Close to the small town of Dia (or Jà) in the western Inland Niger Delta of Mali lies the Dia Mound Complex, comprising three extensive occupation mounds, similar to the Jenne-jeno mound complex close to the city of Djenné near the southeastern marginof the Inland Niger Delta (the names 'Jenne' and 'Djenné' are interchangeable for the city, with Djenné having been favoured by the French colonial government, and Jenne by the post-colonial government of Mali, although the name has never been officially changed - both would be pronounced as 'Jenny' by an English-speaker).
Many peoples in the region claim ancestry from Dia, including the numerous Bozo and Marka groups. Like Djenné and Timbuktu, Dia forms part of the ancient trade network of the region, although unlike these cities it appears to have never been a major commercial centre, but instead is a religious centre to both the Muslim and Pagan peoples of the western Sahel region. The Marabouts (Islamic scholars) of the town are thought to manufacture especially powerful amulets, while in the Pagan Mande origin story, the rainbow which marked the north-south axis of the world at the beginning of time had its northern end at Dia.
Dia is thought to have flourished as an important religious centre after the fall of the ancient Ghana Empire, with clerics from the city spreading Islam westward to the Senegambia region, southward along the Niger, and eastward toward Hausaland. Dia may be the city of Zagha, which was captured by the Songhai Emperor Askia Muhammad I, although the oral histories of the city do not mention being part of the Songhai Empire. The recorded history of the town effectively begins with the advent of the (Pagan) Jawara Dynasty in the fifteenth century; this dynasty lasted until the mid-nineteenth century, and this long period of Pagan rule may have led to the high degree of emphasis being placed upon the magical powers of the city's Islamic scholars.
In the mid 1980s the Dia Mound Complex was the subject of a series of excavations led by Susan Keech McIntosh of the Department of Anthropology at Rice University, and Roderick McIntosh of the Department of Anthropology at Yale University, although the limited available dating technologies of the 1980s, combined with the small number of excavations carried out at other sites in the western Sahel at that time, limited the interpretation of the excavated material. In a paper published in the journal Azania: Archaeological Research in Africa on 1 December 2023, McIntosh and McIntosh revisit the results of their 1980s excavations at Dia, and interpret the material recovered in the light of improved dating techniques and information from subsequent archaeological work in the Inland Niger Delta of Mali.
Map of the Inland Niger Delta and sites and regions mentioned in the text. Top inset shows archaeological sites investigated in 1986–1987. McIntosh & McIntosh (2023).
McIntosh and McIntosh carried out excavation work at two large mounds in the Dia Mound Complex, Mara and Shoma (the town of Dia sits on the third large mound). All three large mounds, and a number of smaller ones, sit on a levee system about 5 km from the river Diaka, a major branch of the Niger. The mound complex, and town, sit at the northwest of a floodplain which is inundated between August and December, and utilised for the cultivation of African Rice.
The Shoma Mound has not been occupied at any time within historical memory. The Mara mound was occupied until the nineteenth century, when the town was taken over and Islamised by Seku Amadu, founder of the Massina Empire, and the settlement was moved to its current location the Dia mound. Cotton was grown on the Mara and Shoma mounds in the 1960s, but this had ceased by the 1980s. Part of the Mara mound is covered by a cemetery used to this day.
McIntosh and McIntosh excavated three trenches in 1986 and 1987, two on the Shoma Mound, and one on the Mara Mound.
The first trench, Sh7, was made at a high point on the western part of the Shoma Mound, where there was an extensive surface scattering of pottery fragments, including examples of styles from all occupation phases at the Jenne-jeno mound complex. A 2.0 by 1.5 m trench was excavated with a north-south orientation, cutting through 15 natural layers to a depth of 4.5 m, although in the lowest parts the size of the trench was reduced to 2.0 m by 85 cm. Levels 1-5 (where Level 1 is the uppermost), 7-9, and 11-14 showed remains of active occupation, while layers 6 and 10 contained significantly less traces of Human activity.
The second trench, SCW, intersected with a periphery wall exposed on the western edge of the Shoma Mound. At the site chosen for the excavation, a section of the wall had been exposed by a narrow gully, and partially collapsed, exposing a mass of irregular sun-dried bricks. This wall was apparently made up of two layers, with loose bricks between. The outer wall was made up of irregular cylindrical bricks averaging about 20 cm long and 9 cm in diameter (cylindrical bricks, known as 'Djenné ferey' are still used in the region). The bricks of the loose fill were orientated randomly, and averaged 15.5 cm in length and 10.3 cm in diameter. The whole wall appears to have been about 3.5 m thick, comparable to the excavated city wall at Jenne-jeno, which was 3.4 m thick. In the 1990s increased rainfall in the region exposed a wall on the northern perimeter of the mound with a zig-zag configuration; this may also have been the case on the western side of the mound, where the orientation of the wall, at 45° to the margin of the mound, was considered hard to interpret at the time of the excavations.
Shoma SCW: photograph of the excavation area. McIntosh & McIntosh (2023).
The SCW trench measured 2.5 m by 75 cm, and extended downwards for 1.98 m, 95 cm beneath the base of the wall. In the bottom part of the trench, the size was reduced to 1.40 m by 75 cm. Nine layers were excavated in this trench, the uppermost five yielded grey clay and heavy loam with elongated and cylindrical bricks, layers 6-8 produced a light clay, upon which the wall had apparently been built, while layer 9 was a friable pale brown sandy loam, with occasional scattered artefacts, which made up the bottom 28 cm of the trench, and presumably some way deeper.
The third trench, D6, was excavated on a high part of the Mara Mound, to the east of the cemetery. This area again had abundant surface pottery, which showed similarities to pottery from phases IV and V at Jenne-jeno, which has been dated to the second millennium AD. The trench measured 2 m by 2 m and was orientated north-south, extending downwards for 4.05 m, passing through 14 sediment layers. In the lower parts the dimensions of the trench were reduced to first 1 m by 2 m, orientated east-west, and then 1 m by 1.25 m. The first five layers at this site exposed cylindrical mud bricks, from an apparent wall collapse, as well as a number of refuse pits. Layers 6-8 were made up of a clay material, which might have been formed by the melting of mud bricks, with no other artefacts present. Layers 9-13 were composed of a homogeneous loamy material, with fluctuating levels of pottery fragments, and an increasing amount of limonite (an iron oxide material which forms in waterlogged soils) in the lower layers. Layer 14 was comprised of similar material, but with no pottery inclusions, and reached the water table.
At all three sites, cylindrical bricks of the Djenné ferey type were present in the upper layers. These bricks spread along the middle Niger in the second millennium AD, and is believed to have been spread by masons from Djenné. At Sh7 and D6 the lower layers contained thin-walled, well-fired pottery similar in style to that from the earliest stages at Jenne-jeno. At SCW similar pottery was found along with the brick deposits.
In order to establish a chronology for the Dia Mound Complex sites, McIntosh and McIntosh first examined the pottery fragments for signs of time-sensitive trends. Most of the pieces were small, and showed only a single decorative element, so McIntosh and McIntosh sorted these by decorative element (having first identified pieces likely to have come from the same pot, to avoid double-counting). Decorative elements present include lip, variou sforms of twine/fibre rouletting, paint (red, white, black) and plastic techniques, such as channelling (created by incising with a multi-toothed tool), comb-impression, stamping, fingernail impression and shallow comb-dragging.
Shoma and Mara: decorative motifs on sherds from the 1986–1987 excavations. Top row (early): braided twine, fish vertebrae roulette, folded fibre strip and twisted twine roulette, fine folded fibre strip roulette overlain by stem dragging, same on Ev1 rim, cord-wrapped stick; middle row: fibre strip roulette overlain by comb drag, polychrome (black or white in channelling or over roulette plus red slip), white on red geometric; bottom row (late): beaded carinated rim with finger nail impression, stamping with white fill, red paint splashed. McIntosh & McIntosh (2023).
Rim fragments were also sorted into groups, using a classification derived from one used on the Jenne-jeno material, with all rim sherds from each level in each excavation unit described in terms rim angle, rim diameter (where determinable), slip (presence/absence, colour, position on sherd) and decoration (details of twine/fibre roulette, paint, plastic decoration and their position on the sherd).
Shoma and Mara: rim forms recorded 1986–1987. McIntosh & McIntosh (2023).
The pottery at sites Sh7 and D6 showed significant changes over time, with some layers showing similarity between the two sites, and some showing similarity with Jenne-jeno. Twine/fibre rouletting is present on over half of the fragments where only a single style of decoration is present. The patterns are similar at the two sites in the lowest layers, but in the upper layers the styles diverge significantly. Multiple styles of decoration on the same piece of pottery are quite rare, but where present some changes over time can be seen. This is most commonly found on rim fragments, with early examples from Sh7 showing rim pieces with two different forms of roulette decoration in adjacent, slightly overlapping fields. This is also present in early layers at Jenne-jeno. Curiously, at D6 this pattern is present at low levels throughout the occupation, while it disappears at both Sh7 and Jenne-jeno. Painted pottery with decoration is present at both sites, with lower layers at Sh7 showing black and white patterns painted over a comb-dragged decoration, while at D6 upper layers produce a pattern of stamps over comb decoration.
The pottery in the lowest layers at both Sh7 and D6 is remarkably well-made, from a well-mixed and homogeneous paste comprised primarily of clay and silt with fine sand and finely ground grog. These pots have thin, even walls typically between 3 mm and 8 mm thick, and with a very smooth finish. This type of pottery has also been found in the earliest layers at Jenne-jeno, although here it only makes up about 10-15% of the assemblage. This fine pottery disappears in later layers at all three sites, being replaces with a thicker-walled pottery with a grog temper and a large number of voids. The fine-pottery producing layers at Jenne-jeno have been dated to the last two centuries BC, and the early centuries AD. Fine pottery of this type is more prevalent in higher layers at SWC, though McIntosh and McIntosh attribute this to old pottery fragments being incorporated into clay bricks, then eroding out as the bricks weather.
Rim fragments from the lower layers at Sh7 and D6 are almost exclusively made from the fine paste material, with a roulette impression pattern below the lip, which may or may not be covered by a slip. Three basic types of pottery are common: Bowls with simple, unrestricted rims, quite often with cross-hatched red paint around the rim, globular jars with restricted rims that are slightly rolled outward at the lip, and ovoid jars with small, everted rims that are flattened at the lip. The bowls and globular jars are also common in the lowest layers at Jenne-jeno. Potlids, long-necked bottles and globular vessels that differ mainly in lip shape are also present.
The nature of this pottery assemblage changes in layers 6-10 at Sh7, with the three earlier forms being replaced by an assemblage in which potlids and two classes of carinated pots are the most common items. Both types of pots are frequently decorated with multiple channels filled with white and/or black paint and/or geometric painted designs.
Shoma and Mara: pottery phases recognised in the 1986–1987 excavated assemblages. Top two rows: late; middle three rows: polychrome; lower three rows: early fineware. McIntosh & McIntosh (2023).
In layers 4 and 5 at Sh7, carinated rims remain common, but paint decoration becomes extremely rare. Instead, a new form of decoration appears, with fingernail impressions on the carination, beaded lips, and comb incised channels. Similar pottery is known from Phase IV at Jenne-jeno, which has been dated to between 900 and 1400 AD. However, in the Jenne-jeno material, braided twine replaces fibre strip as the dominant form of roulette, which never happens at Sh7.
Surprisingly, the rim decoration in the upper layer at D6 on the Mara Mound is quite different from that present in Sh7 on the Shoma Mound. Here plates, potlids and carinated rims, including the form of decoration appears, with fingernail impressions on the carination, beaded lips, and comb incised channels seen at Sh7, but there is also a large number of pieces with everted rims, something not seen at Sh7 and very rare at Jenne-jeno. No polychrome decoration is present on any pottery at D6, while stamping and comb impressions are common. Similar stamp motifs appear at Jenne-jeno, where they show a pattern of development; the impressions are small from 900 to 1400 AD, with larger impressions after 1500 AD. These late, large stamp impressions persist into the historic 'Sahona Tradition' of the southern Inland Niger Delta. The presence of everted rims, large amount of undecorated pottery, and common use of braided twine and large stamp decoration, occasionally with a white fill, strongly suggests that this pottery was made after 1400 AD. Similar pottery is known from the fourteenth to twentieth centuries at Jenne-jeno/Djenné, and from the nineteenth century at Hamdallahi on the eastern Inland Niger Delta. However, the D6 site has not yielded any Tobacco pipes, which are common throughout the Middle Niger from the seventeenth century onwards, suggesting that this assemblage predates the introduction of Tobacco to the region.
McIntosh and McIntosh also obtained a series of new radiocarbon dates for the assemblage, to better date the sequence. The oldest date obtained came from a burned Fish bone from the lowest level of collapsed brick at site SCW, which yielded a date of between 795 and 752 BC (with a 78% probability). This is surprising, as the layers beneath this yielded cultural items which would not be expected before the first millennium AD, which also show considerable patination and erosion, suggesting that they lay on the surface for some time before being buried. This all suggests that the wall was built on top of an accumulation of older material which had been washed downslope over a period of centuries. It is thought that the wall was probably built in the fourteenth century AD, the date obtained for the north wall at the site, although further investigation would be needed to confirm that the two walls are actually part of the same system. Unfortunately, a survey carried out in 1998-2003 did not detect a wall on the western part of the mound, possibly suggesting that it had been destroyed by a catastrophic erosion event since the 1980s.
The pottery fragments recovered from SCW are made from a fine material, and were thought to have been examples of ancient pottery incorporated into the bricks, which had then been eroded out. McIntosh and McIntosh now also note that this assemblage is dominated by pottery with flattened, everted rims, which is also common in the lowest levels at Sh7 and D6, supporting the hypothesis. They note that there appear to be two phases within the fineware assemblage at both sites, with the earlier one showing flattened everted rims. They suggest that this earlier fineware phase with everted, flattened rims, probably dates from 800-400 BC (which would support a common origin for the pottery fragments and Fish bone at SCW), while the later phase is likely to date from about 400 BC to about 200 AD. The material from the second fineware phase is almost identical to the material from the earliest phase at Jenne-jeno. No radiocarbon dates were obtained from the lowest layers at Sh7 and D6. However, excavations carried out by archaeologist Noémie Arazi on Shoma Mound in 1999-2002 produced similar pottery from layers which yielded radiocarbon dates of 2450 years before present and 2220 years before present, strongly supporting the idea that these earliest layers date to the first millennium BC.
Dia fineware from Shoma and Mara shown next to almost identical sherds from Phase I at Jenne-jeno. McIntosh & McIntosh (2023).
The middle levels at Sh7 (6-9) are curious in that they show evidence of settlement in a period otherwise apparently absent on Shoma Mound. A radiocarbon date for Level 9 was obtained in the 1980s, giving a result of between 892 and 1193, to which McIntosh and McIntosh add a new date, of 1040–1165 for Level 7. The only other dates from this period come from Azari's excavations on Shoma Mound, and comprise a date of 940 years before the present from a piece of charcoal found in a sandy layer beneath a level of collapsed brick dated to 460 years before present, and a fragment of Wheat within a pit dated to 1070 years before the present. A layer of sediment dated to between 2070 and 940 years before the present yielded no cultural materials, suggesting the mound was abandoned during this time. Interestingly, the cemetery on Mara yielded several burials dating to the tenth and eleventh centuries AD, although quite where these people lived has remained unclear. Layers 6-9 at Sh7 have yielded a distinctive polychrome pottery, not seen elsewhere in the Dia Mound Complex, apparently indicating a distinctive cultural presence, not detected elsewhere. The higher layers at Sh7 produced Phase IV at Jenne-jeno, which has been dated to between 900 and 1400 AD, as mentioned previously.
At the D6 site, a new date of between 466 and 386 BC was obtained from a piece of burned bone at the top of Layer 10 or bottom of Layer 9, lending further support to the idea that the material in these layers dates to the first millennium BC. Dates from higher in the sequence lead to the conclusion that there were two later phases of occupation, one in the eleventh century AD, and one n the fourteenth to eighteenth centuries AD. This sequence appears to have been slightly disturbed by pit digging around the site, something not detected during the 1980s excavations, but consistent with the use of bud-brick architecture.
Based upon comparison of the sequence of pottery to other sites on the Inland Niger Delta and the new radiocarbon dates, as well as data from the 1998-2002 excavations on Shoma and Mara mounds, McIntosh and McIntosh propose a chronology for the Dia Mound Complex.
Occupation began in the first millennium BC, by agropastoral and fishing groups using distinctive, thin-walled pottery. These are represented by layers 11-14 at Sh7 and 10-13 at D6, as well as in the lowest layers excavated in 1998-2002. The earliest evidence of occupation in Sh7 and D6 is on loamy clay levee deposits, while the earliest remains uncovered in 1998-2002 are on floodplain clays. The data from the 1986-87 excavations suggests that these peoples preferentially settled on the top of levees on the Diaka floodplain. In October 2013, flooding on this plain reached between the 270.5 and 271 m contours. Accurate surveying equipment was not available to the 1986-87 expedition, but it is thought this would have been above the lowest excavated layers in Sh7.
The lower layers at Sh7 and D6 contain alternating layers of light loams and light clays containing iron oxide inclusions that precipitate in standing water, suggesting that the area was prone to periodic flooding. The density of pottery in the lowest layers at Sh7 was 3-8 times as much pottery as at D6, suggesting D6 was only the occupied on a sporadic, short term basis. The level of water flowing through the Diaka and its associated floodplain appears to have varied over time. At D6 the pottery from Layer 12 has a thick surface deposits of dark iron or manganese oxide precipitate, suggesting a prolonged or repeated immersion in water. This suggests a variable environment with periodic flooding. The 1998-2002 excavations recovered several grains of African Rice, Oryza glaberrima, dated to this early period.
Fragments of the fineware from this early period continue, at a low level, at many sites on the Dia Mound Complex deposits. The evidence from SCW, of pieces of this earlier material being incorporated into bricks made almost 2000 years later, gives an explanation for this stratigraphic mixing.
The style of this early pottery has been linked to the earlier proto-Mande Tichitt Culture of eastern Mauritania, suggesting that these earliest settlers in the region migrated southward from the Tichitt region in the early first millennium BC.
This earlier occupation was followed by a hiatus, with little sign of Human activity until the first millennium AD. A distinctive layer of aeolian sand, lacking any artefacts, is present at four of the six sites excavated in 1998-2002, but absent from the other two of these sites, as well as Sh7 and D6. Rice grains derived from these sands have yielded dates of 1996 and 1878 years before the present. There are no signs of any occupation during this period, although it is possible that people were present at least seasonally, using the floodplain to grow Rice. The uppermost layers of the early occupation at D6 yield pottery fragments covered by a thick carbonate deposit, suggesting that they were exposed on the surface for a long period of time, with periods of immersion.
At Sh7 a metre-deep layer producing evidence of occupation has been dated to between 900 and 1165 AD, while a similar deposit at D6 has yielded dates of between 1025 and 1150 AD. These deposits are particularly interesting for two reasons; firstly, no evidence of occupation during this period was found during other excavations at the Dia Mound Complex, and secondly, the material at the two sites is quite different. The Sh7 material is dominated by potlids and carinated pots, similar in style to material from Jenne-jeno which has been dated to between 800 and 1000 AD. At D6 a different style of pottery is present, dominated by pots with everted and thickened rims, and with different forms of decoration to that seen at Sh7.
The reason for this difference is unclear. It may simply be the result of pottery being used for different purposes at the two sites, or represent some difference in social standing between the peoples of the two sites. The Dia Mound Complex is close to a river system which formed an important transport corridor, and it is not unreasonable to expect pottery found in such an environment to reflect wider regional interactions. The polychrome pottery of Sh7 shows strong similarity to pottery from the same period at Jenne-Jeno, which in turn is part of a sequence of pottery found over the wider eastern upper Inland Niger Delta over the period 400-1400 AD. The pottery of D6, however, is quite different, showing cultural affinity with the Méma region to the north.
The presence of pottery at Sh7 with affinities for that of Jenne-jeno and the eastern Inland Niger Delta is interesting, and such pottery has not come from any other excavation site on Shoma. McIntosh and McIntosh speculate that this may represent a small group of people with ties to the eastern Inland Niger Delta. During the 1998-2002 excavations, it was noted that several sites had erosional surfaces which would coincide with this period, possibly suggesting that more material has been lost.
Carinated polychrome pottery resembling that from Phases II and III at Jenne-jeno (which have been dated to between 200 and 900 AD) was excavated from the Diohoun site at the confluence of the Diaka and Niger in the 1990s. The uppermost layers at Diohoun yielded material similar to the material from Sh7 and Phase IV at Jenne-jeno, and Arazi reported layers dating to the second half of the first millennium AD, which yielded significant quantities of carinated pottery and some black- and white painted ware on the Mara Mound, making this absence of pottery from other sites on Shoma more mysterious.
This phase appears to have been followed by another hiatus, then the appearance of a new inhabited phase on both Shoma and Mara, distinguished by extensive mud brick architecture. A number of radiocarbon dates have been recovered from these layers at sites on both mounds, ranging from 1275 to 1500 AD, which would coincide with the height of the Mali Empire.
No radiocarbon data were recovered for this upper layer at Sh7, although the nature of the pottery found strongly suggests a thirteenth or fourteenth century date for these layers. At D6 a wall was constructed in the fifteenth century, and a number of pits contain pottery with decorative elements which suggest having been made post-1500, based upon comparison to material from elsewhere in the Niger Delta. This pottery was found alongside remains indicating a mixed economy based on fishing, domestic stock and the exploitation of wild resources, especially Birds. Similar faunal remains were recovered from later, post 1500, layers excavated in 1998-2002. These later excavations also recovered a small amount of material considered to be trade goods, including four cowries, one copper pendant and four other fragments of copper and 18 glass beads. This is a surprisingly small amount of such material, given the extent of the excavations, suggesting that Dia was not an important trade hub during this period. This is consistent with the known history of the city, as a religious centre rather than a commercial site, although it is likely that agricultural goods, such as Rice and Cotton, as well as smoked or dried Fish, were also being produced and exported.
Most of what is currently known about the Dia Mound Complex derives from the extensive excavations in 1998-2002. However, re-examination of the much smaller 1986-87 excavations does add somewhat to this knowledge. These digs were located on the edge of the Shoma and Mara mounds, while the later digs were concentrated on the central part, and uncovered evidence of sporadic occupation far earlier than the later sites uncovered, as well as evidence for a climate with a more active hydrological system, when the levee deposits were still being built up and the area was periodically covered by floodwaters.
The Sh7 and D6 trenches from 1986-87 yielded distinct sequences of pottery, from which a distinct progression of styles can be seen, and compared to other sites on the Inland Niger Delta. This was not possible during the 1998-2002 excavations, probably because the stratigraphic sequence had been extensively turned over by pit-digging. Re-use of material for brick-making, as seen at SCW, is also likely to have added to this confusion. Nevertheless, McIntosh and McIntosh recommend that the pottery from the 1999-2002 excavations be re-examined for evidence of progression.
The material from the lowest layers, supports the hypothesis that two distinct phases can be observed in the first millennium BC, one associated with the earliest arrivals at Dia and a second, later one associated with the expansion of colonisation deeper into the eastern Inland Niger Delta, including Jenne-jeno.
The middle layer at Sh7 shows a distinct affinity to pottery from Jenne-jeno, something which was not recognised in any of the 1998-2002 material, and places the Dia Mound Complex within a wider cultural context that includes sites such as Tiébala and Diohoun. Surprisingly, pottery from the nearby D6 site shows closer affinities to the Méma region to the northeast of the Inland Niger Delta.
The absence of pottery from this period at any of the sites excavated in 1998-2002 is surprising. McIntosh and McIntosh suggest that this might be due to greater erosion at these sites on the mound summits, although it is possible that the settlements uncovered at Sh7 and D6 were very local in nature, and that the choice of these sites for excavation was extremely fortuitous.
See also...
Follow Sciency Thoughts on Facebook.
Follow Sciency Thoughts on Twitter.
