A Reconstruction of Nunn’s (2008) Dataset

Guinea-Bissau (0.46). Before 1600 the Cabo Verde trade drew on the rivers of present-day Guinea-Bissau, the Cacheu, Sao Domingos, and Geba. Green’s (2012, 23-25) two tables of recorded origins are led by the peoples of that coast: the Biafada are 22.9 percent and the Brame 20.0 percent of the 441 captives recorded for 1547-60, and 28.0 and 30.4 percent across 1560-1600. With the Bijago, Pepel, Nalu, Cocoli, Balanta, and Bainunk-Kassanke, the peoples Green (2012, 33) places in the coastal regions of present-day Guinea-Bissau make up close to half of all recorded origins.

Senegal (0.285). The northern Senegambia between the Senegal and Gambia rivers supplied the Wolof and Sereer, whom Green (2012, 31) locates in the interior to the north and in the creeks of the Saluum delta, with the Casamance peoples to the south; in Green’s (2012, 23) 1547-60 table the Jolof are 19.1 percent and the Sereer 8.2 percent. The catchment also reached a Futa Toro / upper-Senegal interior the coastal tables could not name: the 16th-c records aggregate the Senegal-valley peoples under broad ethnonyms and Iberian scribes logged area-of-purchase rather than homeland, so this deeper Senegal-valley share is credited here on Green’s location of the Wolof and Sereer between the two rivers.

Guinea (0.085). The southern edge of the catchment reached the Sape, Susu, and Nalu coast of present-day Guinea and the Nunez river, which appear among Green’s (2012, 23-25) recorded origin groups, the Sape at 1.1 to 8.8 percent. The upper-Guinea hinterland behind those rivers that the coastal records could not name is included here; the Futa Jallon highland that later dominated Guinea’s trade was not yet a major slaving frontier, so the share stays low.

Mali (0.065). The Mandinga are 8.8 percent of Green’s (2012, 23) 1547-60 table, and Green (2012, 31) places most along the Gambia River, so most were Gambia-Casamance Mande rather than people of the deep Niger interior. A small deep-interior Mande supply that the coastal records aggregated under the Mandinga label is credited here; large-scale upper-Senegal supply from the Niger interior is an eighteenth-century development, so Mali’s pre-1600 share remains small with a wide band.

Sierra Leone (0.05). Green (2012, 5) names present-day Sierra Leone as one of the three Cabo Verde source areas of the sixteenth and seventeenth centuries, reached through the salt-for-slaves circuit Dirck Ruiters described in 1623. The share is modest, far below Sierra Leone’s weight in the nineteenth-century names, which reflect the later Freetown recapture geography rather than the Portuguese-era trade.

Gambia (0.05). Green (2012, 31) places the Mandinka settlements along the Gambia River, and the Gambia-valley provinces supplied captives, but the sixteenth-century trade ran mainly through the Bissau rivers and the Senegal coast, not the English-run Gambia trade of later centuries. The Gambia’s share of the cross-border Mandinka homeland is small in this period.

Mauritania (0.005). A small Saharan-edge share belongs to present-day Mauritania for the Moors of the lower-Senegal north bank and the desert-edge trade. In the 16th c this is minor; the gum-and-slave channel that ties the Trarza and Brakna Moors firmly into the trade is an 18th-c development (Levtzion 1975, 220-21), so the band is kept narrow.

Guinea-Bissau (0.45). Green (2012, 5) treats the seventeenth century as continuous with the sixteenth: captives came from present-day Sierra Leone, Guinea-Bissau and the Casamance region across both centuries. The Cacheu fort, built in 1589, made the Bissau rivers the dominant slaving port (Green 2012, 27), and the coastal peoples of present-day Guinea-Bissau remained the core supply: in Green’s (2012, 24-25) Table I.2 the Brame and Biafada are 30.4 and 28.0 percent of the 1590-1600 sample. Mark and Horta’s work on the contraband trade leads Green (2012, 7) to substantially increase estimates for exported slaves from Upper Guinea in the 1610s.

Senegal (0.25). Wolof, Sereer, and the Casamance peoples of present-day Senegal continued to supply through the Petite Cote ports of Joal and Portudal and the Senegal coast (Green 2012, 5). The catchment also reached a Futa Toro / upper-Senegal interior the coastal labels could not name: the 17th-c records still aggregate the Senegal-valley peoples under broad ethnonyms and the Iberian and French scribes recorded the point of sale rather than the homeland, and the upper-Senegal channel was opening (the French reached Galam in 1685), so this deeper Senegal-valley share is credited here. Lovejoy (2000, 28) records Cadamosto’s report of a Wolof king raising revenue through raids producing many slaves.

Guinea (0.098). The Sape, Susu, and Nalu coast and the Nunez and Pongo rivers of present-day Guinea supplied a modest seventeenth-century flow within the Caboverdean salt-and-dye circuit (Green 2012, 5, 25). The upper-Guinea hinterland behind those rivers is included here; the Futa Jallon highlands that would dominate Guinea’s trade in the eighteenth century were not yet a major slaving frontier, so the share stays low with a wide band.

Mali (0.075). Deep-interior Mande supply up the Senegal and Gambia remained a minority in the seventeenth century, but it grew as the upper-Senegal channel opened: the French reached Galam in 1685 and the fort of Saint Joseph followed in 1700 to tap the sources of gold and slaves of the interior (Levtzion 1975, 219). The Dyakanke and Soninke long-distance networks the coastal records could not name are credited here; the Bambara-Segu raiding that floods the coast with Niger-valley captives still belongs to the eighteenth century, so the share is small with a wide band.

Sierra Leone (0.06). Present-day Sierra Leone remained a named seventeenth-century Cabo Verde source: Green’s (2012, 5) statement that captives came from present-day Sierra Leone, Guinea-Bissau and the Casamance region covers both centuries, and the salt-for-slaves circuit Dirck Ruiters described in 1623 ran Caboverdean salt to Sierra Leone and the goods north to Cacheu. The share is comparable to the sixteenth century and well below Sierra Leone’s nineteenth-century weight.

Gambia (0.06). The London Company for the countries of Ginney and Binney established a factory on James Island in the Gambia in 1618, with the French at Albreda opposite, opening the channel that would carry deeper-interior captives in the eighteenth century (Levtzion 1975, 219). In the seventeenth century the Gambia drew Mandinka of the Gambia-valley provinces, whom Green (2012, 31) places along the river. Volumes were still modest.

Mauritania (0.007). A small Saharan-edge share belongs to present-day Mauritania for the Moors of the lower-Senegal north bank. The gum-and-slave channel through the desert edge was still minor in the seventeenth century, before the eighteenth-century gum boom (Levtzion 1975, 220-21), so the band is narrow.

Senegal (0.30). Present-day Senegal supplied through the coastal warrior-states (Kajoor, Kaymor, Saalum), whose royal power depended on slave soldiers, and the Senegal-valley jihad states of Futa Bondu and Futa Toro, whose holy war enslaved many people particularly in the 1740s and again in the 1780s (Lovejoy 2000, 75). The Futa Toro / upper-Senegal captives the coastal records aggregated under broad labels are credited here; the lower-Senegal gum trade drew further captives from the Tukulor and Wolof of the river (Levtzion 1975, 220-21), and Traore (2023, 29) lists Fuuta-Tooro, Waalo, Kaajor, and Bawol among the upper-Senegal slaving states. The band is wide.

Mali (0.26). The defining eighteenth-century change was the rise of the Bambara state of Segu, on the Niger in present-day Mali. Lovejoy (2000, 75) writes that Segu became a major source of slaves for the European slave-ships in the Senegambia basin, and Levtzion (1975, 220) that the supply of slaves increased considerably in the middle decades of the eighteenth century as a result of wars and raids carried on by the Bambara kingdoms. The broad ethnonym and area-of-purchase labels of the Galam and Gajaaga records mask deep-interior Niger-valley origins, which Traore (2023, 56) names as Kaarta, Segu, Beledugu, Bambuxu and the other Soninke-Malinke diaspora provinces, credited here. The band is wide to reflect this uncertain interior mass.

Guinea (0.18). The Futa Jallon highlands of present-day Guinea became a major eighteenth-century source. Lovejoy (2000, 60-61) records upper Guinea coast exports rising to about 6,000 a year in the 1760s, tied to the second and most violent phase of the Muslim holy war that founded Futa Jallon, with the jihads of the 1740s and 1780s enslaving many people. The Futa Jallon and upper-Guinea hinterland behind the Nunez and Pongo outlets that the coastal records could not name is credited here. The band is wide.

Gambia (0.10). The English Gambia trade, centred on James Island, was a leading eighteenth-century outlet. Levtzion (1975, 220) records that the English encouraged established patterns of trade and were not interested in penetrating the hinterland, relying on African traders to bring captives down to the lower Gambia, and that by 1726 Bambara Cana was reputed on the Senegambia coast to be a very large kingdom. Traore (2023, 106-108) records captives carried through Gajaaga and the Gambia between 1713 and 1757. The Gambia-valley Mandinka are credited here; the trans-shipped Bambara from further inland are credited to Mali.

Guinea-Bissau (0.08). The Bissau rivers and their coastal peoples, the Brame, Biafada, Balanta, Papel, and Bijago, continued to supply in the eighteenth century, but their relative weight fell sharply as the trade’s mass shifted to the deep-interior Bambara and Futa channels up the Senegal and Gambia. Lovejoy (2000, 61) notes the eighteenth-century trade drew on the interior more heavily than before. Green (2012, 33) identifies these Guinea-Bissau coastal peoples, now a minority share of a much larger regional total.

Mauritania (0.05). The lower-Senegal gum coast tied present-day Mauritania into the trade: Levtzion (1975, 220-21) records gum obtained from the Moors at the French stations on the lower Senegal, with slaves moving up the river in exchange. The Saharan-edge Moors (Trarza and Brakna) and the Soninke of Gidimaxa are credited here, on Traore’s (2023, 43) placing of the Senegal River at the crossroads connecting Gajaaga and Fuuta-Tooro to the Arabo-Berber nations and the Berbers and Arabs of the Sahara. The band is wide.

Sierra Leone (0.03). Lovejoy (2000, 60) notes that some eighteenth-century slaves reached the coast via routes from the Sierra Leone-Liberia and Ivory Coast interior. For the Senegambia embarkation region proper the Sierra Leone homeland contribution is small; the large Sierra Leone supply belonged to its own coast. The share is kept low, to avoid carrying the nineteenth-century names’ Sierra Leone weight back onto a period when that profile did not hold.

Sierra Leone (0.30). The nineteenth-century names place Sierra-Leone-homeland peoples (Mende, Temne, Limba, Kissi/Kono) at the top of the Senegambia-attributable origins, 47 percent of the attributable mass in the African Origins database. The share is moderated below that figure here because much of the post-1807 illegal trade embarked from the southern rivers at the Guinea-Sierra Leone-Guinea-Bissau margin and the database is a Freetown recaptive corpus that over-represents its own catchment (Fyfe 1976, 184). This is a coastal homeland share.

Guinea (0.21). The Futa Jallon jihad state and the Rio Nunez and Rio Pongo rivers of present-day Guinea were a leading nineteenth-century origin, 17 percent of the attributable mass in the African Origins database. Guinea is raised here by the western-Sudan Muslim-interior names (the African Origins Arabic/Islamic and Multiethnic catch-alls with no coastal homeland), credited to the Futa Jallon Islamic state and the upper-Guinea Fula behind the southern rivers, the most persistent post-1807 embarkation zone. The band is wide to reflect this uncertain interior mass.

Senegal (0.20). Present-day Senegal carries part of the nineteenth-century attributable share (13 percent in the African Origins database, led by the Mandinka and Fula) and a substantial part of the western-Sudan Muslim-interior names (the African Origins Arabic/Islamic and Multiethnic catch-alls with no coastal homeland), credited to Futa Toro and the upper Senegal. The al-Hajj Umar jihad from Futa Toro added captives from the Wolof, Tukulor, and Mandinka of the valley; the Sine-Saloum and Casamance trade ran late, the groundnut trade replacing it only by the 1860s (Ajayi and Oloruntimehin 1976, 209). The band is wide.

Mali (0.20). The deep-interior Mande, Bambara, and Soninke regions of present-day Mali (Kaarta, Segu, Bambuk, Khasso) kept feeding the upper-Senegal route into the nineteenth century, intensified by the al-Hajj Umar jihad across the western Sudan. Mali takes a large part of the western-Sudan Muslim-interior names (the African Origins Arabic/Islamic and Multiethnic catch-alls with no coastal homeland), on top of its 7 percent of the attributable mass: Traore (2023, 56), drawing on Faidherbe, names Kaarta, Segu, Beledugu, and Bambuxu among the originating provinces of the Soninke and Malinke diaspora still sending captives toward the coast. The band is wide to reflect this uncertain interior mass.

Mauritania (0.04). A Saharan-edge share belongs to present-day Mauritania for the desert-edge Moors of the lower-Senegal north bank (Trarza, Brakna) and the Soninke of Gidimaxa. The lower-Senegal gum coast and the al-Hajj Umar jihad tied this Saharan margin into the nineteenth-century trade; the AO Arabic/Islamic catch-all plausibly includes these Saharan-fringe captives, credited here with a wide band.

Guinea-Bissau (0.03). The Bissau and Cacheu rivers ran a small illegal nineteenth-century trade through the Portuguese ports, with the Bissau-homeland coastal peoples a minor origin, about 2 percent of the attributable mass in the African Origins database. Ajayi and Oloruntimehin (1976, 205) list Bissau among the West African coastal trading stations that persisted through the century. This is a coastal homeland share.

Gambia (0.02). The Gambia became a British anti-slave-trade base early in the century: Fyfe (1976, 183-84) records the settlement on Banjul Island from 1816, renamed St Mary’s, where recaptives were settled as Bathurst. This naval and administrative role cut the river’s part in exports of its own homeland peoples, so the Gambia-valley Mandinka contribution to surviving embarkations was small, about 1 percent of the attributable mass, with the Gambia-valley Mande largely counted under Senegal.

Sierra Leone (0.25). Green (2012, 216) records that the Sierra Leone region was ‘disrupted by violent wars in the 1550s and the 1560s’ and that ‘Records of slave ethnicities from the Americas reveal many Sape slaves as a result of these conflicts (cf. Tables I.1 and I.2).’ The Sape and the peoples known collectively under that name inhabited the coasts and hinterland of modern Sierra Leone. Green (2012, 234-235) further specifies that the Sape territory south of Cape Verga encompassed Tagunchos, Bagas, Sapes, Volons, Temnes, Limbas, and Jalonkés: the core ethnic groups of the Sierra Leone peninsula and its immediate hinterland. Green (2012, 206) describes pre-existing networks linking Bainunk in the north and Biafada and Sape in the south. Together these passages confirm that Sierra Leone itself was the primary source zone within the Mane-Sape conflict zone during the pre-1600 period. The catchment also reached the near-interior Mel-cluster hinterland of the modern Sierra Leone core (Temne, Bullom, Limba, Loko) that the Mane-Sape wars of the 1550s-60s fed to the coast but which the fragmentary American-sale records of this period could not pin to an embarkation point.

Guinea-Bissau (0.22). The only direct sixteenth-century catchment data for the Upper Guinea littoral come from Green’s Tables I.1 (1547-60) and I.2 (1560-1600), which record the ethnic composition of Upper Guinean slaves in the Americas. In Table I.1, Biafara (22.9%), Brame (19.95%), and Mandinga (19.05%), all principally peoples of the Guinea-Bissau rivers, dominate the sample, with Sape (8.84%) and Sereer (8.16%) also present. Green (2012, 33) identifies the coastal peoples of present-day Guinea-Bissau by name: Bainunk-Kassanké, Balanta, Biafada, Bijagó, Brame, Cocoli, Floup, Landuma, and the Nalu. The Bissau rivers were the principal early embarkation zone, yielding the highest proportional share of the pre-1600 Sierra Leone-region catchment.

Guinea (0.21). Green (2012, 234) records that ‘the peoples inhabiting the coastlines of present-day Guinea-Conakry and Sierra Leone had a long history of cultural sharing prior to the arrival of Atlantic traders’ and that the people known collectively as Sapes were ‘mixed in with many other peoples.’ Green (2012, 235) further specifies that the Sape coast ‘south of Cape Verga to Sierra Leone incorporated many different groups, including Tagunchos, Bagas, Sapes, Volons, Temnes, Limbas and Jalonkés.’ Several of these groups (particularly the Bagas, Nalus, and Landumas) occupied the coastline of modern Guinea-Conakry. Rodney (1975, 276) lists among the coastal polities of the region the ‘Nalu, Baga, Limba and Kokoli (all of the Mel language cluster),’ many of whose homelands straddle the Sierra Leone-Guinea border. The catchment ran inland to the Susu, Baga, Nalu and Jalonke of the modern Guinea highlands and Nunez/Cape Verga hinterland, the deepest-reaching named source the coastal records of the near-zero pre-1600 flow could not resolve to a port.

Mali (0.16). Mandinga captives in the sixteenth-century samples trace to the Mali-derived Mande presence in the upper Guinea hinterland. Green (2012, 33) describes how ‘the Mandinka presence operated as an arm of the powerful trading empire of Mali,’ and Green (2012, 34) establishes that ‘The pattern linking Senegambia and Upper Guinea to Mali dates to the latter half of the thirteenth century.’ The deep-interior Mandinka homeland straddles modern Mali and the upper Niger. This Mande network funnelled interior captives toward the coast, a portion of whom originated in the territory of modern Mali. Green (2012, 238) traces the Mane war-leaders to a blacksmith clan, the Camara, of the Konyan highlands on the upper-Niger borderland of modern Mali and Guinea, and the Mali-derived Mande network funnelled interior captives whose specific homeland the coastal samples never named. Wide band, near-zero documented volume.

Senegal (0.10). Green’s sixteenth-century samples include a sizeable Jolof/Wolof and Sereer component alongside a large Mandinga share, peoples principally of modern Senegal. Table I.1 (Green 2012, 23) records Mandinga at 19.05% and Sereer (listed as Berbesi) at 8.16%. Green (2012, 33) explains the political geography: ‘the Jolof polity consisted of five sub-kingdoms, each with a viceroy, whereas the Mandinka presence operated as an arm of the powerful trading empire of Mali.’ Green (2012, 34) traces the link further: ‘The pattern linking Senegambia and Upper Guinea to Mali dates to the latter half of the thirteenth century.’ These Senegambian peoples were thus embedded in the upper Guinea trade network from which the Sierra Leone region drew its captives.

Liberia (0.04). The Kru and Mel peoples of the modern Liberia coast (De, Bassa, Belle, Kru, and Gola) lay at the southern edge of the Mane displacements, but the Atlantic trade had not yet reached deeply into that stretch of coast in the sixteenth century. Rodney (1975, 276) lists ‘the Kwa-speaking De, Bassa, Belle and Kru’ among the region’s coastal polities. Their share is minimal because the Mane advance concentrated captive flows in the Sierra Leone peninsula and the rivers immediately to the north, not yet pushing the trade frontier as far south as Cape Palmas. The Gola, Kissi and Vai of the Cape Mount-Gallinas borderland, displaced by the Mane advance, sat at the southern edge of a catchment whose pre-1600 documentation is near-absent.

Gambia (0.02). Some Mandinka and Wolof captives from the Gambia river states fall within the modern Gambia, representing a small fraction of the early upper Guinea catchment. Green (2012, 33) documents the migration and Mandinka expansion in the rivers region. Rodney (1975, 280) notes that Niumi, Badibu, Niani and Wuli (on the right bank of the Gambia) ‘maintained the same frontiers for centuries,’ confirming these were stable Gambia-region polities contributing to the regional trade. The Gambia’s contribution is small because its captives fed primarily the Senegambia region’s own embarkation ports rather than those of the Sierra Leone region.

Sierra Leone (0.38). By the early seventeenth century the Mane polity had consolidated into ‘a loose association of states, with a few strong kings at Cape Mount, Sherbro, the Sierra Leone peninsula and Port Loko’ (Rodney 1975, 278). The Mane and Loko ‘advanced as ethnicities through penetration of the Gola, Kissi, Bullom, Krim and Temne’ (Rodney 1975, 279), displacing and absorbing the Mel-speaking groups of the Sierra Leone hinterland. The Vai and Koya continued to rule in the Gallinas and Cape Mount (Rodney 1975, 279). Nwokeji (2011, 91) in the Cambridge World History of Slavery, Vol. 3, confirms that the Sherbro’s involvement in the slave trade extended back to the fifteenth century, predating the full Mane consolidation. These overlapping coastal and near-interior polities kept Sierra Leone itself as the principal embarkation zone throughout the seventeenth century. The catchment also reached the modern Sierra Leone near-interior (emerging Mende, Loko, Temne, Bullom, Limba) absorbed by the Mani penetration but unnameable to a port in the sparse 17th-c record.

Guinea (0.28). The northern Sierra Leone-region ports on the Scarcies and the Rio Pongo-Rio Nunez fringe drew on the Susu, Baga, Nalu, Landuma, and Jalonke of the modern Guinea coast and hinterland. Rodney (1975, 280) notes that ‘The strongest representation of the Mande was in the north-west, where the Malinke, Susu and Dyalonke were all prominent,’ and that ‘A few reached as far south as the upper Scarcies, where the Susu of Bena were strong enough to repulse the Mani.’ This places the Susu firmly in the territory of modern Guinea, feeding captives into Sierra Leone-region embarkation points. Green (2012, 27) confirms that by the later sixteenth century the trade’s main accent had shifted, but the upper Guinea coast connections persisted into the seventeenth century. The catchment leaned to the Guinea interior and coast: the Susu, Baga, Nalu and Jalonke of the modern Guinea littoral and the Scarcies-Melakori fringe, the inland reach of the consolidating Mani/Sape world that the thin 17th-c records could not resolve to a named port.

Mali (0.13). Deep-interior Mande and Malinke captives moved down the trade routes from the Mali-derived states. Rodney (1975, 280) observes that ‘A great deal of the trade oriented towards the Atlantic had its origins in the deep hinterland, sometimes passing along well-established trade routes and at other times percolating slowly through the many intervening societies,’ and that this trade meant ‘a meeting of the two powerful forces of the Europeans and the Mande, respectively, often at the expense of the other residents of the upper Guinea coast.’ A portion of these interior captives originated in the territory of modern Mali, funnelled coastward through the Mande commercial network. The deep Mande hinterland feeding the coast down the savanna-to-forest trade routes that Rodney (1975, 280) describes contributed a portion of captives who originated in modern Mali but whose homeland the coastal trade did not record. Wide band.

Guinea-Bissau (0.10). The Bissau-rivers peoples (Brame, Biafada, Bainunk, Balanta, and Pepel) remained a documented part of the seventeenth-century upper Guinea trade, though their share declined as the main commercial axis shifted. Green (2012, 27) notes that ‘although the trans-Atlantic slave trade remained important in Western Africa, the main accent moved gradually to West-Central Africa.’ Green (2012, 33) documents the Guinea-Bissau river peoples as the backbone of the sixteenth-century catchment; their continued presence in the seventeenth century is attested by the persistence of the Cacheu trade and the Biafada networks, though at a reduced fraction of the Sierra Leone-region total.

Liberia (0.07). The Mane displacements pushed Kru, De, Belle, Gola, and Vai toward the Sierra Leone-Cape Mount coast. Rodney (1975, 279) records that ‘The Mende and Loko advanced as ethnicities through penetration of the Gola, Kissi, Bullom, Krim and Temne,’ pushing some groups southward into the Liberia borderlands. Rodney (1975, 279) also confirms that ‘The Vai and the Vai-speaking Koya continued to rule in an unbroken line in the Gallinas and Cape Mount,’ placing Vai-speaking peoples at the Liberia border zone. These populations at the southern edge of the Sierra Leone catchment contributed a modest fraction of the seventeenth-century embarkations, the Gallinas and Cape Mount edge (Gola, Kissi, Vai) at the southern margin of the consolidating Sierra Leone catchment.

Senegal (0.04). A Wolof, Sereer, and Mandinka spill from the northern reaches reached the Sierra Leone-region ports, though the bulk of captives from those peoples fed Senegambia’s own embarkation points. Green (2012, 33) records that ‘the Jolof polity consisted of five sub-kingdoms, each with a viceroy, whereas the Mandinka presence operated as an arm of the powerful trading empire of Mali,’ establishing the Senegambian peoples’ integration into the broader upper Guinea trade network from which the Sierra Leone region drew a small northern fraction.

Sierra Leone (0.40). The eighteenth century is the dominant-mass period for Sierra Leone-region embarkations. Nwokeji (2011, 91) in the Cambridge World History of Slavery, Vol. 3, confirms that ‘When slavery emerged among the sparsely populated Sherbro of Sierra Leone in the eighteenth century, it was in response to the transatlantic slave trade, which the Sherbro had been involved in since the fifteenth century.’ Nwokeji (2011, 98, fn. 6) adds that the Mane invasion of circa 1545 had already ‘displaced the population, driving people closer to the sea,’ creating the coastal settlement patterns that defined eighteenth-century embarkation geography. The coastal and near-interior Sierra Leone hinterland (Temne, Bullom, Limba, Sherbro, and later Mende) supplied the bulk of directly embarked captives, with the Futa Jalon jihad adding interior volume (accounted for under Guinea).

Guinea (0.39). The Futa Jalon jihad (1725-1776) is the defining eighteenth-century supply engine and pushes the catchment deep into modern Guinea. Levtzion (1975, 209) records that ‘Reports from Sierra Leone in 1751 indicated a prodigious trade in slaves, due to an increasing supply from the Futa Jalon.’ Levtzion (1975, 210) documents the jihad’s consequences for the inhabitants of the Futa Jalon itself: ‘The non-Fulani inhabitants of the Futa Jalon — Dyalonke, Tenda and Landouma — were subjugated, and their conquered villages turned into rundes, villages of serfs or slaves.’ Levtzion (1975, 211) confirms the broader effect: ‘Many of the slaves purchased by the Europeans on the upper Guinea coast in the latter part of the eighteenth century were produced by the raids of the Fulani on their neighbours.’ The Futa Jalon lies almost entirely within modern Guinea, making Guinea the dominant second-tier source. The defining engine of the period was the Futa Jalon jihad (1725-1776), whose campaigns Levtzion (1975, 209-211) shows drove a ‘prodigious trade in slaves’ off the Futa highlands of modern Guinea (Dyalonke, Tenda, Landuma, Susu, Kissi, Kuranko), a far larger inland share than the 19th-c recaptive names show and the bulk of what the coastal embarkation records did not name.

Mali (0.11). The Futa Jalon jihad’s wars reached the upper Niger, drawing in captives from the territory of modern Mali. Levtzion (1975, 209) records that in 1762 the combined Fulani-Solima forces were defeated by Konde Birama, ruler of Wasulu: ‘The Wasulunke people were a fusion of Fulani migrants with the Bambara beyond the Sankarani river.’ This places the catchment zone of these conflicts in the upper Niger region of modern Mali. Levtzion (1975, 211) also records that the Futa Jalon’s influence ‘rekindled Islamic militancy in the Futa Toro,’ extending the jihad’s catchment northward. A portion of the resulting captives originated in modern Mali, on the upper Niger where Levtzion (1975, 209) records the Futa Jalon wars reaching the Wasulu (a fusion of Fulani migrants with the Bambara beyond the Sankarani), the deep-Sahel captives the coast could not name.

Liberia (0.06). The southern Sierra Leone-region ports at the Gallinas and Cape Mount drew on Gola, Kissi, and Vai populations whose homelands straddle the modern Sierra Leone-Liberia border. Rodney (1975, 279) confirms that Mende and Loko penetration of the Gola, Kissi, Bullom, Krim, and Temne continued across the seventeenth and eighteenth centuries, pushing some displaced groups toward the Liberia borderland. The Liberia contribution is modest because the Gallinas and Cape Mount ports served primarily as the southernmost extension of the Sierra Leone catchment, not as a distinct large-scale embarkation zone in this period; the Gallinas and Cape Mount ports drew on Gola, Kissi and Vai of the Sierra Leone-Liberia borderland at the southern end of the 18th-c catchment.

Guinea-Bissau (0.02). By the eighteenth century the Guinea-Bissau rivers’ contribution to Sierra Leone-region embarkations had largely faded, as the Cacheu and Bissau trades ran through their own northern ports and were captured by the Senegambia and Upper Guinea regions’ own embarkation statistics. Green (2012, 27) notes that ‘the main accent moved gradually to West-Central Africa,’ reflecting the structural shift away from the Guinea-Bissau littoral that had dominated the sixteenth-century trade. The share reflects continuing but reduced Biafada and Balanta involvement in the wider upper Guinea network.

Senegal (0.02). A Wolof, Mandinka, and Tukulor spill from the north reached Sierra Leone-region ports. The Futa Toro jihad of 1776 was a Senegal-side engine, but its captives fed primarily Senegambia’s own embarkation points. Levtzion (1975, 211) records that ‘influence from the Futa Jalon caused greater Islamic commitment in Bondu and rekindled Islamic militancy in the Futa Toro,’ with the Futa Toro lying in the Senegal river valley. Captives from that conflict filtered south only in small numbers relative to those absorbed by the Senegambia region directly.

Sierra Leone (0.57). The nineteenth-century share is anchored on the African Origins database, which records the names of recaptives liberated by the Royal Navy and landed at Freetown between 1819 and 1845. Fyfe (1976, 181) documents the demographic transformation: ‘The anti-slave trade campaign caused a demographic revolution in Sierra Leone,’ with the ‘Liberated Africans’ brought by naval ships ‘from homelands extending from the Senegal to the Congo.’ The African Origins database shows Sierra Leone as 72.1% of attributable mass in the Sierra Leone-region catchment, dominated by Mende (27.9%), Temne (7.4%), and Mende or Sherbro (7.3%). This is consistent with Eltis and Engerman (2011, 14) in the Cambridge World History of Slavery, Vol. 3, who note that nearly all twins in a sample of 57,000 recaptives landed in Sierra Leone between 1819 and 1845 were on vessels departing Bonny, New Calabar, and Old Calabar, demonstrating the AO database’s coverage of diverse coastal origins funnelled through Freetown. Kono coding correction: the African Origins ‘Kono’ group (5.1%, n250) is mapped to Nigeria in the raw anchor, but Kono is a Mande people of eastern Sierra Leone, not the Nigerian Kono; that mass is re-credited to Sierra Leone, which is why the SLE share rises above the 72.1% attributable-name anchor while the deep-interior catchment is credited inland.

Guinea (0.185). The African Origins database assigns Guinea 8.7% of the attributable mass in the Sierra Leone-region nineteenth-century catchment, carried primarily by the Kuranko (87% Guinea), Kissi (81% Guinea), and Limba (13% Guinea) recaptive groups. Scaled by the 0.73 attributable fraction, Guinea accounts for roughly 6.5% of total Sierra Leone-region embarkations in the 1800s. The largest share of the 27% unattributable 1800s mass (African Origins ‘Multiethnic’ 16.5% + ‘Arabic/Islamic’ 9.8%, names with no resolvable homeland) plus deep-interior captives belongs to the Futa Jalon jihad state of modern Guinea, the Sahelian-Islamic and highland frontier the recaptive names point toward but cannot resolve; this lifts GIN well above its 8.7% attributable-name share. Wide band.

Mali (0.121). The African Origins database assigns Mali 4.2% of the attributable mass in the Sierra Leone-region nineteenth-century catchment, carried by the Fula (Upper Guinea) recaptive group (homeland Mali 100% in the origin mapping), reflecting the deep-Sahelian Fula networks that continued to channel interior captives toward the Sierra Leone coast. Scaled by the 0.73 attributable fraction, Mali accounts for roughly 3.1% of total embarkations. A large part of the unattributable Islamic/Multiethnic catch-all and deep-Mande captive mass belongs to the Sahel interior of modern Mali (Kaarta, Segu, Bambuk, Khasso), intensified by the al-Hajj Umar jihad across the western Sudan, the deepest catchment the Freetown names label only as ‘Islamic’. Wide band.

Liberia (0.095). The African Origins database assigns Liberia 3.6% of the attributable mass in the Sierra Leone-region nineteenth-century catchment, carried by the Kissi-or-Kono (17% Liberia) and Kissi (9% Liberia) recaptive groups. Scaled by the 0.73 attributable fraction, Liberia accounts for roughly 2.6% of total Sierra Leone-region embarkations in the 1800s; the share is lifted slightly toward the anchor to reflect the Gallinas ports’ continued role on the Liberia border. The Gallinas and the Liberia-border hinterland (Gola, Kissi, Vai), the deep-interior southern edge, are under-resolved in the recaptive names relative to the persistent post-1807 Gallinas embarkations.

Senegal (0.025). The African Origins database assigns Senegal 3.4% of the attributable mass in the Sierra Leone-region nineteenth-century catchment, carried primarily by the Mandinka recaptive group (90% Senegal). Scaled by the 0.73 attributable fraction, Senegal accounts for roughly 2.5% of total embarkations, reflecting Mandinka and Wolof captives reaching Sierra Leone-region ports from the north.

Guinea-Bissau (0.004). The African Origins database assigns Guinea-Bissau 0.6% of the attributable mass in the Sierra Leone-region nineteenth-century catchment, carried by the Mandinka group (10% Guinea-Bissau). Scaled by the 0.73 attributable fraction, Guinea-Bissau accounts for roughly 0.4% of total embarkations: a near-token share reflecting the near-complete shift of the Bissau-rivers trade to northern embarkation points by the nineteenth century.

Sierra Leone (0.50). Coastal Sape core of the modern Sierra Leone littoral (Bullom, Temne, Limba, Krim, Sherbro), the dominant source within the near-zero pre-1600 flow. Green (2012, 236) records that the Mane invasions of the 1540s-1570s drove these riverine Sape peoples into the trade; Green (2012, 237, paraphrasing Rodney) notes that by the time Mane forces reached Sierra Leone ‘many were Bulloms and Temnes from that very region.’ The Sierra Leone forest-Mande interior (Kono, Koranko) that the coastal evidence cannot name separately is also credited here; the deeper catchment goes to the highland and hinterland zone rather than to this coastal baseline.

Guinea (0.28). The northern Windward catchment (Guinea-Conakry coast and its highland hinterland): Susu, Baga, Cocoli, Nalu and other Mel and Sape peoples whose homelands lie in modern Guinea, plus the Guinea-highland interior that the coastal records could not name. Green (2012, 234-235) shows the Sape kingdom ran ‘south of Cape Verga to Sierra Leone’ incorporating Bagas, Temnes, Limbas and Jalonkes; Green (2012, 237) records a 1564 ship ‘with a large majority of Sape and Susu slaves.’ The Susu are a Guinea people. The deep catchment behind this stretch of coast was the Guinea highlands, credited here.

Liberia (0.115). The southern Windward catchment toward Cape Mount and Cape Palmas (modern Liberia), and its near-coast interior. Mani military power was ‘located near Cape Mount,’ its influence felt ‘as far east as the Grand Cess river, as a result of accompanying Kru displacements’ (Rodney 1975, 278), pushing Gola, De, Belle and Vai or Quoja peoples toward the coast. Kru maritime peoples themselves ‘resisted enslavement’ (Iliffe 2007, 133), so the credited mass is the displaced interior peoples, not the Kru coast. The Mane displacements opened the Liberian hinterland to enslavement, credited here.

Mali (0.055). A deep-interior Mande component traced to the Mane warriors’ hinterland origin: Green (2012, 238) records that ‘the Mane leaders came from a blacksmith clan in the Konyan highlands, the Camara,’ placing the invasion force in the upper-Niger borderlands of modern Mali and Guinea. The upper-Niger Mande edge is credited here on this basis. The share is speculative for this period and the band is very wide given the absence of documented embarkations.

Cote d’Ivoire (0.02). A thin south-eastern Windward interior component. Rodney (1975, 276) identifies ‘the Dan and Gouro (eastern Mande)’ as ‘the intrusive elements in the hinterland of Cape Palmas,’ the southernmost point of the Windward Coast, and the Cambridge World History of Slavery (Vol. 3) classes ‘especially Liberia and Ivory Coast’ as part of the Windward zone. The Cape Palmas hinterland (Dan, Gouro) of modern Cote d’Ivoire is credited here. The Kru and stateless peoples of this interior ‘resisted enslavement’ (Iliffe 2007, 133), so the share is kept small with a wide band.

Senegal (0.02). Mandinka and Mande hinterland trading networks linked Upper Guinea to the Senegal-Gambia savanna (Green 2012, 2256). For the pre-1600 period the Windward Coast carried essentially no documented embarkations, so this is a thin savanna-edge trace; the deeper catchment goes to the closer highland and forest-interior zone.

Guinea-Bissau (0.01). A small Mandinka and Kaabu-edge component; most Bissau-region peoples (Biafara, Brame, Bainunk) shipped from the Senegambia and Rivers ports, not the Windward Coast, so this is kept near-zero.

Sierra Leone (0.50). Post-Mane Sierra Leone littoral: Temne, Bullom, Limba, Loko, Sherbro of the modern Sierra Leone core. By the 1620s the Mane had ‘settled in relative peace and indeed adopted many Sape customs’ (Green 2012, 236) and the coast consolidated around states ‘at Cape Mount, Sherbro, the Sierra Leone peninsula and Port Loko’ (Rodney 1975, 278). The Sierra Leone forest-Mande interior the coastal records cannot name is also credited here; the deeper catchment goes to the highland and hinterland zone rather than to this coastal baseline.

Guinea (0.28). The Guinea-Conakry coastal zone (Susu, Baga, Cocoli, Nalu) and the Scarcies/Melakori fringe, plus the Guinea highland interior behind it, continued to feed the small 17th-c flow. Rodney (1975, 278) records that ‘the Limba retook the initiative in the seventeenth century’ and that hostilities developed between ‘the Loko and the Koya-ruled Temne’ along the modern Guinea-Sierra Leone boundary. The deep catchment behind this coast was the Guinea highlands, credited here.

Liberia (0.115). Mane displacements pushed Kru, De, Belle, Gola and Vai or Quoja peoples toward the Cape Mount coast. Rodney (1975, 279) records that ‘the Mende and Loko advanced as ethnicities through penetration of the Gola, Kissi, Bullom, Krim and Temne’ and that ‘the Vai ... continued to rule in an unbroken line in the Gallinas and Cape Mount.’ Kru resistance to enslavement (Iliffe 2007, 133) kept the Kru coast outside the trade, so the credited mass is the displaced interior peoples. The Liberian hinterland is credited a substantial share here.

Mali (0.055). A thin interior-Mande tail through Mandinka dyula and Mande-state networks reaching the coast from the savanna hinterland. Rodney (1975, 276) notes the Mande ‘invariably forged powerful ties between the savanna hinterland and the forest zone.’ The upper-Niger Mande edge of modern Mali is credited here. Speculative and near-zero for the 17th c, with a very wide band.

Cote d’Ivoire (0.02). A thin south-eastern Windward interior component. The Dan and Gouro (eastern Mande) were ‘the intrusive elements in the hinterland of Cape Palmas’ (Rodney 1975, 276), the southernmost Windward point, and the Cambridge World History of Slavery (Vol. 3) classes ‘especially Liberia and Ivory Coast’ in the Windward zone. The Cape Palmas hinterland of modern Cote d’Ivoire is credited here; Kru and stateless peoples there ‘resisted enslavement’ (Iliffe 2007, 133), so the share is small with a wide band.

Senegal (0.02). A small Mandinka and savanna-edge tail; near-zero mass for the 17th-c Windward Coast, with the deeper catchment going to the closer highland and forest interior.

Guinea-Bissau (0.01). A small Kaabu and Mandinka-edge tail; most Bissau-region peoples shipped from the Senegambia and Rivers ports, keeping this near-zero.

Sierra Leone (0.47). Core embarkation zone of the peak trade: the Sierra Leone estuary (Bance Island, the Sherbro, the Scarcies) and its hinterland (Temne, Bullom, Limba, Loko, Mende, Koranko, Kono). Levtzion (1975, 208) records ‘a prodigious trade in slaves’ from Sierra Leone in 1751 ‘due to an increasing supply from the Futa Jalon,’ and Rodney (1975, 292) documents rice-victualling of slave ships in the hinterland ‘after such developments had been greatly intensified by the Futa Jalon jihad.’ The Sierra Leone forest-Mande interior is also credited here; the deeper catchment goes to the Futa Jalon (Guinea) engine that drove this period rather than to the coastal baseline.

Guinea (0.28). Guinea carries the largest interior catchment because the dominant 18th-c supply engine was the Futa Jalon highland theocracy. Iliffe (2007, 145) records that ‘Futa Jalon ... became a major slave exporter, with the most state-controlled economy in West Africa and exceptional dependence on agricultural slaves,’ and Rodney (1975, 292) that ‘the Futa Jalon jihad displaced some of the Susu and Dyalonke and pushed them towards the coast.’ The deep highland catchment (Susu, Yalunka, Kissi, Kuranko, Baga, Limba) that the coastal records could not name is credited here, raising Guinea well above the 19th-c recaptive anchor.

Liberia (0.13). The 18th-c Windward Coast ‘especially Liberia and Ivory Coast’ (Price 2011, 518). Rodney (1975, 293) records that ‘the Malinke entrenched themselves as far east as the Lofa river, where the important state of Kondo at Bopolu dominated the Loma people’ and ‘controlled the route from Cape Mount to the deep hinterland, via Kissi country.’ These are modern Liberia peoples (Loma, Kpelle, Vai, Gola, Kissi). The Kondo/Bopolu interior route is the channel credited a substantial interior share here, above the 19th-c anchor.

Mali (0.065). The Futa Jalon raids and Mande trade networks reached the upper-Niger savanna of modern Mali. Levtzion (1975, 208-209) records that in 1762 combined Fulani and Solima forces invaded Wasulu, ‘whose ... Wasulunke people were a fusion of Fulani migrants with the Bambara beyond the Sankarani river.’ The upper-Niger borderland where Futa Jalon raiding overlapped with Bambara country is credited here. Inland edge of the catchment, wide band.

Cote d’Ivoire (0.025). The Voyages Windward region extended to the Ivory Coast (Price 2011, 518, ‘especially Liberia and Ivory Coast’). Rodney (1975, 276) identifies ‘the Dan and Gouro (eastern Mande)’ in ‘the hinterland of Cape Palmas,’ the southernmost Windward point. The Cape Palmas hinterland of modern Cote d’Ivoire is credited here. Kru and stateless peoples there ‘resisted enslavement’ (Iliffe 2007, 133) and the coast was primarily an ivory rather than a slave coast, so the share is small with a wide band.

Senegal (0.02). A thin Mandinka and savanna-interior tail through Mande trade networks; near-zero mass for the 18th-c Windward Coast, with the deeper catchment going to the closer highland and forest-interior zone.

Guinea-Bissau (0.01). A small Kaabu and Mandinka-edge tail; most Bissau-region captives shipped from the Senegambia and Rivers ports, keeping this near-zero.

Sierra Leone (0.627). Anchored on the African Origins 19th-c recaptive data (Sierra Leone 69.1% of attributable mass: Mende, Temne, Loko, Mende-or-Sherbro, plus the mis-coded Kono cohort re-credited to Sierra Leone). The part of the ‘Islamic/Multiethnic’ recaptive-name pool that, on the Windward Coast, belongs to the Mande-speaking forest interior of Sierra Leone (Kono, Koranko, Mende hinterland) rather than to a distant Sahel homeland is credited here. The bulk of that pool still goes to the highland and Sahel interior rather than to the coastal anchor.

Guinea (0.176). Anchored on the African Origins data (Guinea 11.3% of attributable: Kissi, Kuranko, Susu, Yalunka). Guinea takes a large share of the ‘Islamic/Multiethnic’ pool because the Futa Jalon theocracy was still actively raiding into the 19th c (Iliffe 2007, 157: ‘a major revolt began in Futa Jalon in 1845 when slaves flocked to a dissident marabout’), so the no-homeland highland and Fulbe captives entering via Futa Jalon are credited to Guinea’s highland catchment.

Mali (0.087). Anchored on the African Origins data (Mali 4.9% of attributable, carried by the Fula Upper Guinea recaptive group). Mali takes a large share of the ‘Islamic/Multiethnic’ recaptive-name pool, which on the Windward Coast is the deep-Sahel Fulbe and savanna captives entering through the highland routes via Futa Jalon - precisely the Muslim-Sahelian names the African Origins linguists could not assign to a coastal homeland. Wide band reflecting the thinly documented interior.

Liberia (0.074). Anchored on the African Origins data (Liberia 4.5% of attributable: Kissi, Kissi-or-Kono, Vai or Gola around the Gallinas and Cape Mount). The Liberian Mande-forest hinterland (Loma, Kpelle, Gola, Kissi), the deep catchment behind the Gallinas and Cape Mount that the recaptive name codes could not place, is credited here. The Kru coast stayed outside the trade (Iliffe 2007, 133).

Senegal (0.022). Anchored on the African Origins data (Senegal 2.9% of attributable, via the Mandinka recaptive group), reflecting savanna and Senegambia-edge captives reaching the Windward Coast through interior trade routes.

Cote d’Ivoire (0.01). A thin south-eastern interior component for the 19th c. The Voyages Windward region extends to the Ivory Coast (‘especially Liberia and Ivory Coast’), whose Cape Palmas hinterland held the Dan and Gouro (eastern Mande) (Rodney 1975, 276). The Cote d’Ivoire forest interior is credited a small share here; the African Origins data show no measurable Ivory Coast recaptive origin, so the share is kept small with a wide band.

Guinea-Bissau (0.004). Anchored on the African Origins data (Guinea-Bissau 0.5% of attributable, a trace Mandinka and Kaabu-edge component). Near-zero share.

Nigeria (0.00). The African Origins database’s nominal 5.5% Nigeria attribution is a homeland-coding artifact: the ‘Kono’ group (n=20, coded Nigeria 100%) and ‘Kissi or Kono’ (n=9) are the Sierra Leonean Kono of the Mande-speaking forest zone, not the distant Kono of the Nigeria-Cameroon Mandara hills the homeland code points to. A genuine Nigerian origin for a Windward Coast recaptive is historically implausible (the Bight catchments shipped from their own ports). The mid-point is set at zero with a small upper bound for coding ambiguity.

Ghana (0.78). Through the sixteenth century the Gold Coast was a net importer of slaves, not an exporter: the Akan goldfield states bought between ten thousand and twelve thousand captives from the Portuguese between 1500 and 1535 to carry imports inland and clear forest (Iliffe 2007, 133), and transatlantic exports did not begin until the 1640s (Nwokeji 2011, 90). There is essentially no measurable outflow before 1600, so this cell carries near-zero export mass. Any tiny trickle of captives shipped from the region would have been local Akan war captives generated within the immediate forest and coastal zone, which lies overwhelmingly within modern Ghana; the band is wide to reflect that the figure is a prior over an essentially nonexistent flow.

Nigeria (0.22). The Akan imported eastern captives in this period rather than exporting them: after Benin ceased to export male slaves in 1516, most slaves sold to the Akan came from the Niger Delta and Igbo country to the east (Iliffe 2007, 134). The region was receiving these captives, but any individuals re-exported in the negligible pre-1600 outflow would carry an eastern-Nigerian origin. The Gold Coast’s import pool in this period contained a documented eastern-Nigerian component; the band is wide because the import flow is clear but the re-exported fraction is not, and the export mass is in any case near zero.

Ghana (0.78). Transatlantic exports begin in the 1640s and reach only about one thousand per year by the 1670s, and the captives actually exported were war captives generated by the rise of the militarised forest and coastal Akan states: Akwamu captured Accra in 1680 and powerful states such as Akwamu, Denkyira and later Asante emerged, their armies consisting of slaves in part (Iliffe 2007, 147; Nwokeji 2011, 91). The homelands of these states lie in the Akan forest and immediate hinterland of modern Ghana, so the dominant component of the seventeenth-century export stream was Akan war captives from within modern Ghana.

Nigeria (0.13). The Gold Coast in the seventeenth century was still importing eastern captives on a large scale: the Dutch purchased captives at Allada and Angola for sale on the Gold Coast during the second and third quarters of the century, and the English likewise bought captives at Allada for enslavement by Gold Coast indigenes (Nwokeji 2011, 90). Some of these eastern captives passing through Gold Coast ports could have been re-shipped to the Americas, so a Nigerian component is included as the documented eastern import source. The band is wide because the import flow is well documented but the re-exported fraction is not.

Togo (0.07). Akwamu’s late-seventeenth-century expansion reached east across the lower Volta toward the Ewe and Anlo districts of modern southeastern Ghana and Togo (Iliffe 2007, 147; Rodney 1975, 314). A small share of the early export stream originated among cross-Volta Ewe-speaking peoples whose homeland straddles the modern Ghana-Togo border. The band is wide because the Ewe homeland is split across the border and the cross-Volta catchment in this early period was modest and not separately quantified.

Burkina Faso (0.02). A small northern-frontier component applies. Even in the seventeenth century the militarised Akan states drew some captives from beyond the immediate forest, and the wider Mossi-Dagomba/Voltaic savanna to the north was already a zone of raiding and warfare that fed captives southward toward the emerging Gold Coast markets (Levtzion 1975, 184-185). The share is very small and the band wide because the seventeenth-century northern corridor was nascent compared with its eighteenth-century peak; this Voltaic savanna interior, not the Mali/Niger Sahel, fed the Gold Coast Atlantic ports, because that Saharan-facing interior fed the trans-Saharan trade instead.

Ghana (0.73). The eighteenth century carries about 84 percent of the region’s whole-period export mass and is dominated by Asante imperial warfare, which generated captives from conquered and tributary peoples whose homelands lie overwhelmingly inside modern Ghana: the metropolitan and inner-ring Akan chiefdoms, the conquered Akan states including Bono-Mansu (1722-3), and above all the outlying non-Akan tributaries of Gonja and Dagomba, from which Asante demanded a thousand slaves a year (Iliffe 2007, 147; Levtzion 1975, 184). The dominant part of the export stream is retained here, because the Gonja and Dagomba tributaries and the inter-Akan wars are within modern Ghana.

Burkina Faso (0.09). Asante’s northward thrust and the wider Mossi-Dagomba war zone fed captives southward from the Voltaic savanna. The Mossi states of Yatenga and Wagadugu (modern Burkina Faso) were a zone of persistent raiding, forcing the Dogon into the Bandiagara mountains and fighting the Samo, while the Bariba of Borgu raided caravans along the routes through Kotokoli to Salaga (Levtzion 1975, 185, 190). In this consequential northern-frontier century the Salaga corridor’s captive pool drew on Voltaic and Mossi country beyond the modern Ghana border. The band is wide because the sources document the raiding but cannot apportion how much of the corridor’s throughput originated in modern Burkina Faso rather than northern Ghana; this Voltaic zone, not the Mali/Niger Sahel, which fed the Saharan trade, is the relevant catchment.

Togo (0.09). Asante expanded east across the Volta into Eweland, Kwahu, Kete Krachi and the Kotokoli (Tem) districts (Rodney 1975, 314), and the northern slave corridor ran through Salaga toward the Kotokoli of modern northern Togo, where the Bariba of Borgu raided caravans along the route through Djougou and Kotokoli (Levtzion 1975, 190). Cross-Volta Ewe and Tem-Kotokoli captives entered the Gold Coast export stream in this frontier century. The band is wide because the Ewe and Tem catchment is split across the Ghana-Togo border and the sources do not partition it.

Nigeria (0.05). A small eastern component persists in the eighteenth-century Gold Coast export stream. Gold Coast exports lagged behind those from the Bight of Benin until the 1770s, so cross-shipping with the Bight was active during most of the century; resident Gold Coast brokers in Whydah to the east acted as correspondents to Gold Coast-based traders, and some captives acquired through this eastern connection were shipped from Gold Coast ports (Nwokeji 2011, 90). The nineteenth-century recaptive names show a persistent roughly fifteen-percent Nigerian signal at Gold Coast ports, which plausibly extends back into the late eighteenth century.

Cote d’Ivoire (0.04). Asante’s wars against Gyaman, Buna and the western Brong frontier reached into territory at the far west of the Gold Coast catchment: from Kong the route to the Akan forest passed through Buna and Bonduku, and Asante pressure on Gyaman culminated in Opoku Ware’s victory at Bonduku around 1740, intensifying a southward Brong trek into the forests that gave rise to the Agni (Anyi) and Baule ethnicities of the modern Ivory Coast southeast (Levtzion 1975, 183; Rodney 1975, 314). This western war zone is credited here. The band is wide, reflecting a western-frontier spill less prominent than the northern Gonja-Dagomba-Mossi catchment.

Ghana (0.66). The nineteenth-century African Origins database gives Ghana 85.1 percent of the attributable names (Akan names 45.5 percent, all with a Ghana homeland), with an attributable fraction of only about 53 percent of all recorded names; the remainder are catch-all Arabic, Islamic and multiethnic recaptive names. At a Gold Coast port whose surviving outflow rode on the Salaga northern-slave corridor, those catch-all names are overwhelmingly the Voltaic and northern-Ghanaian (Gonja/Dagomba) interior, so most of that mass belongs to Ghana, which holds the bulk of that northern catchment and the continuing Asante war captives. The band is wide to reflect that this rests on a thin name sample anchored at 0.851 times 0.53.

Burkina Faso (0.13). A Salaga-corridor share belongs to Burkina Faso. The catch-all Arabic/Islamic names that dominate the non-attributable mass are exactly the signature of the northern, Muslim-influenced Voltaic savanna that the nineteenth-century Salaga slave market drew on (Levtzion 1975, 190; Cambridge History of Africa Vol5). The thirty-three-name sample cannot resolve this northern corridor, so it is credited here as a small but real Voltaic/Mossi component. The band is wide, and this Voltaic zone rather than the Mali/Niger Sahel, which fed the trans-Saharan trade and not the Gold Coast Atlantic ports, is the relevant catchment.

Nigeria (0.12). The African Origins database gives Nigeria 14.9 percent of the attributable names at Gold Coast ports, driven by a Yoruba and Tiv signal; applied to the roughly 53-percent attributable fraction this yields about 0.08. The Yoruba wars of this period made the Bight of Benin the major slaving region with Lagos as its headquarters, and some Yoruba and eastern captives appear among the names embarked via the Gold Coast littoral (Inikori 2011, 668). The share is lifted modestly above the scaled value because the tiny n=33 sample plausibly under-counts the eastern signal; the band is wide accordingly.

Togo (0.09). The cross-Volta Ewe and Tem-Kotokoli homeland extends into modern Togo and remained part of the Gold Coast’s eastern catchment in the nineteenth century, as it had been since Asante’s expansion into the Eweland districts in the eighteenth century (Rodney 1975, 314). This component is not separately resolved in the thirty-three-name African Origins sample (Ewe captures map partly to the Akan-coastal signal in the anchor), so it is credited here as a small under-sampled border component. The band is wide to reflect the under-resolution.

Benin (0.82). The Bight of Benin Atlantic trade did not exist at scale before about 1640; the earliest ‘Slave Rivers’ attestation is late fifteenth-century and applied to the area east of Lagos (Law 1991, 13), and Manning’s (1982, 31) export series begins only in 1641. If any captives left before 1600, the near-coastal Aja peoples of modern Benin (Hueda, Allada, Hula/Popo, and the Fon and Aizo behind them) would dominate, as they did once the trade began. This row carries essentially no embarkation mass. The catchment also reached the Aja and Mahi interior of modern Benin behind the named coastal suppliers, consistent with the Aja-dominant trajectory.

Togo (0.10). The western Slave Coast ports at Anlo and Keta and the Popo towns straddle modern Togo and southeastern Ghana, and Law (1991, 14) places the Ewe homeland across the modern Ghana-Togo boundary, so any pre-1600 western-port captives would have drawn from this zone. The catchment reached the Ewe and Tem interior of modern Togo behind the western Slave Coast; this period carries essentially no measurable mass.

Nigeria (0.08). A small Yoruba and eastern component is geographically plausible for any pre-1600 trade through the Lagos and Benin channels, but Ajayi and Smith (1971, 123) record that Yoruba enslavement was inconsiderable before the nineteenth century. The interior Yoruba country of modern Nigeria is included for that channel; this period carries essentially no measurable mass and the catchment is undefinable from contemporaneous evidence.

Benin (0.813). In the seventeenth century the Bight of Benin trade was dominated by the Aja kingdoms of modern Benin. Allada, with Offra and later Jakin as its ports, was the ‘principal supplier of slaves to the coast’; Law (1991, 186) reports the Fon (‘Foin’) first named as a source of slaves in 1680, when ‘the most part’ of slaves sold at the coast came from there. Manning’s Table 2.2 (1982, 31) puts the Aja peoples at 94.9 percent of 1641-1700 Bight of Benin exports (217,900 of 229,700), and Manning (1982, 340) locates the entire Aja base population in modern Dahomey/Benin: ‘Aja - 656,000 (all in Dahomey)’. The catchment also reached the Aja and Mahi interior of modern Benin behind the named coastal suppliers.

Togo (0.094). The western Slave Coast ports at Anlo/Keta and the Popo towns (Little Popo and Great Popo) draw on the Ewe and Hula homelands of modern Togo and southeastern Ghana. Law (1991, 15) describes the ‘entire littoral from Aflao in the west to Jakin’ as the Hula/Popo zone, and Eltis’s port series in Falola and Childs (2004, 24) shows Popo departures totalling roughly 104,000 over the era, with 95 percent at Little Popo (modern Togo). The catchment also reached the western Ewe and Tem interior of modern Togo.

Nigeria (0.073). A small Nigerian component in the seventeenth century comes from two sources: Manning’s Table 2.2 (1982, 31) gives Yoruba at 5.1 percent of 1641-1700 exports (11,800 of 229,700), and Law (1991, 185) notes that some slaves reaching the coast came via ‘merchants from the kingdom of Benin, meaning perhaps from Lagos’. Eltis in Falola and Childs (2004, 29) judges that before 1726 ‘no more than 10 percent’ of Bight of Benin deportees were Yoruba speakers. The catchment also reached the western Yoruba (Anago) interior and the Lagos-channel hinterland of modern Nigeria.

Ghana (0.02). The Ewe homeland straddles the modern Ghana-Togo border (Law 1991, 14), and the extreme western Slave Coast sometimes extended west of the Volta to include the Adangme/Laduku zone of modern southeastern Ghana; Gold Coast immigrants and traders were also resident at Ouidah. The African Origins database records a persistent approximately 2.5 percent Ghana component in nineteenth-century Bight of Benin recaptive samples, corroborating a standing western-fringe contribution from this coastal fringe.

Benin (0.64). The eighteenth century remained Aja-dominated, but the catchment deepened into the interior of modern Benin. After Dahomey conquered Allada and Hueda (1724-1730), the Whydah trade was ‘supplied effectively solely from Dahomey’s own slave-raiding’ (Law 1991, 190), directed especially against the Mahi to the north. Manning’s Table 2.2 (1982, 31) puts Aja at 78.3 percent of 1701-1800 exports (873,500 of 1,115,100), and his Appendix 3 (1982, 340) places all 656,000 Aja in modern Benin; he locates 420,000 of 610,000 Eastern Voltaic people (the Bariba and Atacora peoples) in modern Benin too. The catchment reached the deeper Mahi and Bariba interior of modern Benin behind the Dahomian raiding frontier.

Nigeria (0.267). The Nigerian component rose through the eighteenth century as Oyo opened the eastern ports (Porto Novo, Badagri, Epe) and brought captives from far inland. Manning’s Table 2.2 (1982, 31) gives, for 1701-1800, Yoruba at 11.4 percent (127,400), Nupe at 0.7 percent (7,900), and Hausa at 0.9 percent (10,500). Law (1977, 226) confirms that ‘many slaves were purchased, specifically from the northern neighbours of Oyo, the Nupe and Bariba’, and (1977, 227) that Adams ‘refers to the large numbers of Hausa slaves brought to Porto Novo by Oyo traders’. The catchment reached the interior Yoruba, Nupe, Bariba, and Hausa country of modern Nigeria fed to the coast by Oyo.

Togo (0.074). The western Aja and Ewe homelands of modern Togo (Anlo, Little Popo, Great Popo) continued to supply captives in the eighteenth century through the Popo ports; Eltis in Falola and Childs (2004, 24) shows ‘Popo departures amounted to just one-fifth of Ouidah’s’ in 1726-50. Manning (1982, 31) identifies the Atacora mountains of ‘northern Togo and Benin’ as a source of Eastern Voltaic slaves supplied ‘almost entirely in the eighteenth century’. The catchment reached the Atacora and Tem interior of modern Togo.

Ghana (0.015). A small southeastern Ghana and Gold Coast component supplied the western Slave Coast ports in the eighteenth century. Law (2004, 75) documents Gold Coast immigrants and traders at Ouidah, including families from Cape Coast and Accra, and the Ewe homeland straddles the Ghana-Togo border (Law 1991, 14). The African Origins database records a persistent approximately 2.5 percent Ghana component in Bight of Benin recaptive samples from this coastal fringe.

Niger (0.004). A very small far-northern component reached the eighteenth-century coast through the Oyo-fed Porto Novo and Badagri channels. Manning (1982, 32) locates Hausa captives in ‘the far interior mostly at the time of the formation of the Sokoto Caliphate’, and Law (1977, 227) records ‘large numbers of Hausa slaves brought to Porto Novo by Oyo traders’; the deep-interior fringe of that Hausa stream extends into modern Niger. This far-northern Hausa and Sokoto-frontier interior carries little mass and a wide band.

Nigeria (0.822). The nineteenth century is decisively the Yoruba period. The collapse of the Oyo Empire in the 1820s-1830s and the ensuing Yoruba civil wars flooded the Bight of Benin ports with Yoruba captives; Verger (1976, 7) documents ‘the almost total absence of the Nago-Yoruba until the beginning of the nineteenth century, and their presence en masse around 1830’. The African Origins database records a 96.4 percent Nigeria component (Yoruba 85.9 percent plus Hausa, Edo, and related groups) of attributable Bight of Benin recaptive names for this period. Ajayi and Smith (1971, 123) establish that ‘the great expansion of the trade in the 1820s... [was] the result, not the cause, of the collapse of the Old Oyo empire.’ The catchment also reached the interior Yoruba and Bariba country of modern Nigeria.

Benin (0.125). The Aja and Fon of modern Benin (the ‘Gégé’ or Dahomeans) remained a real but minority component in the nineteenth century. Castelnau’s 1848 report in Verger (1976, 6) records the ‘Gégé or Dahomeans who form a powerful nation and are quite numerously represented in Bahia’, and Verger (1976, 7) confirms ‘the Dahomeans are present under the name Gégé from the very beginning of the register’. Manning’s Table 2.2 (1982, 31) puts Aja at 24.8 percent of 1801-1870 exports (124,300 of 501,400). The catchment held the continuing Aja and Mahi home interior of modern Benin.

Togo (0.033). The western Popo ports of modern Togo collapsed in relative importance as Lagos surged in the nineteenth century; Eltis in Falola and Childs (2004, 24) shows Popo departures falling to 7,200 in 1801-25, 11,900 in 1826-50, and 1,100 in 1851-65, while Lagos rose to 114,200 and 170,600. Manning (1982, 340) notes ‘modern Togo was not as heavily involved in slave trade as surrounding regions’. The catchment held the continuing western-Aja, Ewe, and Tem interior of modern Togo.

Ghana (0.02). The African Origins database fixes the nineteenth-century Ghana component directly: approximately 2.5 percent of attributable Bight of Benin origins in the recaptive samples, via the Akan group (approximately 2.2 percent, homeland Ghana) and the southeastern Ghana Ewe fringe of the western Slave Coast. This is the period where the African Origins anchor is most reliable, so the band is narrow on this coastal-fringe share.

Nigeria (0.988). The Bight of Biafra drew almost exclusively on the Cross River basin and its immediate hinterland in what is now southeastern Nigeria. Nwokeji (2010, 5) establishes that the Igbo and Ibibio ‘had long provided most of the export captives,’ and Latham (1973, 28) records that the earliest slaves sold at Old Calabar were ‘the unlucky members of lower Cross River tribes captured in war,’ the majority Igbo supplied by the Aro. The pre-1600 trade was confined to this narrow Cross River/Ibibio belt before the Aro had expanded inland, so Nigeria’s share is near-total. The catchment reached the deeper Igbo and Cross River interior the early coastal evidence could not name, which the historiography places overwhelmingly within modern Nigeria.

Cameroon (0.012). A small share belongs to what is now Cameroon, reflecting the presence of borderland peoples in the immediate hinterland of Old Calabar. Latham (1973, 5) identifies the Efut, who ‘migrated from the Cameroons,’ among the peoples immediately north of Old Calabar, alongside the Qua (an Ekoi group). The Cameroon trade had not yet developed at this date (Nwokeji 2010, 156, fn. 15), so the share is minimal but non-zero, capturing the Cameroon-margin element of the early Cross River catchment.

Nigeria (0.978). The seventeenth century saw Old Calabar consolidate as the region’s principal port, still drawing overwhelmingly on Igbo and Ibibio captives from the Cross River basin within modern Nigeria. Nwokeji (2010, 34) describes ‘a concentration of Aro activities in neighboring Ibibioland’ in 1651-75, ‘at a time when the Aro had not yet expanded westward and northwestward, and when Old Calabar was still Biafra’s principal port.’ Latham (1973, 28) confirms the supply ran on Igbo captives through the Aro, with the main slave market at Itu. The catchment reached the deeper Igbo/Cross River interior the coastal records could not name; with the Aro not yet expanded inland the catchment remained tightly Nigeria-weighted.

Cameroon (0.022). A modestly larger Cameroonian share applies for the seventeenth century, reflecting the borderland position of Cameroon-origin peoples (the Efut and Ekoi groups; Latham 1973, 5) in the Old Calabar hinterland. Nwokeji (2010, 156, fn. 15) notes the Cameroons trade had not yet come into its own, so the figure remains small, representing the Cameroon-margin component of the supply zone.

Nigeria (0.947). The eighteenth century was the peak of the Bight of Biafra trade, driven by the expansion of the Aro merchant network into central Igboland and the rise of Bonny over Old Calabar. Nwokeji (2010, 1) dates the ‘huge increase in the numbers of captives leaving the Bight of Biafra after 1740’ to this expansion, and (2010, 51) records that ‘Bonny received its captive supplies almost exclusively from central Igboland.’ The Igbo were ‘probably the largest single African group arriving in North America and several Caribbean destinations for much of the eighteenth century’ (2010, xiii). The catchment reached the deeper central-Igbo and Cross River interior opened by the Aro; the great bulk of the eighteenth-century catchment lay within modern Nigeria.

Cameroon (0.053). The Cameroonian share grows to roughly 0.05 in the eighteenth century as the Aro network extended eastward and supply routes from the Cameroon Highlands began to develop. Nwokeji (2010, 156, fn. 15) cautions that ‘the proportion of Cameroonians among captives leaving Old Calabar in the eighteenth century would have been much smaller than is suggested by the nineteenth-century sample... the Cameroons slave trade did not come into its own until after the mid-eighteenth century.’ This reflects the emerging Cameroon-margin supply, with the upper band left wide because the eighteenth-century Cameroonian proportion is poorly constrained.

Nigeria (0.874). Nigeria retains a large majority share in the nineteenth-century trade, though it declines as the Cameroon Highlands supply matures. The Igbo and Ibibio of southeastern Nigeria remained the principal source: Latham (1973, 29) records liberated Africans from Old Calabar coming ‘from Ibibioland, and the tribes of the upper Cross River, such as the Akunakuna, Yako, and Ekoi,’ all within modern Nigeria, while Bonny still drew ‘almost exclusively from central Igboland’ (Nwokeji 2010, 51). The direct nineteenth-century recaptive-names evidence puts Nigeria at 86.8% of attributable origins (Igbo alone 65.6%, n=8148). The Igbo/Cross River interior carries the Islamic/Multiethnic catch-all names that have no single homeland, in the same attributable ratio (~0.886:0.114 Nigeria to Cameroon).

Cameroon (0.112). The Cameroonian share rises substantially in the nineteenth century as the Cameroon Highlands supply route through Old Calabar became well established. Nwokeji (2010, 156, fn. 15) cites a recaptive database showing ‘no less than 10 percent of captives leaving Old Calabar between 1822 and 1837 came from western Cameroon,’ and Latham (1973, 29) records that as late as 1888 most slaves at Old Calabar came from ‘the Bamenda Grassfields of the Cameroons Highlands.’ The direct names evidence puts Cameroon at 11.1% of attributable origins (Tikar, Duala, Bamum, Akoose homelands all CMR). The share is routed in the attributable NGA:CMR ratio, with the upper band raised to reflect the genuine uncertainty in the Grassfields share.

Democratic Republic of the Congo (0.014). A small share belongs to the modern Democratic Republic of Congo, the outermost reach of nineteenth-century supply networks tapping the Congo borderlands at the southern edge of the region’s coast. The direct recaptive-names evidence puts this origin at 1.4% of attributable mass. The share is held at its vetted level, since the unattributable catch-all tracks the Nigeria/Cameroon core of the catchment rather than the Congo margin; the band stays narrow.

Angola (0.54). The Kongo kingdom’s core districts whose enslaved population fed Sao Tome lay substantially in modern northern Angola; by 1526 the kingdom exported two to three thousand slaves a year with its own subjects as victims, and from 1576 Luanda opened the Ndongo/Kimbundu plateau (central Angola) as a separate node (Iliffe 2007, 134; Lovejoy 2000, 41; Domingues da Silva 2017, 85). On top of the vetted coastal share, the 16th-c catchment was the Kongo-kingdom/Sao Tome/Ndongo trade whose larger population base lay on the Angolan side of the cross-border Kongo zone, so Angola is the modal origin here; the band is very wide because the Angola-DRC split of the Kongo heartland is undocumented for this period.

Democratic Republic of the Congo (0.31). The lower-Congo/Kongo-Central portion of the Kongo kingdom and the Malebo Pool markets (modern Kinshasa) are in modern DRC, and Vansina (1990, 201) is explicit that the inner basin, not the coastal lands, was first and most deeply affected by the slave trade, with Tio captives from north of the Congo River also supplying Kongo (Lovejoy 2000, 42). The inner-basin/Malebo-Pool share of the 16th-c Kongo trade is credited here; the band is very wide (0.12-0.55) because the Pool-versus-Kongo-heartland split is undocumented and the modern Angola-DRC border bisects the Kongo cultural zone.

Republic of the Congo (0.12). The Malebo Pool’s north and west bank, the lower-Kwango relay at Okango, and the Vili/Loango coastal zone fall in modern Congo-Brazzaville (Vansina 1990, 201). The COG-side Pool relays are credited here, but the Loango coast exported only a trickle before 1600, so this stays a minor early share.

Gabon (0.03). Gabon was negligible before 1600, a small importer rather than a source until the seventeenth century (Vansina 1990, 205). Only a token sliver is carried here.

Angola (0.64). The Kimbundu of the Luanda plateau became the earliest and most direct victims after Luanda’s foundation (1576), with Kasanje on the Kwango (modern Angola) emerging in the 1630s as the great broker funnelling Nganguela, Songo, Shinje and Ovimbundu captives to Luanda (Cambridge History of Africa Vol.4, 350-355; Domingues da Silva 2017, 85). Angola is the modal origin because the Kimbundu/Ndongo wars made the near-coastal Luanda plateau the dominant 17th-c source; the band is very wide because the interior-catchment boundary with the DRC-side is imprecise.

Democratic Republic of the Congo (0.27). The continuing Kongo-kingdom zone (lower Congo and Kongo-Central, modern DRC) and the Malebo Pool markets remained major suppliers to the Luanda network, with the inner-basin canoe trade reaching Lake Mai Ndombe and the Alima confluence before 1612 (Vansina 1990, 202). The DRC-side basin and Pool share is credited here; the band is very wide (0.10-0.52) to span the cross-border ambiguity of the Kongo zone.

Republic of the Congo (0.08). The Loango and Vili coast (modern Congo-Brazzaville) was still a minor slaver before the 1660s, exporting only about 300 slaves per year between 1630 and 1650 (Vansina 1990, 202). The COG-side relays are credited here as a small share.

Gabon (0.01). The Gabon Estuary remained a small importer rather than a source in the seventeenth century (Vansina 1990, 205); only a token share is carried here.

Angola (0.53). The Luanda-Benguela axis tapped the Kimbundu interior via Kasanje and drew in the Ovimbundu of the Benguela highlands as Umbundu captives grew through the 1700s (Cambridge History of Africa Vol.4, 350; Domingues da Silva 2017, 85). The trade is shared between Angola and the DRC-side because the simultaneous Loango boom shifted much of the trade to the Congo basin; Angola remains the plurality and the band is very wide because the Loango-versus-Luanda balance is imprecisely documented.

Democratic Republic of the Congo (0.29). This is the period with the largest DRC share: the dominant 18th-c Loango coast drew slaves from wide areas of the central Congo basin via the Bobangi, and the Kwango-Yaka-Suku-Pende zone of the Lunda westward expansion straddles the DRC (Cambridge History of Africa Vol.4, 348-351, 374; Vansina 1990, 204). The basin/Bobangi share is credited here; the band is very wide (0.14-0.52) to carry the cross-border Kongo/Kwango ambiguity.

Republic of the Congo (0.13). The Loango, Malemba and Cabinda ports and their Vili/Kunyi/Teke hinterland (modern Congo-Brazzaville) were the dominant 18th-c outlet, shipping 10,000+ slaves a year, probably more than Angola (Cambridge History of Africa Vol.4, 347-348; Vansina 1990, 204). A genuine share for the COG-side coast and Teke plateau is credited here.

Gabon (0.04). In the eighteenth century slave trading reached the Gabon Estuary and Cape Lopez and the Loango network tapped supplies toward the upper Ogooue (Vansina 1990, 205; Cambridge History of Africa Vol.4, 346). A small share reflects this furthest-north 18th-c frontier.

Zambia (0.01). The Lunda south-eastern expansion drew in the Ndembu of north-western Zambia, whose chief Kanongesha sent slaves north as tribute (Cambridge History of Africa Vol.4, 375-376). Lunda’s external trade was small (under ~3,000/yr), so this stays a token share.

Angola (0.63). Domingues da Silva’s (2017, 85-86) direct 19th-c origins data (1831-1855) show the catchment retracted to within ~400 km of the coast, overwhelmingly Kimbundu (central Angola), Umbundu (highlands) and Angolan-side Kikongo, all west of the Kwango; Table A.1 (total 767,215) is dominated by Kimbundu groups. Angola is the modal origin on the broader Angolan source base, but the band is very wide (0.28-0.80) because the African Origins database attributes a high share of the same recaptive cohort to the DRC, so an adversarial reader can land near the DdS Angola-dominant reading or much lower.

Democratic Republic of the Congo (0.24). This cell carries the AGO/ZAR cross-border dispute: the African Origins database attributes ~89% of WCA recaptives to the DRC (Luba 31%, Kongo 29%, Kanyok 11%, all mapped 100% to ZAR), while Domingues da Silva (2017, 169) reconstructs the same cohort as Angola-dominant and finds no slaving frontier moving deep into the interior. The DRC share is real (cross-border Kikongo of Kongo-Central, the Yaka/Suku/Pende of the Kwango, the Cokwe-Ruund Kasai tail). The DRC is credited a substantial share alongside Angola; the band is the widest in the matrix (0.08-0.72) so it spans the DdS Angola-dominant reconstruction at the low end and the AO DRC-dominant reading at the high end.

Republic of the Congo (0.06). The northern Loango/Vili/Kunyi/Lumbu/Yombe coastal zone (modern Congo-Brazzaville) continued to embark captives, with Cabinda and Molembo as major illegal-era ports (Domingues da Silva 2017, 32, 85). The COG share is smaller than in the 1700s because the Loango interior reach had collapsed; only a sliver is credited here, the African Origins database putting COG near 2%.

Mozambique (0.05). A 19th-century-only cell: traders searched for other South Atlantic sources such as Mozambique to keep West Central Africa volumes stable (Domingues da Silva 2017, 168), and the African Origins database records Mozambique near 5.6% among captives associated with this embarkation region via Swahili and East African names re-shipped through the southern system. A genuine Southeast African substitution flow, left unchanged.

Gabon (0.02). A token 19th-c share for the northern Kota and Njebi fringe of Domingues da Silva’s (2017, 85) catchment, the Gabon-estuary trade having declined from its 18th-c peak; the African Origins database puts Gabon near 0.6%.

Mozambique (0.55). Before c.1600 the Atlantic-bound trade from this coast was effectively nil; gold and ivory, not slaves, were the staples of the Sofala-Kilwa economy (Alpers 1975, 44), and what enslavement occurred fed the Portuguese garrison economy from the immediately adjacent coastal hinterland, the Makua of Macuana behind Mozambique Island (Alpers 1975, 82-83). The lower-Zambezi and Makua-adjacent hinterland that the coastal records do not resolve to a named people is credited here. Because the traded weight is effectively zero this is a prior with a wide band.

Malawi (0.32). The Maravi confederacy (Chewa, Nyanja, Nyasa-speaking peoples), whose heartland is modern Malawi, was expanding through the sixteenth century and dominated the interior of the Mozambique hinterland; the Kalonga ‘thoroughly dominated’ the other interior peoples (Alpers 1975, 83). Any captives drawn from beyond the immediate coast in this period would have passed through Maravi-controlled territory, which the coastal evidence cannot name, so the deep-interior catchment is credited here. Negligible volume, so the band is wide.

Zimbabwe (0.08). The Zimbabwean plateau (Mutapa/Shona) supplied the gold that dominated the pre-1600 Sofala-Kilwa economy; enslavement was incidental to gold and ivory production (Alpers 1975, 44). The far-interior plateau periphery that the coastal records cannot resolve is credited here. Weight is minimal and the band is wide.

Zambia (0.05). A small share is placed in the eastern Maravi and Nsenga zone of modern eastern Zambia, which bordered the lower-Zambezi trade axis and lay within the Maravi tributary network (Alpers 1975, 83). No direct provenance evidence exists for this period, so the share is a geographic prior with a wide band.

Mozambique (0.55). Through the seventeenth century the trade remained gold- and ivory-dominated; the small slave component came from the Makua coastal hinterland of Macuana and Uticulo (Alpers 1975, 82-83) and the lower-Zambezi prazo zone, whose Sena and Tonga labour the Portuguese estates drew on (Beachey 1976, 13-14). The lower-Zambezi prazo hinterland the coastal records do not resolve to a named group is credited here. Wide band given the absence of direct provenance data.

Malawi (0.34). The Maravi confederacy, whose heartland is modern Malawi, dominated the interior trade of the Mozambique hinterland throughout the seventeenth century; the interior peoples ‘were thoroughly dominated by the Kalonga’ (Alpers 1975, 83), and the Yao had not yet displaced the Maravi as the principal carriers. The Maravi were the channel through which any interior captives reached the coast, which the coastal evidence cannot name, so the deep-interior catchment is credited here. Wide band.

Zambia (0.08). The eastern Maravi and Nsenga zone of modern eastern Zambia bordered the lower-Zambezi trade axis (Alpers 1975, 83). The Maravi-tributary frontier west of the lower Zambezi is credited here; no direct evidence places Zambian peoples among 1600s coastal slaves, so the share rests on geographic priors with a wide band.

Zimbabwe (0.03). A small share is placed on the Zimbabwe-plateau and southern-Zambezi periphery, which lay at the far inland edge of the lower-Zambezi trade axis (Alpers 1975, 83; Beachey 1976, 13). Enslavement remained incidental to the gold and ivory economy; weight is minimal and the band is wide.

Mozambique (0.51). The eighteenth-century trade grew suddenly after 1785 on French Mascarene demand concentrated at Mozambique (Alpers 1975, 185); the Makua of Macuana, in the coastal hinterland behind Mozambique Island, were most exposed to the intensified raiding, and the coastal hinterland ‘remained the primary region from which slaves were taken for exportation’ (Alpers 1975, 242). The Makua-adjacent hinterland the records do not resolve to a named people is credited here.

Malawi (0.27). By the eighteenth century the Yao had emerged as the principal long-distance carriers, drawing captives from the Maravi-speaking peoples of the Lake Nyasa region, whose heartland and much of the Yao homeland are in modern Malawi (Alpers 1975, 15, 84). The Yao-Maravi network drew its captives from the Lake Nyasa interior that the coastal evidence cannot name, so the deep-interior catchment is credited here. Wide band.

Tanzania (0.10). The Yao homeland extends into southeastern Tanzania north of the Ruvuma, and Kilwa was the secondary eighteenth-century export outlet whose nearer hinterland contributed captives (Alpers 1975, 15). The SE Tanzanian Yao/Makonde zone and the Kilwa hinterland north of the Ruvuma that the southern (Mozambique-centred) records do not capture are credited here. Wide band.

Zambia (0.09). From the 1790s the Bisa carried slaves, ivory, and copper from the eastern Lunda state of Kazembe in the Luapula valley of modern eastern Zambia to the Zambezi Portuguese, supplying the Yao (Alpers 1975, 122, 178; Birmingham 1976, 243). This Bisa/Kazembe deep-interior catchment, which the coastal evidence registers only as part of an undifferentiated interior flow, is credited here. Wide band.

Madagascar (0.03). The sudden post-1785 growth of the Southeast African trade was ‘almost wholly the work of the French’ supplying the Mascarene Islands (Alpers 1975, 185), and the French Indian-Ocean demand drew on Madagascar alongside the Mozambique coast. Madagascar is credited here as a genuine origin entering the system in the French-demand era. Madagascar carried no Atlantic-side trade before this period; weight is small and the band is wide.

Mozambique (0.52). The nineteenth century carried most of the region’s Atlantic embarkations, overwhelmingly from Mozambique ports, with the Makua coastal hinterland the primary origin: ‘most of the slaves who were sent to Brazil came from this part of Macuana’ (Alpers 1975, 220), and the coastal hinterland ‘remained the primary region from which slaves were taken’ (Alpers 1975, 242). The Makua-adjacent hinterland the port records do not resolve to a named people is credited here.

Malawi (0.23). As the coastal hinterland was depopulated, the frontier moved inland to the Lake Nyasa region; Rigby’s 1860 report put 15,000 of 19,000 Zanzibar-bound captives from ‘the neighbourhood of the great lake of Nyassa’, mainly Nyasa, Yao and Manganja (Alpers 1975, 239), the same Lake Nyasa catchment that fed Quelimane and Ibo to the Atlantic (Alpers 1975, 231, 252). The deepest 19th-c caravan catchment was the Lake Nyasa interior of modern Malawi, credited here. Wide band.

Tanzania (0.11). The Yao homeland, the Makonde, Ngindo and Mwera of southeastern Tanzania, and the Nyamwezi caravan network supplied captives originating north of the Ruvuma to Kilwa and thence via Ibo and Angoche to the Atlantic; von der Decken found most Kilwa slaves ‘were Bisa and Yao’ (Alpers 1975, 240) and Sheriff (1987, 162) names the Makonde, Ngindo, Nyasa and Yao in the Kilwa orbit. This SE Tanzanian and Kilwa-hinterland flow the southern records do not capture is credited here. Wide band.

Zambia (0.10). As the frontier deepened it reached west of Lake Nyasa to the Bisa, Nsenga, Bemba and Chikunda of modern Zambia; von der Decken named the Bisa among the principal Kilwa origins (Alpers 1975, 240), and Abdallah lists ‘Nyasa, Bisa, Senga... and Chikunda’ as sources for the Yao trade (Alpers 1975, 241). This deepest Zambian caravan catchment, which the coastal evidence registers only as part of an undifferentiated interior flow, is credited here. Wide band.

Zimbabwe (0.03). The Zambezi prazo and Chikunda raiding zone and the periphery of the Shona plateau south of the Zambezi contributed a small share to the Sena-Tete-fed Quelimane trade; Beachey (1976, 13) documents raiding pressure across the Zimbabwe-Zambia-Mozambique borderlands, and the Chikunda are listed among Yao slave sources (Alpers 1975, 241). The share is retained unchanged; the deep-interior catchment is the Lake Nyasa and Tanzania zones rather than the small southern-plateau periphery. Wide band.

Madagascar (0.01). Madagascar continued as an Indian-Ocean source feeding the same French and South-Atlantic demand that linked the Southeast African coast to the Atlantic trade in the suppression era (Alpers 1975, 210, on Brazil-directed Mozambique-coast demand; the French-island and re-flagged South-Atlantic flows). Madagascar is credited here as a genuine origin in the system. Weight is small and the band is wide.