West Africa between Rome and medieval Mali: the pre-Islamic sites of Loropeni, Kissi, and Oursi in Burkina Faso.

In 1996, the discovery of an elite necropolis at Kissi in Burkina Faso containing grave goods that included Roman items dated to the early 1st millennium CE provided the first definitive evidence for long-distance trade in luxury goods in pre-Islamic West Africa.

In the early 2000s, excavations uncovered evidence for pre-Islamic nucleated settlements with monumental architecture at the sites of Oursi and the impressive ruins of Loropeni. These later sites feature a double-storey house complex and a group of walled towns containing flat-roofed structures, all of which point to an independent origin for West Africa's architectural styles.

Just as recent research has pushed back the emergence of complex societies in West Africa dates to the late 3rd millennium BC at Dhar Tichitt, studies of the sites of Kissi, Oursi, and Loropeni show that cultural developments that were previously associated with the arrival of Islam, such as long-distance trade with North-Africa and monumental architecture, predated the emergence of the earliest Muslim societies such as Medieval Mali.

This article explores the latest research on the pre-Islamic sites of Burkina Faso and their significance in reconstructing the pre-Islamic history of West Africa between the late Roman empire and Medieval Mali.

(Top) South wall of Loropéni enclosure, looking north-east. (bottom). Reconstruction of Oursi Hu-beero looking north.

A brief background on the recorded history of Burkina Faso: The Mossi kingdoms against medieval Mali and Songhay 

For much of the pre-colonial era, the modern country of Burkina Faso was populated by mostly non-Muslim societies which were on the periphery of the better-known empires of medieval Mali and Songhai.

The powerful cavalries of the Mossi kingdom of Yatenga in northern Burkina Faso, for example, appear in the Timbuktu Chronicles as the only 'pagan' challenger of Mali and Songhai, sacking the cities of Timbuktu and Walata in 1343, the 1430s, 1477, and 1480, taking away loot and captives southward to their capital. Eventually, the punitive expeditions by the Askiya rulers of Songhai in 1483, 1498, and 1549, broke their power.

The Mossi of Yatenga were last mentioned defeating a Songhay force in 1561 and intimidating the armies of Askiya Dawud in 1570. The threat was such that the Askiyas created the office of Hombori-Koi to govern the province of Hombori at the Niger Bend, facing the hostile Yatenga-Mossi states.

(Timbuktu and the Songhay empire pg 11-12, 38 n.4, 39, 97-99, 106-107, 146, 150, 341)

This antagonistic relationship presented by the 17th century Muslim chroniclers based in the city of Timbuktu and Djenne however, conceals a long history of intercultural exchanges and regional trade that significantly predates the ascendance of the Muslim empires of Mali and Songhai. These exchanges, revealed by the archaeological discoveries at Kissi and Oursi, indicate that societies along and near the Niger bend were part of a broad cultural continuum that laid the foundations for the emergence of the empires that appeared in later Islamic sources.

Location of the sites mentioned in this essay;

• Loropeni, Oursi and Kissi in red; the medieval Muslim capitals of Gao (Songhay empire) and Kumbi Saleh (Ghana empire) are in green, as well as the city of Tadmekka (that was attacked by Ghana) and Timbuktu and Walata (that was attacked by the Mossi); the capital of Roman north Africa, Carthage, and Rome's southernmost province of Jarma are shown in orange. Included is the pre-islamic site of Marandet where some of Kissi's copper was likely obtained.

Map showing the empires of Medieval west Africa.

The elite necropolis of Kissi and its Roman connections. (300BC-1200CE)

The site of Kissi consists of a cluster of about a hundred Iron Age settlement mounds with various clusters of stone structures and a cemeteries that were permanently occupied between the 3rd century BC and 12th century CE. Spreading over an area of more than 300 hectares, the archaeological site lies on the northern shore of the Mare de Kissi lake, where sedentary agro-pastoralists were established in the 1st millennium BC.

(Burial and Society at Kissi, Burkina Faso pg 375-377)

Kissi was part of the many Iron Age societies of the western Sahel during the 1st millennium BC, as evidenced by its oldest site; Kissi 49, where Iron slag and related material were found dated between 410-200BC. However, significant occupation of the site began between the 1st and 4th century CE when ironworking and agriculture based on pearl millet and small livestock became fully developed, along with some regional trade.

(Some Aspects on the Iron Age sites of Kissi pg 80-81)

The Eastern Niger Bend and the location of Kissi, note the location of Oursi to its west. Map by S. Magnavita.

Settlement of the Kissi sites increased significantly during the period between the 4th to the 8th century CE, where most graves are dated and the area covered extended over 100 ha. The grave goods associated with the inhumated persons leave no doubt as to the cosmopolitan nature of life in Kissi. The location of the graves within the vicinity of inhabited areas, rather than completely separated from them, implies that the dead were held in high regard.

The graves of Kissi are marked by dressed stone slabs, similar to those found in other ancient West African cemeteries like at Oursi, at the city of Gao in the Niger Bend region, and even further at Dhar Tichitt. Many burials were accompanied by grave goods such as jewelry and weaponry, and there is a substantial contrast in the number of grave goods associated with some buried persons: with some graves having no grave goods, while a few contain an extremely numerous amount of objects.

(Burial and Society at Kissi, Burkina Faso pg 379, Some Aspects on the Iron Age sites of Kissi pg 83-85)

A typical mound at Kissi, and another one near the car in the background. the remains of domestic structures of the settlement were found at the foot of these mounds but were heavily eroded. Image by P. Breunig

Stone slabs marking graves at Kissi. Image by S. Magnavita

The vast majority of the grave goods were beads. A total of 5,000 were found during excavations in the cemeteries Kissi 3, 13, and 14; about half of which were made of semi-precious stones; while the rest were glass beads and iron beads. Besides beads, other common ornamental objects included cowrie shells as well as bracelets, rings, and anklets made of iron and copper-alloys.

Grave 10 in cemetery Kissi 3 in particular contained numerous iron rings and arrows, as well as two daggers, a sword, two brass anklets, four finger rings and several other rings, all made of copper alloy, a bead necklace consisting of 165 pieces, including carnelian and 39 glass beads and cowrie shells; as well as wool textiles were indicated within this grave.

The large arm daggers with curved blades that were likely shoulder blades based on where they were found. A common type of dagger was one with an iron handle forged into an oval ring and found next to the feet of the buried persons. All of these were dated to between the 5th to 7th century CE.

(Some Aspects on the Iron Age sites of Kissi pg 86-89, Burial and Society at Kissi, Burkina Faso pg 386-387)

A selection of different types of beads found in Kissi. Image by S. Magnavita.

Examples of looped daggers: a) Kissi 13, Grave 9; b) Kissi 3, Grave 10 (restored); c) an archer, equipped with bow, quiver and looped dagger, depicted on a ‘Benin-beaker. Images and Captions by S. Magnavita.

Metal jewellery. a: upper arm iron bracelets (Kissi 3, grave 10); b: lower arm iron bracelet (Kissi 3, grave 1); c: brass anklets (Kissi 3, grave 10); d: copper alloy ear(?) ring (Kissi 3, grave 14); e: copper alloy anklet (Kissi 14C, grave 7). Images and Captions by S. Magnavita.

Textile remains were also discovered in some of the graves, along with fragments of a woollen rope, which were likely preserved due to the large amount of metal objects in the graves which stopped their natural decay. All garments were manufactured in the plain weave technique and were made of wool or fine animal hair, and are dated to between the 1st and 7th century CE, making them the oldest preserved west African textiles so far. Its unclear whether they were produced locally or imported from further afield.

(The Early History of Weaving in West Africa 194-203)

Fragment of a musical instrument from Kissi, Burkina Faso, cemetery Kissi 3, Grave 11, with fine animal hair or wool fabric in weft-faced plain weave. Image by S. Magnavita

The artifacts recovered from the settlements and cemeteries hint at the kinds of commercial contacts the Kissi people were involved in: regional, interregional, and long-distance. The semi-precious stone beads likely originated from the Tilemsi valley of eastern Mali; some of the glass beads found at Kissi resemble those manufactured at the site of Ife in Nigeria that are dated to the 9th century CE, and were also found at Igbo Ukwu.

Some of the copper of Kissi corresponds well with that from the contemprenous site of Marandet in Niger and to an extent with that found at Igbo Ukwu. The other significant find were the glass beads, which, while mostly dated to the second half of the 1st millennium CE, were produced in the Near East on the eve of the Arab conquest or during its early expansion, with atleast one HAHL glass bead being made from Ife in Nigeria, which signficantly pushes back the date of early trade contacts along the Niger river.

(Some Aspects on the Iron Age sites of Kissi pg 90-94, Contacts between West Africa and Roman North Africa pg 130-133, Glass beads from Kissi)

The most impressive evidence for external contacts however, is presented by some of the metal objects which were most likely imported from Roman North Africa as ingots. This is based on chemical analysis of the objects whose ore had a western Mediterranean origin, in particular the copper chainmail links which were likely imported from Spain. The similarities between some of the published Roman and Punic Tunisian objects and the Kissi objects are apparent and suggest that these objects may have shared a similar trade pool of metal sources.

According to the archeo-metallurgists who studied the Kissi metals; "Contacts between Kissi and Roman Carthage in the early 1st millennium AD are supported by the evidence presented here. This level of contact between West Africa and the western Mediterranean and beyond at this time period has not previously been documented and represents an important breakthrough in understanding long distance contacts in Africa in the 1st millennium AD. "

(Contacts between West Africa and Roman North Africa)

The concentration of trade goods in particular elite graves indicates that Kissi was a hierarchical society, in which trade and trading networks played an important role. The fact that exotic items such as brass jewellery, swords, cowries and glass beads reached the site of Kissi before the Arab-Islamic conquest of North Africa, show that that community was not cut off from the outside world but was rather bound into a larger network of contacts and exchange.

(Burial and Society at Kissi, Burkina Faso pg 378)

According to the archaeologist Sonja Magnavita who carried out the excavations at the site of Kissi; "it can be therefore hypothesised that at least a low-scale, informal exchange of ‘northern’ for ‘southern’ commodities was being carried out between the nomads of the desert and the peoples living in the Sudan prior to the 8th century AD". She adds that the Kissi settlements were possibly just one of several locations adjoining the Niger Bend, which were inhabited by agricultural communities organised along similar lines, and engaged in regional trade. She argues that given the dearth of archaeological research in the west African sahel, its unlikely that Kissi was exceptional.

Her observation would be proven correct by the discovery in 1997 of a site just 40km north east of Kissi, called Oursi Hu-Beero, whose excavators described it as having "a lot in common" with Pompeii because it was rapidly buried under debris and subject to minimal post-depositional disturbance.

The House-complex of Oursi in the 11th century.

Oursi Hu-Beero is a large house complex that is part of a settlement mound complex that was occupied between the late 1st and early 2nd millennium CE. The region around Oursi contains a wide array of archaeological sites that date back to the Stone Age, although more permanent settlement mounds appear around the start of the Iron Age in the mid-1st millennium BC. These settlement mounds, which reached their height at around 2ha at the turn of the 2nd millennium CE would flourish until the 14th century CE, long after Oursi was abandoned around 1100CE.

(Oursi Hu-Beero pg 39-41)

Location of Ouris Hu-Beero, map by J. Eisenberg.

The house at Oursi consists of 28 clustered rooms constructed in four phases with sun-dried rectangular mudbricks with a building size about 300m2. The units are not free-standing but are built adjacent to one another, sharing enclosing walls and separated by intermediate walls. The house had an upper storey supported by seventeen rectangular mudbrick pillars, sitting ontop of a roof made of wooden timbers.

The upper floor was itself covered by a clay roof, and it was used for most domestic activities as indicated by the roof debris, which included weapons, copper bracelets, beads, and other jewellery, and grinding stones. This use of an upper floor for domestic activity, and the flat-roofed houses, is similary observed among the Lobi in Burkina Faso (see section on Loropeni below), the Dogon of Mali, and the ancient sites of Jenne-Jeno and Dia. The building complex and surrounding sites made up a village that was occupied by a few hundred people, estimates vary from 200 to 800 persons.

(Oursi Hu-Beero pg 35-38, 43-76, 203-209)

The ruins of Oursi Hu-beero and the shelter built to protect them.

Layout of the Oursi House showing the walls and pillars (red) and the rooms (black).

Reconstruction of Oursi Hu-beero, looking north-west.

Finds of spindle whorls, iron slag, iron and copper objects (arrowheads, needles, nails, knives, bells, weapons) and numerous local pottery indicate significant manufacturing activity at the site, which is further evidenced by fragments of textiles found preserved next to one of the iron objects. Organic material from the site was dated to between 1020 and 1070CE.

(Oursi Hu-Beero pg 101-102, 105-112, 173,174)

The occupants of Oursi were likely engaged in regional trade as indicated by finds of knotted string roulette pottery similar to that made in Nigeria, at Dhar Tichitt in Mauritania, and in Mali's Tilemsi Valley. The five cowry shells ultimately originating from the Maldives were found at the site, likely indicating early trade links with contemporary West African centers (possibly Gao and Bentiya) that were connected to trade with Egypt.

(Oursi Hu-Beero, pg 88, 110-111)

some of the items found at Oursi

The end of the occupation of Oursi is represented by the massive fire that burned much of the mudbrick building and the three human skeletons found in the ruins. The three skeletal remains, which were excellently preserved like a modern crime scene, all showed signs of heavy trauma; the woman was struck on the right side of her head, while the man and infant were crushed by the heavy roof fragments, likely related to the same attackers who killed the woman. The site was then deliberately set on fire and permanently abandoned, with the human remains left unburied between the charred remains of their house.

(Oursi Hu-Beero, pg 170-171, 211-213)

Archaeologists who have worked at Oursi suggested that this attack, which was a deliberate act of strategic political violence as no loot was taken, could have been connected to the social upheavals of late 11th-century West Africa. This period was relatively extensively documented, detailing the social upheavals associated with the expansion of the Ghana and Almoravid empires, whose armies on horseback rode eastwards towards Gao and the Niger Bend region, and attacked the town of Tadmekka.

(Oursi Hu-Beero pg 213-214)

Reconstruction of Oursi Hu-beero looking east.

This attack also coincided with the end of the final phase of settlement at Kissi between the 9th and 12th century CE, where only a few graves were made. After this, the permanent human occupation at Kissi, inhabited for more than 1500 years, ceased. Archaeologists suggest that the arrival of nomadic herders, such as the Peul (Fulani) who currently occupy the region, likely contributed to the abandonment of the settlement by its sedentary farming population.

(Some Aspects on the Iron Age sites of Kissi pg 83)

These upheavals were not influenced by the expansion of the Mali and Songhay empires which emerged much later in the 13th to 16th centuries, and did not control the region, since their closest province was Hombori, 150km north-west.

Known historical polities in the region appear much later in the 19th century, particulary the Fulani Emirate of Liptako which was engaged in several battles with the Tuaregs. Before the massive arrival of the Fulani and other pastoralist groups in the mid-second millennium CE, the region was inhabited mainly by Voltaic farmers who were Gur speakers according to what is known from oral traditions.

(Burial and Society at Kissi, Burkina Faso, pg 390-391)

The mysterious rise and collapse of the pre-Islamic settlements at Kissi and Oursi, is paralleled by the history of the much larger site of Loropeni, about 400 miles to its south-west that is also attributed to non-Muslim Gur-speaking groups.

The enigmatic ruins of Loropeni (11th to 19th century)

The World Heritage site at Loropéni is the best-preserved of about a dozen large quadrangular sites enclosed by stone walls in the southwest of modern Burkina Faso. They are part of the larger Lobi Ruins, a 120-mile-by-60-mile cultural landscape spanning lands that cross the modern borders of Burkina Faso, Côte d’Ivoire, and Ghana. It includes similary large ruins such as Obire, Karankasso, and Lakar that are about the same size as Loropeni. The area is today mainly settled by the Gan (to the west) and the Lobi (to the east).

(Fouilles archéologiques dans le compartiment pg 89-90.)

Map showing Loropeni and important trade towns in medieval west Africa.

Aerial view of the site, east side

Plans of enclosures at same approximate scale. Loropéni, Karankasso, 1km west of Obiré, Obiré village, Obiré ouest, Lakar, Olongo, Yérifoula and Loghi. Image by Henry Hurst.

The Loropeni ruins consist of a massive quadrangular-shaped stone and earthen rampart complex measuring 105 × c. 115 m, which includes laterite stone perimeter walls that reach 6m high and are 1.4m thick. Inside the walls are ruins of twenty rectilinear house structures whose walls are preserved to a height of 2-3m.

Traces of beam locations and remains of holes regularly aligned on the ground to receive posts suggest that the roofs of these rooms were flat. This was similarly observed in the settlements just outside the walls of Loropeni, which featured flat-roofed rectilinear muti-roomed buildings constructed with stones and earth, some within enclosures and some isolated.

(Les ruines de Loropéni pg 53, Approches archéologique et architecturale Prg 4, Stabilization of the Ruins pg 11-12)

Structures in north part of Loropéni interior, looking south-east. Image by Henry Hurst.

Partial view of a structure (Lor-sat 27) of Loropéni exposed in its northern and southern parts. Image by H. Farma.

Superposition of structures of different plans, southern compartment, sector 11b. Image by L. Kote.

Layout of Loropeni and the rooms excavated by L. Kote.

The excavations uncovered buried walls which suggests that the ruins were occupied, abandoned, then reoccupied. This multi-period occupation began in the 11th century, based on dates obtained from charcoal remains obtained during early excavations, and the site was only abandoned in the 19th century.

More recent dates obtained from excavated sectors next to the domestic structures in the interior of Loropeni provided dates ranging from; 1040-1420CE, 1420-1660, 1400-1850. while the peripheral settlements just outside Loropeni's walls provided dates subdivided into three occupation phases; Phase 1 from the 14th to 15th century, phase 2 from the 15th to 17th century and phase 3, from the 17th to the 18th century.

Finds made in the excavations included local pottery; iron slag and metal pieces metal objects (arrowhead, needle, hammerhead, and a knife with curving blade). The settlements just outside the ruins of Loropeni also yielded material such as local pottery, as well as cowries and stone and glass beads in phase 2; and a copper alloy ring in phase 3.

(Les Ruines de Loropéni au Burkina Faso pg 260, Fouilles archéologiques dans le compartiment pg 97-98, 102-103, 109, Approches archéologique et architecturale Prg 3-5)

Attribution of the sites to specific groups has been primarily done through studying the oral history and anthropology of the present populations at the site, who consist of two groups. The autochthons; Gan, the Koulango, the Lorhon, and the later arrivals; the Touna; and the Birifor, the Dyan, the Dyula and the Lobi.

According to the most recent study by the archaeologist Lassina Simpore: "the anteriority of occupation of the region is recognized by the Lorhon, the Touna and the Koulango. They would have settled before the 15th century, we can say in the current state of knowledge of the Lorhon Koulango-Touna that they are the builders of the stone ruins of the region, notably those of Loropéni, The Gan who would have arrived after them, would have appropriated their material heritage"

(Les Ruines de Loropéni au Burkina Faso pg 256-257, Archaeology of Greece and Rome pg 369-370, Fouilles archéologiques dans le compartiment pg 118)

The construction of the walls without any features that would have been used for defending the occupants rules out its use as a fortification., unlike the 19th century sites of Hamdullaye (in Mali) and Sati (in Burkina faso) that were explicitly constructed for that purpose, despite their stricking resemblance with Loropeni. The presence of large domestic structures inside the walls likely indicates that it was occupied by elites.

Other authors connect the ruins with gold mining, noting that Loropéni is situated on medieval trade routes between Koumbi, Gao and Mopti (in modern Mali) in the north and Salaga (Ghana) and Bondoukou (Côte d’Ivoire) in the south.

(Fouilles archéologiques dans le compartiment pg 91, Les Ruines de Loropéni au Burkina Faso pg 261)

The ruins of Hamdallaye and Sati, Images from the 2009 Burkinabe report to Unesco.

View of the interior of the Loropeni enclosure, North compartment. Image by Henry Hurst.

The Mande-speaking Dyula/Juula traders from medieval Mali are known to have penetrated the upper basin of the Volta river during the late middle Ages, establishing the city/town (or simply gold 'mine') of Begho in the 14th century according to local chronicles from the 17th and 18th centuries. This textural evidence for a southern expansion of Muslim traders from Mali was confirmed by findings of Islamic material culture, burials and long-distance trade goods in Begho dated to 1400-1700. Begho's collapse led to the dispersion of many of the Juula groups who are credited with the establishment of the towns of Bondouku and Buna during the 17th/18th century.

However, while evidence of external trade is doubtlessly indicated by the presence of cowrie shells and glass beads, evidence for gold mining or trade at Loropeni and similar sites in the country of the Lobi has yet to be uncovered, and the Muslim Juula traders arrived much later.

Historical accounts describing the expansion of the Watara empire of Kong which controlled part of the region during the 18th and 19th centuries, mention that the Lobi became especially known as a population that resisted the Watara warriors by way of fortified villages and open battle. However, it was only during the early colonial period that the Dyula/Jula gold traders started settling in some larger Lobi settlements like Loropeni, long after its heyday in the early to mid-2nd millennium.

(Gold Mining and Juula influence pg 43)

The Lobi's historical traditions emphasize that their predecessors were the Gan, speakers of the Gur language whose rulers 'arrived' into the region from northern Ghana and joined (or were joined) by the autochtonous Koulango and Lorhon. Although this arrival event is dated to the 15th century, the anthropologist Madeleine Pere who has studied the ruins and Gan society more extensively, argues that the 'king lists' of the Gan may extend much further back. (future athropological research will be needed to clarify the historical relationship between the Koulango& Lorhon, versus the Gan)

She adds that another Gur-speaking group; the Kulango of Buna, who were renowned gold traders associated known in the Asante kingdom of modern Ghana likely aided the Gan in mining and trading the gold, since the Gan weren't typically associated with either activity. The later upheavals associated with the arrival of the Lobi, and their wars with armies of the Wattara and Samory doubtlessly affected the Gan's settlement patterns and architectural styles, resulting in the abandonment of Loropeni and similar sites, as well as ending the regional trade that sustained them.

(Vers la fin du mystère des ruines du Lobi pg 83-89)

Conclusion: a window into the civilizations of pre-Islamic west Africa.

The above analysis of these three, seemingly disparate pre-Islamic sites in Burkina Faso reveals the complex sequence of cultural developments in West Africa since antiquity.

The discoveries at Kissi showed that luxury goods from various parts of Africa and the wider world were finding their way into the West African Sahel earlier and, more importantly, on a larger scale than previously thought.

The House complex at Oursi provides the earliest evidence that monumental architecture associated with Islamic urban centers like Jenne and Timbuktu, with rectilinear mudbrick structures with an upper storey, was can also be found in non-Muslim societies. Similar archaeological digs in northern Ghana also uncovered evidence of massive rectilinear storey houses dated to the 15th century before the arrival of Muslim traders.

Finally, the walled towns of the Lobi ruins with their planned architectural layout, flat-roofed stone houses, and massive enclosure walls whose monumental function bears resemblance with the famous ruins of Great Zimbabwe, in a largely non-Muslim region, provides further proof for the independent emergence of several cultural developments typically associated with the arrival of Islam.

Hopefully, future discoveries in the region will uncover the extent of pre-Islamic civilizations in West Africa and show that their inventions, which are so often misattributed to external influences, had an indigenous origin.

References:

Timbuktu and the Songhay empire, translated by John Hunwick

Oursi Hu-beero: A Medieval House Complex in Burkina Faso, West Africa, edited by Lucas Pieter Petit, Maya von Czerniewicz, Christoph Pelzer

Archaeology of Greece and Rome : Studies In Honour of Anthony Snodgrass, edited by John Bintliff

Burial and Society at Kissi, Burkina Faso by Sonja Magnativa

Some Aspects on the Iron Age sites of Kissi by Sonja Magnativa

The Early History of Weaving in West Africa by Sonja Magnativa

Contacts between west Africa and Roman North Africa by Thomas R. Fenn, David J. Killick, John Chesley, Sonja Magnavita & Joaquin Ruiz

Glass beads from Kissi (Burkina Faso): Chemical analysis and archaeological interpretation. by P Robertshaw, S Magnavita, M Wood, E Melchiorre, R Popelka-Filcoff

Fouilles archéologiques dans le compartiment sud du site des ruines de Loropéni by Lassina Kote

Les ruines de Loropéni, Burkina Faso by Lassina Simporé

Approches archéologique et architecturale des enceintes et structures en pierres au sud-ouest du Burkina Faso by Hantissié Hervé Farma

Stabilization of the Ruins of Loropeni (UNESCO, 2009 report)

Les Ruines de Loropéni (January 2009 report by Ministère de la Culture, Burkina Faso)

Les Ruines de Loropéni, premier site burkinabé patrimoine mondial de

l’humanité by Lassina Simporé

Gold Mining and Jula Influence in Precolonial Southern Burkina Faso by Katja Werthmann

Vers la fin du mystère des ruines du Lobi by Madeleine Père.

link