Gao from the second half of the eighth century6 (Lewicki 1961; Hunwick 1985: 6, 1994: 257).
Fig. 2. Gao in the historical map of North West Africa
- 7 Historians have tried to clarify this point by comparing different kinds of historical documents. S (...)
10We are also faced with another polemic on the identification of the kings commemorated in the epitaphs of Gao Saney with the list of kings mentioned in the two Tarikhs of Timbuktu7.
11Why are there so many controversies about the early days of Gao? One of the reasons, as mentioned above, is the scarcity of information given by early Arabic writers. To this can be added two other reasons: the diversity of the historical documents on the early days of Gao, and the lack of extensive archaeological research in Gao and its surroundings.
- 8 A complete list of these epitaphs was given for the first time by de Moraes Farias (2003) who trans (...)
12In fact, there exist three categories of historical records that provide information on the early history of Gao. The first is the writings by medieval Arabic authors, which modern historians generally consider to be primary sources. Nevertheless, these are only external and second hand documents, being written by authors who had never been to West Africa, with the exception of Ibn Battuta who visited the Mali Empire in the fourteenth century. The second is the two Tarikhs of Timbuktu, which were written by local authors and therefore provide good and approved documentation (es-Sa’di 1981; Kati 1981). Nevertheless, as they were written in the seventeenth century, they have the defect of not being contemporary with the early days of Gao. The third is a considerable number of epitaphs discovered in the region of Gao (at Gao, Gao Saney, Es-Souk and Bentya). These epitaphs have the high advantage of being contemporary and internal records, kept untouched since then8. Unfortunately, the information they yield is destitute of contextualization so that it needs to be related to other information given by other documents. The differences in emphasis of these historical records have made quite controversial the interpretations as for the early days of Gao.
13If the historical records cannot fulfill our requirement for historical reconstruction, archeological research would furnish a solid foundation for writing the history of Gao. In fact, number of research projects have been carried out since de Gironcourt (1920) visited the area at the beginning of the twentieth century to engage in archaeological works. In the 1950’s, Raymond Mauny (1951) excavated in the northern part of Gao to find, according him, a mosque constructed by Mansa Musa on his return from Mecca in 1325 or 1326. Recently, British archaeologist Timothy Insoll (1996, 1997, 2000) carried out two archaeological researches in Gao city and its surroundings in 1993 and in 1996. These excavations did discover some important materials that shed light on the economic life of early Gao. They did not, however, succeed in yielding a sound foundation for tracing the process of evolution of the Gao kingdom, as the research was limited and short-term.
14As for Gao Saney, the archaeological works of Colin Flight (1975, 1979) in the 1970’s was of great importance. Through his excavations in 1972, 1974, and 1978, he could identify the buildings made of fired-brick in two different places: one in the Muslim cemetery of Gao Saney and another in the southern fringe of the mount (Fig. 3). As for the dating, he considered these buildings to be “of about 1100” (Flight 1979: 36). Disturbances in the region, however, made him unable to extend his planned excavations at the principal archeological site of Gao Saney.
Fig. 3. Plan of the site of Gao Saney
15To compensate these historical and archaeological deficiencies, we began to engage in archeological investigations in the region of Gao in 2001. In 2001-2002 season, we conducted two excavations on the tell of Gao Saney. From 2003-2004, we moved to Gao city (so-called Gao-Ancient) to obtain a comparative view. In Gao-Ancient, we found two large buildings constructed completely of stone in the area where, local people say, Kankan Musa had constructed his mosque on his return from Mecca (Fig. 4), as was mentioned in Tarikh al-Sūdān (es-Sa’di 1981: 14).
Fig. 4. Plan of the site of Gao-Ancient (2005)
16These buildings are so immense that we have not yet finished exposing them (Fig. 4 & 10). The objective of the present article is therefore relatively limited. Its main purpose is not to provide a sufficient account on our field researchs, as we are preparing a book in which readers will find full data and a complete analysis of the materials found in our excavation. Our objective here is to offer sufficient data to give a new light on the formation of the early Gao kingdom and its implications for the trans-Saharan trade.
17Gao Saney is located seven kilometers east of Gao city, surrounded by the Wadi Gangaber downstream from the Tillemsi valley, which in the past must have flowed into the Niger River at Gao (Fig. 1). This site is very popular among historians of West Africa because of its royal cemetery, which contains notably some marble epitaphs. These epitaphs were discovered in the 1930’s and have since been the subject of much scientific work (Sauvaget 1950; Viré 1958; de Moraes Farias 1990, 2003). Close to this cemetery, there exists a tell 700 m long, 400 m wide, and 11 m high, totally covered with archeological remains such as ceramics, glass beads, and iron and copper fragments. These remains are associated with up to 2,000 extensive clandestine diggings to obtain glass beads.
18The devastation of the site made it difficult for us to find a place to excavate, when we targeted it in 2001-2002 season. Nevertheless, careful exploration of the surface of the tell enabled us to find two undisturbed areas, one on the top of the tell, and another about 100 m away on the northern flank, which slopes down one meter from the first site. As the workable space was limited, the excavation at the top (hereafter gs-3) measured two meters by three meters and that on the northern slope (hereafter gs-1) measured three meters by three meters (Fig. 3). We changed the level according to the natural layer of soil. After two months of work, we arrived at culturally sterile soil. The depth of gs-1 was 6.35 m at the fifteenth level, and that of gs-3 was 7.30 m at the seventeenth level.
19In terms of the stratigraphy of gs-1, the upper part down to 1.60 m seemed to have been used as a dump. There were a large quantity of remains such as shards of pottery, glass beads, and small clay crucibles, but no structural remains. The soil was a mixture of yellow sand and clay, containing much ash. This phase (phase 1) was followed by another (phase 2) that consisted of yellow sand and clay mix, 4.75 m in depth. Some rectangular banco bricks were found there, but no visible structures remained.
20In contrast, the gs-3 trench uncovered three layers of structural remains. After removing disturbed soil 0.35 m in depth (phase 1), we found the first structure made of rectangular banco bricks 0.6 m long and 0.3 m wide (Fig. 5).
Fig. 5. Construction made of rectangular dried banco bricks
- 9 The origin and the dissemination of the African rice (Oryza Glaberrima) were fully discussed by Rol (...)
21This structure would have been a solid one with walls 0.65 m thick. Beneath this structure, at a depth of 2.28 m, we found two convex thin banco walls of 1 m long, 0.5 m wide and 0.05 m thick. These structures were filled with incinerated Glaberrima rice of West African origin9, and would have been used as corn lofts. After this, appeared the remains of another structure with wall 0.4 m wide, made of rectangular banco bricks. The total depth of this phase containing three layers of structures (phase 2) was 2.60 m. Its soil was yellow sand and clay mix in distinguishable from that of phase 3 beneath. In the latter phase we found a large number of shards of pottery and beads, with no structural remains.
- 10 Some “collapsed rectangular bricks” were found at Jenné-Jeno from the structure whose dating is 103 (...)
22The discovery of the structure made of rectangular dried mud bricks is interesting, as this building technique is unfamiliar in “medieval” West Africa. The houses of the archaeological site of Jenné-Jeno from the third century b.p. to the fourteenth century a.d. were generally made of cylindrical dried mud bricks10 (McIntosh & McIntosh 1980: 104-110; McIntosh 1995: 64-65). The people of Jenné still distinguish these two kinds of brick, calling rectangular mud bricks “tubabu ferey”, which means “white men’s bricks”, and cylindrical bricks “jeney ferey”, “Jenné’s bricks”. This suggests that the diffusion of the rectangular mud brick in the region of Jenné took place in the last decades of the nineteenth century. As the Gao Saney site is considered to date from the eighth to the tenth centuries, this use of rectangular mud brick is one of the earliest in West Africa (see footnote 16 of this article).
23The findings from Gao Saney are composed, in the first place, of more than two thousands shards of pottery of which we analyzed a quarter, that is, 611 shards. Of these, 26.2% were classified as cord and twine-decorated, 25.3% as painted, and 13.6% as undecorated, while 34.9% were not classifiable. The cord decorated pottery, which frequently contains sand grains, is generally bigger and more rustic than the painted pottery. Our previous surveys in the Niger Bend demonstrated that the most frequently encountered Iron Age pottery is monochromatic, painted in light reddish brown or reddish yellow. At Gao Saney, however, much of the pottery is multicolored and more beautifully decorated than was standard (Fig. 6). The variety of decorations and diversity of manufacturing techniques suggest that either this site was occupied by peoples of different origins or pottery was imported from various areas. In spite of this diversity, the types of pottery were not differentiated in accordance with the depth of their finding. This implies that this site was not occupied over the ages.
Fig. 6. Some kinds of decoration of the pottery of Gao Saney
- 11 This information was provided by Mutsuo Kawadoko, Director of the Middle East Cultural Center, Toky (...)
24Among 121 shards of pottery of distinguishable form, 79% are bottles, 11% lids or dishes, 5% pots or jars. The high percentage of bottles is the most remarkable particularity of the pottery found at Gao Saney, because the ratio of bottles that our previous surveys conducted in Mali revealed is less than 2%. Other earthenware utensils include two earthen lamps of the same form as that used in North Africa (Fig. 7)11, and 26 earthen crucibles that would have been used to mould copper, glass or gold.
Fig. 7. Earthen lamp of Gao Saney in the same form as those of North Africa
- 12 H. Lhote (1972: 436) describes these “trois petites barres de cuivre à section hémisphérique, longu (...)
25In terms of metal objects, 333 fragments of copper were found in gs-1 and 116 fragments in gs-3. Most of these objects can not be identified, with the exception of a few bracelets and earrings. Some fragments are considered to be portions of semicircular copper-alloy ingot identified by Timothy Insoll (1997: 21) as currency. As he notes, one earthen mould and three copper ingots found at Marendet in northern Niger by Henri Lhote (1972: 436) in the 1960s seem to have the same form and the same dimension as the copper ingots of Gao Saney12. The excavation conducted by Lhote could expose more than 30,000 crucibles, dating from the sixth to the tenth century a.d. (ibid.: 449-450, 453). This corresponds almost exactly to our dating of the tell of Gao Saney. So, it is very probable that copper was imported from Marendet, located 800 kilometers east of Gao, in the form of semicircular ingots that would have been circulated as currency and melted down at need to make copper products.
26Iron slag was found in every layer of our two excavations. It is therefore evident that iron was manufactured in large amounts at this site. The 195 iron items found include 24 fragments of swords or knives, more than 100 nails, and two sets of two big rings attached to each other. The latter can be considered to be portions of harness.
27The most important, and most impressive, finding from Gao Saney is the many different kinds of beads. Of 388 beads in total, there are 11 bone beads, 102 earthenware beads, 22 beads of precious stone such as carnelian and crystal, 243 glass beads. The glass varies in color from dark blue to bright blue, grayish-white, yellow, green and red in quantitative order. The glass and stone beads were undoubtedly imported from North Africa and elsewhere, whereas the earthen and bone beads would have been local. As some of the glass beads found are in the process of manufacture, the inhabitants of Gao Saney must have not only imported but also remade glass beads.
- 13 These samples, numbers GS1-2 and GS3-1, are too early in relation to other samples. It is possible (...)
- 14 The date proposed in this study is 1000±70 B.P. analyzed in the mid 1980’s (Raimbault & Sanogo 1991 (...)
28During our excavations, no drastic change was noticed regarding either the soil or the pottery according to the depth of the layers. This led us to believe that the occupation of this site must not have been long-standing. This idea was confirmed by the C14 analysis of seven pieces of charcoal sampled during our excavations (Fig. 8). Although the dating of two samples appears anomalous13, if these are discounted, the deepest layer indicates dates of 753±37 a.d. and 776±42 a.d. and the layer above 877±54 a.d. and 920±42 a.d. We can therefore date the occupation of this site from the early eighth century to the mid-tenth century. This dating is almost three centuries earlier than that of between 1042 and 1280 or 1299 deduced from the epitaphs of the cemetery of Gao Saney (Sauvaget 1950; de Moraes Farias 1990, 2003: xxxiv, 32). Our dating, however, corresponds approximately to another proposed in 1980’ by researchers of the Malian Institut des Sciences Humaines, which indicates the late tenth century14. We may therefore situate the occupation date of the tell of Gao Saney between the early eighth century and the late tenth century.
Fig. 8. Radiocarbon ages of the site of Gao Saney
29The most outstanding feature of the Gao Saney tell is the overwhelming influence of North African civilization. In fact, our 15 m2 excavation produced 388 beads, including 243 glass beads, an exceptional number in terms of archeological surveys carried out in West Africa. The Jenné-Jeno excavations could find only several dozen of glass beads in total (McIntosh & McIntosh 1980: 164; McIntosh 1995: 250), and the Kumbi Saleh excavation of 130 m2 found 220 glass beads (Berthier 1997: 91). As it is generally accepted that glass was not manufactured in sub-Saharan Africa, this assemblage of glass beads places Gao Saney as one of the most important centers of trans-Saharan trade in West Africa from the eighth to the tenth centuries.
- 15 T. Insoll and T. Shaw build up a hypothesis that the splendid findings of Igbo-Ukwu in the Southern (...)
30No less impressive are the existence of the 26 crucibles that would have been used to melt down of glass or gold, of the slag and iron goods found at every level of our excavations, of the more than 300 copper items, and of the very sophisticated pottery of different kinds. All these findings suggest that Gao Saney was not only a mercantile town but also a manufacturing one, as proposed by Insoll (1997: 23). If this reading is accepted, our findings will cast new light on industrial development in West Africa from the eighth to the tenth centuries15.
- 16 It should be compared with the cases of Tegdaoust and Kumbi Saleh, which are located in the Sahara (...)
- 17 The excavations at Tegdaoust discovered more than 100 spindle whorls, only in a layer dating its la (...)
31Moreover, the use of rectangular mud bricks is not only exceptional, but also the earliest in West Africa16. Also exceptional are earthen lamps of the same shape as those used in North Africa, sophisticated spindle whorls that are among the earliest in West Africa17, and the high and exceptional percentage of bottles among the pottery found at this site. All these peculiarities suggest that the main inhabitants of Gao Saney would have been merchant-manufacturers from North Africa who would have coexisted with the local population, as suggested by the divers forms and techniques of pottery.
- 18 According to Michael Brett (1969: 357), “Saharan trade along three routes may be envisaged as large (...)
- 19 In 1352, Ibn Battūta met “a company of white men who are Kharijites of the Ibādī sect called Saghan (...)
32According to Tadeusz Lewicki (1961, 1971), a specialist in Ibādī history, the first agents of the trans-Saharan trade were Ibādī merchants expelled from Iraq in the second century of Islamic expansion. The Ibādī sect, which claimed the equality of all Muslims, was welcomed by the Berber tribes who had been oppressed by the Arab-centered administration of the Umayyad dynasty. Supported by them, the Ibādites accumulated power on the northern fringe of the Sahara and established Rustam dynasty in Tāhart in 776. By the last decades of the eighth century, they must have opened a commercial route to Sudanese cities such as Ghāna and Gao (ibid.: 1961, 1971: 118)18. The relationship between Tāhart and Gao must have been very close, as is illustrated by an anecdote reported in an Ibādī document, which says that a son of the second imam of the Rustam dynasty, al-Wahhāb (784-823), tried to go to Gao (Levtzion & Hopkins 1981: 90). It will therefore not be too great a stretch of the imagination to conceive that the Ibādī merchants-manufacturers would settle in Gao Saney19 in the second half of the eighth century, as suggested already by John Hunwick (1985: 6).
33In 2003, we decided to turn to Gao city to expand our understanding through a comparison with our discovery at Gao Saney. The northern area of Gao city is called Gao-Ancient, and many archeological sites remain there, including the monument of Askia Muhammad built in the last decade of the fifteenth century. In this area there is an open space of about 40,000 m2 where, according to local tradition, the emperor of Ancient Mali, Mansa Musa, constructed a mosque on his return from Mecca in 1325 or 1926, as noted in Tarikh al-Sūdān (es-Sa’di 1981: 11). Some French colonial officers and Raymond Mauny (1951: 844) carried out archaeological investigations there in 1949 and 1950 to find “une mosquée à mirhab” constructed with dried mud bricks. As Mauny found nearby a tombstone dated to 1362, he concluded this structure to be the remains of the Mansa Musa’s mosque (Fig. 4).
- 20 During his excavation in 1996, T. Insoll (2000: 4-8, 26-27) hit one end of the stone walls. Unfortu (...)
34Besides this mosque, Mauny noted some structural remains made of stone, one of which measured 30 m long and 1.2 m thick. Believing these stones to be the foundations of modern houses, Mauny affirmed that “Des sondages effectués à plusieurs points de cet important bâtiment m’ont donné des résultats négatifs: il ne s’agit pas de murs à proprement parler, mais de fondations” (ibid.: 845). We do not know if Mauny actually carried out excavations there to make this declaration. What we know is that this claim has gone unchallenged. In spite of their apparent importance, nobody has tried to find out what exactly these stones structures represent20.
35After having swept all the surface and exposed structures of this open space, we chose one area in which to dig. This area was defined by the ends of two parallel ranges of stones 2 m apart. We envisaged them to have formed the entrance of an ancient building, and decided to establish one unit of 5 m by 5 m to include all the visible stones. As the trench was large, we decided to change level every 20 cm. Several days’ work brought us discoveries that surpassed all our expectations (Fig. 9).
Fig. 9. Entrance of the big building of Gao-Ancient
36The stone structures proved not to be building foundations, as proposed by Mauny, but the actual walls of a big building buried underground. The entrance was decorated with fine fired bricks. The thickness of the walls was consistently 1.2 m, made of a combination of flat schist slabs and round sand stones, reinforced by clay plaster. We dug down to the bottom of the walls at 1.2 m deep. During our excavation of this first unit, we realized that the building was much bigger than we had expected. We therefore decided to put the highest priority on obtaining an overall view of the building by opening other units in the direction of the stone-wall structures.
- 21 A modern house now stands on top of the western wing of the building, making it impossible to enlar (...)
37Our excavations in this season took two and a half months to give us an approximate plan of the building (Fig. 10). The final excavated area of this season, 28 m wide and 20 m long, was unable to uncover the whole structure. The thickness of the wall is consistently 1.2 m, made of a combination of flat schist slabs and round sand stones. The distance between the walls is not large, which varies from 2.5 m to a maximum of 4 m, with no arches used in the building technique in West Africa. The eastern wing of the building is composed of four ranges of stone walls, so the building proves to be very immense21. Inside and outside the building were several walls and floors made of fired or dried mud bricks, which, owing to differences in depth, we interpreted as the remains of later housing that would have reused the abandoned stone walls.
Fig. 10. Plan of the buildings of Gao-Ancient
2010, The central units (GKM1-GKM8) were excavated in the 2003-2004 season.
- 22 Bentya is a small village located 140 km south of Gao, on the route to Niamey. Bentya is famous for (...)
38As for the two kinds of stones used construction, sand stones can be easily found on the river bend of the Niger River near Gao, but flat schist slabs are totally absent. When we visited Bentya-Kukiya22, 140 km away from Gao and considered as the first capital of the Gao kingdom according to the local tradition collected by Jean Rouch (1953: 156-165), we discovered some structural remains made of the same kind of stones. Gao being linked with Bentya by the Niger River, it is very probable that these stones were carried from there by boat. Did the rulers who had left Bentya, their old capital, to settle in Gao-Ancient tried to construct their new residence with the same materials? This seems very plausible, but will remain a hypothesis until an extensive survey will be carried out in the region of Gao including Bentya. In any case, an immense mobilization of labor would have been needed to transport a large amount of stones, as Gao is 140 km from Bentya and the building in Gao-Ancient is very immense.
39In relation to the scale of the construction, our findings during this season were relatively limited. More than 400 pottery shards were found, with decorations almost the same as those of Gao Saney, except for black or brown burnished wares of the finest quality. The most marked difference in the forms of pottery found was the scarcity of bottles in Gao Ancient in comparison with Gao Saney. Other findings included several dozen spindle whorls, a large quantity of iron products, some copper articles, and very few iron slag. The number of glass beads was relatively limited (493 items) for the site’s size, while a dozen large carnelian beads were found. In spite of these findings of great value, no dump and very few charcoals were remarked during our excavations.
40Excavations in Gao-Ancient have continued year in year out since 2003 with some interruptions. The eastern wing of the big building was finally uncovered to reveal four ranged walls 36 m long. These walls are 2.5 m apart, with the right wing measuring 12.5 m from front to back. If the left wing has the same length as the right wing, the total length of this building will be 73 m. In terms of the height of the building, actual houses constructed with dried mud brick walls 0.4 m thick are generally 3 m high. So if we treble this figure in accordance with the triple thickness of the wall, the building would stand over 10 m high. The immensity of the building and the mobilization of massive labor force to construct it made us to think of the citadel or royal residence mentioned by al-Muhallabī and al-Bakrī of the tenth and the eleventh centuries (Levtzion & Hopkins 1981: 87, 174).
41On the north boundary was found another building containing some rooms. This building, or rather residence, is smaller than the other. Its main room is 11 m long and 5.5 m wide, with walls 0.5 m thick made with flat schist slabs (Fig. 10). Inside the room are seen two ranges of columns made of smaller schist slabs (Fig. 11).
Fig. 11. Interior of the small building of Gao-Ancient
42Two layers of white and red plaster are painted inside the walls, meaning that this room must have been very beautifully decorated. This residence contains two other rooms, one of which is a bathroom equipped with a drain pipe under a corner of the floor made of pebbles (Fig. 12).
Fig. 12. Bathroom of the small building of Gao Saney
43How were these two buildings constructed? The walls of the bigger building are found to be set on a foundation made of banco, 0.3 m thick and tightly beaten. The foundation of the other building is not so thick or tight. The fact that the wall of the small residence is deeper than that of the bigger one by 20 cm gives rise to two possible interpretations: either both buildings were constructed at the same time, or the smaller building was built first, followed by the other after an interval of several years or decades. In any case, the buildings are situated so close to each other that the bigger building must have been built while the preceding residence was occupied.
- 23 Sample number GMK4-6 was taken from the excavation of the bigger building, and sample number GMK11- (...)
- 24 T. Insoll (1995: 327) found 53 hippopotamus ivories in a layer called the “fifth period” (1.60 m in (...)
44For dating, seven charcoal samples were taken from the two buildings (Fig. 13). The charcoals of sample number gmk4-6 and gmk11-7 are taken from the deepest layer over the floor23, and those of gmk4-5 and gmk11-6 from the next deepest layer. Their dates indicate that the two buildings were constructed in the first quarter of the tenth century (between 908±50 a.d. and 938±40 a.d.) and abandoned in the last decade of the same century (between 978±50 a.d. and 1018±40 a.d.)24. The other charcoal samples, numbers gmk11-4, gmk20-3, and gmk29-3, are taken from the layer at almost the same depth as the annexed floors made of fired or dried mud bricks. We interpret these floors as the reuse of the abandoned stone walls, meaning that the posterior constructions equipped with these floors were built in the last half of the eleventh century or the first half of the twelfth century24.
Fig. 13. Radiocarbon ages of the site of Gao-Ancient
- 25 During his excavation in Gap-Ancient in 1996, Insoll hit a corner of these stone walls. Unfortunate (...)
45This dating of the stone buildings of Gao-Ancient merits comparison with that of Gao Saney. In fact, the construction of Gao’s main buildings coincides with the later period of occupation of the tell of Gao Saney from the eighth to the tenth centuries. It is therefore clear that the buildings of Gao-Ancient and those of Gao Saney coexisted during the tenth century. This agrees exactly with the writings of al-Muhallabī and al-Bakrī of the second half of the tenth century25, which affirm that the capital of the Gao kingdom was a twin cities, one called “Sarnāh” and dedicated to the markets and trading houses, and the other, the residence of the king located “to the west of the Nile” (Levtzion & Hopkins 1981: 84, 174). The collation of the results of our excavations carried out in Gao-Ancien and Gao Saney offers evidence for the assertion that “Sarnāh” is the actual site of Saney and that both parts of the twin cities of the Gao kingdom were located on the east side of Niger. The “Nile” of these writings should not be interpreted as Niger River, but as Waji Gangaber which separates Gao to its west from Saney to its east (Fig. 1), as was interpreted by Dierk Lange (1991: 258, 1994: 284).
- 26 Al-Bakrī finished a part of his book concerning the Bilād al-Sūdān in 1068, but he exploited the wr (...)
- 27 Among the glassware, two kinds of coloring are recognizable: greenish and colorless. Chemical analy (...)
- 28 The number of fragments of glassware and luster pottery found in Gao-Ancient is far greater than th (...)
46Both quantitatively and qualitatively, our findings in Gao-Ancient are more important than those of Gao Saney. Most of the valuable goods were found in the main room of the smaller residence, including a round brass trim, an iron sword inlaid with some brass asterisks, a bracelet made of several hundreds of brass small rings, and a probable small jewel box decorated with a brass ornament. There are also 40 fragments of luster pottery imported from North Africa and from China, some of which were typical pottery of Cairouan under the early Fatimid dynasty 416 fragments of glassware, including several dozen of small bottles that would contain perfume or balm, and one cupping glass for drawing blood (Fig. 14)26. These findings are important as nothing similar has been found in West Africa, even in large trading cities such as Tegdaoust and Kumbi Saleh27. Moreover, two small shreds of gold weighing 2.50 g and 2.25 g are found in the same room, indicating that the export of gold was practiced in Gao28.
Fig. 14. Glassware imported from North Africa or Middle East
47As for the pottery, some differences are remarked between Gao Saney and Gao-Ancient. The most remarkable difference is that a hundred pieces of black or blown burnished pottery of the finest quality are found in Gao-Ancient while the high percentage of bottles remarked in Gao Saney is not reproduced here. Other earthenware items include 46 spindle whorls of the same material and geometric decoration as the black burnished wares, whose provenance we cannot yet identify. The absence of bottles and earthen lamps that are the most characteristic feature of the pottery of Gao Saney and can be considered as the indicator of the existence of the North African people gives an idea that the inhabitants of Gao-Ancient differed from those of Gao Saney. In contrast, the diversity of decoration of the pottery found at Gao-Ancient suggests that Gao was a center for the exchange of the local pottery fabricated in different localities.
- 29 Gold has not been commonly found by the archeological surveys realized in the Gao region. Only one (...)
48The findings from Gao-Ancient are more important than those of any other archeological site in West Africa other than Igbo-Ukwu (Shaw 1970)29. The richness of the site is shown in the number of findings in each category: The total number of beads is 9,743, including two bone beads, 34 metal beads, 114 earthenware beads, 150 precious stone beads (agate, carnelian, crystal, etc.), and 9,411 glass beads. There are also 2,098 pieces of iron, 1,889 pieces of copper, 416 fragments of glassware, and 41 fragments of luster pottery imported from North Africa, the Middle East, or China, but few iron slag. The ratio of the number of findings from the small building to those from the larger one is roughly 4:1. This ratio is even higher for precious items such as copper articles, glassware and luster pottery.
49The tenth-century main buildings made of stone on a scale never before seen in the West African savannah, the centralized power required to mobilize massive labor for the transportation of schist slabs, and the intensive transport by camels of precious and very fragile goods such as small glass bottles, all lead us to deduce that the small building of Gao-Ancient was a royal residence protected by a kind of large stone-built citadel. This interpretation seems more probable given the fact that no dump and a very limited quantity of iron slag were found here. It demonstrates that this area was not occupied by the common people, but reserved for the nobility who did not need to work or cook.
50Our excavations at the archaeological sites of Gao-Ancient and Gao Saney have produced many discoveries more important and more controversial than we had ever expected. Some of them, notably the new dating of the two sites and the position of the royal residence of the early Gao kingdom, diverge from commonly accepted views. Yet, if our interpretations are accepted, they will surely deepen and widen our understanding of the process of the formation of the early Gao kingdom. Which kind of new light can our discoveries cast on the history of Gao and of West Africa? Four points are to be noted.
- 30 T. Insoll and T. Shaw (1997) tried to compare the findings from Gao and Gao Saney with those from I (...)
51Firstly, our discovery of the royal palace of the tenth century, if our interpretation is accepted, is the first discovery of a “medieval” royal palace in West Africa. Since the beginning of the twentieth century, when West African historiography began (Shaw 1905; Delafosse 1912), many efforts have been made to find the capitals of ancient Ghāna, Mali, and Gao kingdoms, without any success30. The twin cities of Gao composed of the royal residence and the Muslim commercial town, were mentioned by Arabic writers of the tenth and eleventh centuries such as al-Muhallabī and al-Bakrī. But the absence of any archeological proof until now has meant that these writings have caused controversy among modern scholars rather than shared understanding.
52Our discovery has shown the very existence of the twin cities of Gao in the tenth century, both located on the east bank of the Niger River. Furthermore, the scale of the buildings made of stone and the richness of the imported and locally manufactured items which exceed generally accepted concepts in “medieval” West African history suggest that the inhabitants of Gao-Ancient would be the holders of the centralized power. And their power must have been consolidated by the coexistence of the flourishing commercial and manufacturing center of Gao Saney where the merchants-manufacturers from North Africa would cohabit with the local population.
- 31 For excavations of ancient Ghāna, see Thomassey and Mauny (1950), Berthier (1997). For ancient Mali (...)
53Secondly, the new dating of Gao-Ancient and Gao Saney confirmed by our excavations will require a revision of the early history of the trans-Saharan trade. The epitaphs found at the cemetery of Gao Saney dating from 1042 to 1280-1299 have made scholars consider that the tell contiguous with it was contemporaneous, despite the lack of any archeological investigation there. This presumed idea must be revised in the light of the results of our excavations, which situate the occupation of the tell from the eighth to the tenth centuries a.d.31 Our excavations suggest also a high probability for the existence of a colony of North African merchants and manufacturers in this place.
- 32 This does not exclude the possibility that Gao Saney was inhabited earlier or later than this perio (...)
54A large quantity of glass beads, copper ingots, iron goods and iron slag, and miscellaneous earthenware including crucibles, spindle whorls and earthen lamps, all demonstrate the intensity of the trans-Saharan traffics and manufacturing activities in Gao Saney. If this place was inhabited by the merchant-manufacturers of North African origin as early as the eighth century, as we believe, the trans-Saharan trade must have existed in considerable intensity in this century32. To identify the main items exported from Gao is beyond the scope of this article. But gold and hippopotamus ivories shipped by the Sorko, a riverine branch of the Songhay, would have constituted an important part of the products exported to North Africa. Situated at the confluence of the Niger River and Tillemsi Valley and in the contact zone between the Sahara and the savannah, Gao must have been a privileged target for both foreign merchants and the local populations who wanted to engage themselves in the world-wide economy via North Africa.
55Thirdly, the confirmation that the occupation of Gao Saney preceded the construction of the main buildings of Gao-Ancient by two centuries will shed new light on the process of formation of the twin cities of Gao. The oral tradition of the Sorko collected by Jean Rouch (1953: 156-165) in the 1950s affirms that their forebears originated far to the south, in the W region on the frontier between modern Benin and Nigeria (Fig. 2).
56Their mythic ancestor, Faran Maka Bote, ascended the Niger River to settle in Kukiya-Bentya where he found the residents dominated by a demon in the form of a gigantesque fish. It was after Faran Maka Bote, or Za-al-Ayaman after Tarikh al-Sūdān (es-Sa’di 1981: 6), had killed this demon with a kind of harpoon that he was welcomed by the local population and became king there (Rouch 1953: 168; es-Sa’di 1981: 8; Kati 1981: 326). Thus, the first kingdom of Gao would have been founded in Kukiya-Bentya, which must have expanded its power through a monopoly of the fluvial activities of the Niger River. This legend may have reflected the historical evidence that the first kingdom of Gao emerged from the combination of the local agricultural population and the immigrant riverine people who monopolized the fishing, hunting of aquatic animals, and fluvial traffic of the Niger Bend and beyond.
- 36 The fact that the epitaphs of Es-Souk precede those of Gao Saney suggests that this custom was intr (...)
- 34 Because of the existence of the epitaphs in Bentya necropolis from 1272 to 1489 which coincide appr (...)
57We have no historical record of when this event happened. Yet, it is very probable that the later development of the small kingdom of Kukiya-Bentya was due to the intensification of the commercial and manufacturing activities of Gao Saney that had flourished since the eighth century. On the other hand, the consolidation of power of the small kingdom in Kukiya-Bentya would be felt to be beneficial to enlarge their settlement in Gao Saney by the foreign merchant-manufacturers who had decided to live there33. If this interpretation is accepted, it implies that the first twin cities of Gao consisted of Kukiya-Bentya and Gao Saney since the eighth century. Some decades or centuries later, the ruler of Kukiya-Bentya would decide to construct his new residence in Gao in the early tenth century at the latest, to engage his kingdom more deeply in the trans-Saharan trade and to acquire more of its benefits from it34.
58Fourthly, if the peak of the early Gao kingdom was achieved by the Songhay people during the tenth century when they constructed big buildings at Gao-Ancient, a question must be posed regarding its continuity: Is it the Songhay who held the hegemony of Gao since then? To answer this question, we must examine carefully the changes of the religious lives and of the material culture, observed in Gao-Ancient and Gao Saney. As shown above, the main buildings of tenth-century Gao-Ancient were made of stone, while the structures of the next layer above dating of the eleventh and twelfth centuries were equipped with floors made of fired mud bricks. This technique was unfamiliar in the Gao region, with the exception of the “caveaux” of the cemetery of Gao Saney and the “structure Q” in the southern fringe of the tell of Gao Saney, which were considered to be “about 1100” (Flight 1975, 1979: 36) (Fig. 3). What does this affinity between the later construction at Gao-Ancient and the “structures” at Gao Saney mean?
59This change of the building techniques should be related to another one realized in the religious lives: appearance of the epitaph in the Gao Region. In spite of the overwhelming influence of the Muslim civilization, the first epitaph appeared in 1013/1014 at Es-Souk and in 1042 at Gao Saney (de Moraes Farias 2003: xxxiii-xxxiv). The absence of epitaph before these dates would be due to the repulsion toward its construction by the Ibādites who monopolized the trans-Saharan traffics before the tenth century. Since the second half of the tenth century, nevertheless, the influence of the Sunnī and Shī’ī Muslims who were not against epitaph construction gradually got to predominate in this region (ibid.: cxlv-cxlvii). In fact, the Almoravids who were the advocates of Sunnī extended their power, after having conquered Sijilmāsa in 1054, to the north to occupy Almeria in 1091, as well as to the south to take Ghāna in 1076, and Tādmakka in 1083.
60Some studies propose that the marble epitaphs of Gao Saney, made in and transported from Almeria on the southern coast of Spain, were closely related to the Almoravid domination in the Gao region (Hunwick 1980: 420-423). So the marble epitaphs of Gao Saney, which were introduced for the first time in West Africa around 1100 to commemorate deceased kings and queens, represented exactly the two extremities of the Almoravid dynasty. As A. D. H. Bivar (1968: 7) has suggested, this new ruler of Gao who introduced the Spanish marble epitaphs would not be “one of the Almoravid princes but a federated or vassal local ruler”. He or she must have felt a great need for it, because his/her emerging authority would not have been recognized easily by the local population (ibid.: Hunwick 1980: 429-30).
- 35 This settlement process was described for the first time by Hunwick (1985: 5-7). Tarikh al-Sūdān co (...)
61If the new ruling class of Gao, of presumed Sanhāja origin35, introduced marble epitaphs from Spain, it is very probable that to consolidate their authority they also introduced other items more visible and more overwhelming: structure made by a new building technique, fired mud bricks. So, the new rulers established their royal houses in Gao-Ancient and Gao Saney by using a new building technique, while standing the epitaphs in the cemetery of Gao Saney between 1042 and 1280-1299. As a consequence, the ancient Songhay rulers of Gao-Ancient would be obliged to go down the Niger River far south. This new dynasty of provably Sanhāja origin, succumbing to the domination of Mali Empire in the Gao region during the mid-thirteenth and the fourteenth centuries, would retreat to Bentya where they constructed the epitaphs between 1272 and 1489 (de Moraes Farias 2003: liv). From far south the descendants or the compatriots of the ancient Songhay rulers were obliged to wait several centuries before they regained Gao with the assumption to the throne of Sonni Ali-Ber.
62❖
63Our excavations in Gao-Ancient and Gao Saney could furnish rich and solid materials that will deepen and widen an understanding not only of West African archaeology but also of the history of West Africa.
64The most urgent necessity imposed on us is to carry out further excavations in order to expose the rest of the buildings still covered under-ground at the site of Gao-Ancient. Our hope is to continue this task. Nevertheless, we have to recognize that this is not an easy task, as our excavation of about 1600 m2 at Gao-Ancient only represent ten percent of the possible site which may occupy about half of the open space of 40,000 m2, called site of Mansa Mūsa’s mosque.