法国奴隶贸易概论

The French Slave Trade: An Overview

David Geggus
Jan,2001,Wiliam & Mary Quarterly

IN comparative studies of race and slavery, France's colonies often appear in an intermediate position between Iberian and Anglo-Saxon extremes. The institutions of the French colonies exhibited in attenuated form the Catholic absolutism of the Iberians, whereas the colonies' social structures closely resembled that of the North European colonies. Scholars who discuss colonial slave laws, the frequency with which slaves were freed, or racial integration often situate French slave society somewhere between northern and southern European polar types. In the history of the Atlantic slave trade, too, we find the French occupying an in-between position. Whether the subject is the length of time the French actively participated in the slave trade (as importers or as traders), their share of its total volume, or the pace of their withdrawal, the mortality suffered by the captives they carried or the balance between males and females among those they transported, French activity tended to have that same intermediate character. In the history of the slave trade, the Iberians are often grouped at one end of the spectrum and the Northern Europeans at the other, but, invariably, the French appear somewhere near the middle.
1
This article begins by addressing the contribution of the Du Bois Institute's dataset to existing knowledge of the French slave trade. It then examines the general contours of the trade, notably its temporal and spatial distribution, its mortality rates, and the age and sex composition of the people transported. The second half of the article focuses on France's Caribbean colonies, which were the principal destination of 90 percent of French slave ships. It identifies a pattern of variations among different colonies and regions in the ethnic, age, and sex composition of their African migrants. This pattern is attributed to factors that include planter preferences determined by crop type, differing chronologies of colonial expansion, and the commercial aspects of markets that were valued by merchants. Finally, the cultural implications of this pattern for different areas are briefly considered.
2
The Du Bois Institute dataset lists 4,033 slaving voyages by French-registered ships destined for the Americas that sailed between 1669 and 1864. It excludes a large number of voyages whose final destination lay in the Indian Ocean. The systematic efforts of Jean Mettas and Serge Daget in the 1970s and 1980s to inventory the post-1706 French slave trade left the compilers of the Du Bois Institute database less scope for extending knowledge of France's contribution to the Atlantic slave trade than that of most other participants.1 In a 1990 article, David Eltis argued that the Mettas-Daget Répertoire of eighteenth-century slaving expeditions was at least 90 percent complete.2 To judge from the sources it lists, the new database adds to this corpus primarily a dozen new cases found in Lloyd's Lists, five others collected by Robert Stein, and French voyages to Spanish America drawn from the research of Herbert Klein and Elena Studer. In terms of new voyages, the Du Bois Institute material contributes most to knowledge of French activities during the "illegal" period that began in 1814. From British archives alone, Eltis, Stephen Behrendt, and David Richardson have added at least twenty expeditions that did not appear in Daget's study of "la traite illégale."3
3
For the little-studied seventeenth century, some data recently uncovered by Clarence Munford and others are combined with material from older works by Elizabeth Donnan, Abdoulaye Ly, and John Barbot. The compilers note, however, "much of the seventeenth century French traffic is missing."4 4 A large part of France's slave trading was then clandestine, conducted by interlopers challenging royal monopoly companies. The other period for which coverage is least complete is that of the Consulate (1799–1804). This neglect is both suggested and partly remedied by Eric Saugera's recent publications, which uncover a number of hitherto unknown expeditions of the nineteenth, and also the seventeenth, centuries.5 Some extra scraps of information regarding several late eighteenth-century voyages (slaves disembarked or sold, African places of trade, revolts, owners and captains, and the like) can be found in the correspondence of the Cap Français merchant house Foäche, Morange et Hardivilliers, but it reveals no new voyages.6 Fortunately, those periods for which the database provides the least coverage are those when the French slave trade was least active. We have, then, a catalogue of slave-trading expeditions that overall seems remarkably complete.
4
Some of the database's projected figures need further refinement.7 Even so, in those instances where we can compare projections incorporated in the database to summary statistics found in official correspondence, the match up is extremely close. The imputed number of Africans introduced into Martinique from January 1714 to August 1721 (11,376) suggests a 2.4 percent undercount, compared to the 11,651 reported by the Intendant.8 The database's imputed number of arrivals in St. Domingue during the period 1785–1789 is, for individual years, slightly lower or higher than the figures reported by colonial officials, but in the aggregate it differs very little (137,298, as compared to 136,857).9
5
The French voyage data are also in certain respects more extensive than those for other sections of the Du Bois Institute database. The ship's port of departure is known in 96 percent of cases (as opposed to 75 percent for the total sample). The slave's place of embarkation and of disembarkation is known for 75 percent and 78 percent of French voyages (compared to 57 percent and 75 percent overall). The numbers of Africans embarked and disembarked are known in a proportion of cases (28 percent, 55 percent) that is average for the dataset. Although the proportion of French expeditions reporting deaths among captives is below average (16 percent compared to 22 percent), mortality rates can be imputed for 27 percent of cases.
6
Almost 3,950 expeditions are listed that, altogether, embarked from Africa an estimated 1.25 million captives. Given current assumptions about the volume of the Atlantic slave trade and a reasonable allowance for unrecorded voyages, French carriers must have been responsible for about one-eighth of the total traffic. Of the Africans transported, two-thirds were males, perhaps 27 percent were children, and an estimated 13 percent died in transit—proportions typical of the trade in the period covered.10 They left Africa in groups that averaged close to 320 per ship. If decades of slave trade scholarship have rendered such average figures banal, the surreal quality that attaches to some of the extreme cases can still perhaps seize our attention, as with the Antoinette that sailed into Rio de Janeiro with 900 captives and ten crew members; or the voyage on which 408 Africans died; the Middle Passage that lasted 310 days; or the "cargo" of which 92 percent were children.
7
Needless to say, the vast majority of French voyages originated in France. Few departures from the Americas or Africa have been recorded for the period before the nineteenth century. In the years 1801 to 1825, 12 percent of expeditions set out from the Caribbean, and in the following quarter-century, as the French trade withered, that proportion rose to 43 percent, with another 7 percent of French vessels leaving from Brazil. Nantes was far and away the most important port of departure (see Map I). It fitted out no fewer than 1,708 slaving expeditions that account for 44 percent of all French expeditions of known origin. This amount is more than the combined share of the five next most important ports: Le Havre, La Rochelle, Bordeaux, St. Malo, and Honfleur. Even so, nearly twenty French ports participated in the trade, and so obscure a participant as the tiny port of Vannes fitted out at least thirteen voyages.
8
In Africa, the port most frequented by French vessels was Whydah on the Bight of Benin. It appears as the point of embarkation for twice as many captives as those embarked from Malembo in West Central Africa, the second most prominent port. As the third and fourth most important shipping points (Cabinda and Loango) were similarly on the Loango coast, West Central Africa formed the main source of captives for the French slave trade. It attracted one-third more ships than did the Bight of Benin and, because it also attracted the largest French vessels, supplied 50 percent more slaves (see Table I). This ordering of ports is subject to some uncertainty, because the listed site of trade of many expeditions was only a vague "Angole" (applied to all West Central Africa) or "Côte d' Or" (the equivalent in French usage of "Guinea"). The widespread usage of "Angole" raises the issue of French penetration of the Portuguese possessions south of the Loango coast (highlighted by Joseph C. Miller) and therefore of the identity of the slaves whom French colonists called "Congo."11 It is certain that Angolan ports are very rarely mentioned, yet an extensive clandestine trade beyond those ports' limits could well be cloaked beneath usage of the regional term (see Map II).12 West Central Africa does not appear at all in the seventeenth-century record, when French captains purchased most of their slaves in Senegambia and the Bight of Benin. In the first quarter of the eighteenth century, West Central Africa became the second most important source of captives for the French, although it remained a long way behind the Bight of Benin. Only during the boom years of the last half of the eighteenth century was the Loango coast the French trade's leading supplier. During the nineteenth century it played a very minor role, while Biafra and Sierra Leone, hitherto unimportant regions, became the foci of French activity (see Table II).
9
The exact sex and age ratios of the Africans transported is difficult to pin down. The mean proportions per vessel of males and children (generally reckoned to be under age fifteen) recorded at the start of the Middle Passage are respectively 67.6 percent and 19.0 percent, but they derive from samples of only 96 and 94 expeditions. The proportions recorded at the point of disembarkation derive from much larger samples (699 and 693 voyages) and show proportions of 62.9 percent and 27.8 percent. It is uncertain how much of the difference between the two sets is the product of sampling and how much reflects differential mortality during the Atlantic crossing. The available data suggest that, on the average ship, the proportion of males fell 1.8 percentage points and the proportion of children rose by 0.5 percentage points, but these figures derive from samples of just 60 and 58 voyages.
10
The number of captives loaded per vessel rose from an average of 261 in the seventeenth century to a plateau of 325 in the period 1726–1775, before peaking at 340 in the last quarter of the eighteenth century. After 1800, the average number of Africans embarked fell to 276.13 This decline was owing to increased concentration on the Upper Guinea coast, the region that attracted the smallest vessels, and a shift away from Central Africa, which was served by the largest. In the Bight of Biafra, the main trading region of the illegal period, average cargo size in fact increased between the late eighteenth and nineteenth centuries. French loading rates (slaves per ship) in the different African markets broadly mirrored those of other nations, with the exception that French vessels in the trade to Southeast Africa carried notably fewer captives than was normal for that region.
11
On the Upper Guinea coast, where the smallest vessels traded, the supply of slaves was limited and crossing times to the Americas were also shortest. The average duration of the Middle Passage fell sharply between the eighteenth and nineteenth centuries. In the first half of the eighteenth century, however, average voyage times from most regions rose, and from the Gold Coast, Bight of Benin, and West Central Africa, this rise continued into the third quarter of the century. The reasons for this pattern are not at all obvious, but it parallels a similar increase and decline in loading times on the African coast noted by Eltis and Richardson.14 Crossing times for the seventeenth century are unavailable.
12
Duration of the Middle Passage was a significant but not major determinant of mortality rates and their variations between different trading regions.15 One quarter of the captives transported from Indian Ocean ports did not survive the voyage, whereas about one-tenth died on the shorter routes from West Central Africa and Upper Guinea. The exceptionally long transit from the Bight of Benin and Gold Coast did not result in death rates greatly above the average. Moreover, the Bight of Biafra stands out as a high-mortality region for reasons that go beyond those of voyage time.16 Neither the region's dominant modes of capture (judicial enslavement and kidnaping) nor its high proportions of females and children can have been responsible. The latter proportions fluctuated independently of mortality rates in the Biafran trade, and as seen above, males and adults were more likely to die in the crossing. This contrast between Biafra, on the one hand, and the Gold and Slave Coasts, on the other, is reminiscent of these captives reputations among French and British colonists in the Caribbean: people from the Gold and Slave Coasts were considered to be extremely hardy and robust, and at least the Ibibio of Biafra were referred to as sickly.17 Although the relative nutritional value of yams and maize is controversial, it is tempting to agree with the eighteenth-century commentators who suggested that differences in local diet underpinned this contrast.18
13
The regional contrasts and the average mortality rate revealed in the French data are similar to those found for the rest of the Atlantic slave trade.19 The extent of regional variation was greater, however, with death rates on the voyage from Southeast Africa being considerably higher. Although the long-term trend in mortality rates was downward, the progression was far from smooth. As Table III shows, from one quarter century to another rates rose as often as they fell. Fluctuations in the matrix of regional sources of supply were in part responsible, but average voyage times per quarter-century period were not implicated.20 A long-term, if irregular, shift away from low-mortality regions toward those associated with highest mortality perhaps blunted the impact of improving health care and business organization by traders (see Table II).
14
Behrendt has shown that death rates per voyage were higher for crew members than for slaves in the eighteenth century, although mortality per day was much higher for slaves, who spent a shorter time on board. Crew mortality was concentrated during the African phase of the voyage, although one in four crew deaths occurred in the Americas.21 Crew mortality rates declined steadily throughout the eighteenth century, unlike those of slaves.22 Behrendt also notes a striking pattern whereby French crews suffered consistently lower death rates than British crews. He suggests that longer loading times, less attention to hygiene, and large proportions of inexperienced (therefore non-immune) landsmen in the British trade may explain this difference. The greater concentrations of French ships in the region of lowest mortality, West Central Africa, and of British ships in the deadly Bight of Biafra, doubtless also contributed. Slave mortality and crew mortality were related, as Klein observed twenty years ago, even though the deaths of sailors and their captives were concentrated in different phases of the slaving voyage.23
15
The main place of slave disembarkation is known for 3,248 voyages. Africa and mainland Spanish America each account for about 1 percent of these cases; Brazil and North America (chiefly Louisiana) for a little less. The non-French Caribbean accounts for another 6.4 percent of debarkations, with Cuba (137 expeditions) being by far the most prominent destination.24 The principal destination, however, of nine out of ten French carriers was the French Caribbean—St. Domingue, Martinique, Guadeloupe, Guyane, and (until 1763) Grenada (see Maps IV, VI).
16
St. Domingue was by a huge margin the major recipient of captives carried in French vessels, even though it largely ceased to import slaves after slavery was abolished there in 1793.25 At different times the world's leading exporter of indigo, sugar, and coffee, St. Domingue accounted for more than three quarters of French expeditions to French colonies and for at least 80 percent of the slaves they sold there (Table IV). By 1715, it had overtaken Martinique as France's prime market for slaves. The one port of Cap Français, in northern St. Domingue, was the main destination of more than one-third of all French slaving voyages. In 1790, the peak year of the French slave trade, French ships landed at least 40,000 Africans in St. Domingue and more than 19,000 at Cap Français—a record for any American port to that date.26
17
The figures discussed here, it should be stressed, do not represent total imports, as France's colonies imported more slaves than its merchants could supply. The high prices charged by French traders left all French Caribbean colonies open to a foreign interloping trade, whose dimensions, distribution, and growth have been extremely hard to gauge. Wartime brought further arrivals of foreign vessels, both those captured by French privateers and those protected by foreign military occupations.27 In 1783, moreover, the French government legalized slave imports by foreign ships into Martinique, Guadeloupe, and, temporarily, southern St. Domingue, because they were so neglected by French commerce.28
18
Several factors attracted slave ships toward St. Domingue and especially its north province, leaving the south Caribbean colonies short of labor. Chief among them were the size of the local market and the prospects of rapid sales and reloading. Also important were local prices, together with the amount of credit local planters expected when buying slaves and the availability of specie, which was needed to grease the wheels of commerce and to pay port taxes.29 These factors placed struggling Cayenne (Guyane) at one end of a scale whose other end was the prosperous port of Cap Français. Cayenne lacked cash, attractive commodities, and solvent clients. Cap Français had a large and diversified hinterland of long-established plantations and favored access to Spanish silver.30 In between were the Windward Isles (Martinique and Guadeloupe), whose relatively small, creolized plantations generated only limited demand. Guadeloupe had the particular disadvantage of being legally subordinated to the merchants of Martinique for much of the eighteenth century. Martinique was plagued by ants through the 1770s, and both Windward Isles suffered when their main product, semi-refined sugar, fell from favor in the 1780s.
19
It is especially interesting that this hierarchy of markets was clearly reflected, not just in the numbers of ship arrivals and the quantity of slaves sold in each, but also in the sex and age ratios of the African slaves they imported and, in a less straightforward manner, in the ethnic composition of those Africans. St. Domingue absorbed not only far more slaves than the others, but also more males, fewer children, and fewer Africans from Sierra Leone, Biafra, and the Windward Coast, those least appreciated by the colonists. Neglected Guyane and Guadeloupe, on the other hand, received easily the largest shares of captives from these regions, the smallest numbers of males, and generally speaking, more children (Tables IV, V).31 These contrasts did not result from the early closure of the trade to St. Domingue, for they were present throughout the eighteenth century.32 The high proportion of Biafran captives in Guadeloupe was merely accentuated by French traders increasing presence there after 1800.
20
Certain aspects of this pattern reappear in contrasts between the three provinces of St. Domingue, north, west, and south (see Tables VI, VII). The dynamic port Cap Français attracted the largest proportion of male slaves and of captives from West Central Africa and the smallest proportions from Sierra Leone, Biafra, and the Windward Coast. The west province, less favored by merchants, absorbed an intermediate proportion of slaves from the unpopular zones of embarkation, but not as many as the neglected south coast, where French slave ships were rare visitors. It is surprising the data do not indicate that children were differentially directed toward the least-favored markets, because there is no doubt slave merchants in Le Cap redirected incoming ships with numerous children as well as with unsold queues de cargaison (usually the sick, the elderly, and the young) to other ports in the west and south.33 This may be because age and sex ratios in the St. Domingue dataset are available for very few Africans sold in second or third ports of call, a role often played by Port au Prince, St. Marc, and Léogâne. In 1790, a third of the slave ships visiting those ports had been sent on from Cap Français; one in four vessels that anchored at Le Cap was sent on to other ports.
21
As described so far, the pattern of distribution of Africans in the French colonies resulted from planters' attitudes toward the gender and ethnicity of slaves and from commercial factors important to merchants. An additional factor was the temporal shifts in the location of French slave trading on the African coast. The numerical dominance of West Central Africans in St. Domingue partly reflects the shift southward of French slave traders in the later eighteenth century, a period when St. Domingue's economy boomed and the other French colonies generally stagnated, importing far fewer slaves.34 The very high proportion of males among captives transported from West Central Africa explains a good part of the regional variations in sex ratio in the French Caribbean (Table I). A strong "Congo" presence brought with it a highly imbalanced sex ratio. Conversely, importations from Sierra Leone, Biafra, and the Windward Coast are associated with high proportions of women and children.
22
A final, and very significant, factor that shaped the peopling of the French Caribbean was the spatial distribution of different crop types. As in Brazil, the distribution of West Central Africans was closely linked to the cultivation of coffee. St. Domingue dominated coffee production, which was concentrated in its north province. The proportion of "Congos" varied among St. Domingue's three provinces, but in each they were 50 percent more numerous on coffee plantations than on sugar estates.35 Sugar planters disliked buying them, although sugar planters generally exhibited a strong preference for male labor. This dislike was due to the West Central Africans below-average height, the ill-health they experienced on lowland plantations, and their traditional gender division of labor that gave the men relatively little agricultural experience. For the demanding work of cultivating cane and making sugar, planters preferred Africans from the Bight of Benin, whose men and women were regarded as robust, good agriculturalists, and capable of taking charge of their own provision grounds. West Central Africans, women, and children were reckoned to be suitable workers under the less-exhausting regime of coffee production. This suitability explains the paradox of why a relatively "unpopular" ethnic group was concentrated in the most selective markets.
23
Nevertheless, sugar planters had more opportunities to be selective in their purchases of slaves than did coffee planters. Coffee planters were not as well established as the owners of larger and older estates; therefore, they purchased in small quantities and with greater urgency to bring their new lands into cultivation. All the ethnic groups with negative reputations among colonists, including those from Biafra, generally show up with greater frequency on coffee than on sugar plantations.36 Less is currently known about indigo estates, but they seem to have had an "ethnic profile" somewhere between that of the other two plantation types, fittingly for a crop whose work regime can be considered intermediate in its severity.37 In deciding on a slave ship's destination in the colonies, merchants took into account the crop mix of different regions. All buyers thus confronted a regional supply shaped by the region's most prominent crop.
24
These factors, crop type, ethnicity, age, and gender meshed with the market factors concerning the number, wealth, and liquidity of local buyers and sellers. Local prices presumably played a significant role but this is not entirely clear. Some evidence suggests Africans purchased in Sierra Leone and Biafra cost less than those purchased elsewhere. The low prices offered for slaves in the Windward Isles were certainly given as a reason for avoiding those colonies by French merchants in the 1780s.38 However, neglected markets surely could have been forced to pay higher prices, as long as the (much cheaper) foreign interloping trade was kept in bounds. In St. Domingue, somewhat surprisingly, slave prices were lower in the selective market of Cap Français than in the colony's ports that were forced to accept the slaves it rejected. One might think it was the predominance of "Congos" in the north province that brought down average prices there. However, the average price per person was a little higher for ships coming from West Central Africa than from the Bight of Benin. Although, all things being equal, an Ewe-Fon slave sold for more than a Bakongo, the greater proportion of males among the latter gave them a higher aggregate price. Two things kept prices relatively low at Le Cap: the degree of competition between sellers seeking rapid turnaround in a large market and the less generous credit terms that were offered to buyers in the north. This probably eliminated any real regional difference in prices in the colony.39 Finally, it may be argued that the longer sailing time to St. Domingue than to the other French Caribbean colonies mandated its receiving the highest value cargoes, hence those with most men and fewest children. Certainly, in Africa itself, captives transported the longest distances exhibited the highest sex ratios.40 Richard Nelson Bean has argued that slave prices were higher in Jamaica than in Britain's east Caribbean colonies because of their relative distance from Africa. Eltis agrees and adds that for the same reason Virginia imported more male slaves than did Jamaica.41 In the French case, however, it seems that the extra sailing time between the east and west Caribbean was generally more than counterbalanced by the speedier turnaround time in Cap Français. It is true that, before 1730, slave ships tended to spend longer at Le Cap than in east Caribbean ports. By the 1740s, however, vessels were completing sales and reloading a full month more quickly in Cap Français than in Martinique. The contrast with Cayenne and Guadeloupe was apparently comparable, although the relevant data are sparse.42 This contrast is despite Le Cap's ability to attract the largest vessels in the French trade, which carried one-half to two-thirds more captives than those trading to the latter two colonies. Bearing in mind the port's other advantages, it is difficult to conclude that the week's extra sailing time from Africa had a critical bearing on the composition of cargoes. The costs of a longer voyage must have been more than recouped in reduced expenses and much higher revenues per day in port.
25
Only a mixture of factors that expressed the interests of merchants and planters, therefore, can explain the Caribbean distribution pattern of the Africans they purchased. Guadeloupe, the colony of a colony, and the isolated south coast of St. Domingue surely did not seek their reputation as markets for slaves unwanted elsewhere. But there were perhaps two sides to their heavy purchases of people from the Upper Guinea coast and Biafra, who combined a reputation as poor workers with the highest proportions of women and children of any exporting zone. Certainly there was merchant exploitation of under-supplied markets, but slaveowners in those markets may also have been more ready to purchase women and children to promote population growth.
26
The more balanced sex ratios of Africans sold in the Windward Isles after 1760 presumably helped accelerate the creolization of their slave populations, as did, in a cultural sense, the larger proportions of children, easier to assimilate, who were purchased there in the same period. Nicole Vanony-Frisch's study of Guadeloupe notarial papers of the 1770s and 1780s suggests that between 60 and 75 percent of the island's slaves were then locally born creoles and nearly one-fifth were of mixed racial descent. These proportions were assuredly higher in Martinique, whose slave population was longer established.43 In St. Domingue, on the other hand, the growth of a creole society could make only little progress when, between the late 1760s and late 1780s, annual arrivals of Africans doubled from 13,600 to 27,500, and the proportions of women and children among them shrank. Then came the massive influx of 1790, an abrupt 50 percent increase in Africans that was concentrated in the north and was followed by the uprising there of August 1791.44 It would be unwise, however, to link the uprising too closely to this upsurge in importations. The insurrection was mainly led by Creole slaves; it broke out in the region of St. Domingue where Creole slaves were most numerous, and those areas where Africans were most numerous were generally the slowest to join the insurrection.45
27
In this second half of the eighteenth century, the presence of West Central Africans became increasingly prominent in St. Domingue. Although outnumbered by Creole slaves on the sugar estates of the plains, they constituted in all regions the largest minority among the Africans, and on the coffee plantations of the mountains they formed a clear majority of the adult slaves. Those who study St. Domingue society are lucky to have a large body of plantation documents that is richly detailed as to slaves ethnic origins. The plantation documents perfectly complement the slave trade data, providing a finer focus on slaves' origins. As regards these West Central Bantu peoples, generally blanketed under term "Congo," however,, plantation documents rarely offer further clues as to ethnic identity. This problem parallels the difficulty with the term "Angole" in the slave trade data.46 While acknowledging this disguised diversity, one might say that St. Domingue's coffee sector provides strong support for the new scholarly emphasis on the cultural homogeneity of American slave societies, whereas the sugar sector gives substantial weight to Philip Morgan's post-revisionist rejoinder about diversity.47 By the end of the colonial period, there were almost as many slaves on coffee plantations as on sugar estates; indigo workforces resembled those of the coffee sector more than the sugar workers.
28
This pattern raises the old question about why the dominant African influence on Haitian culture seems to have been that of the Ewe-Fon people, or Gbe speakers, of the Bight of Benin, whose vocabulary and deities are most evident in the lexicon and pantheon of Haitian voodoo.48 Into the 1970s and beyond, some scholars assumed these "Arada" slaves must have predominated in the colonial slave trade, though by the 1960s the pioneering work of Gabriel Debien had made clear this was not the case.49 If the Yoruba legacy in Cuba and northeast Brazil can be attributed to the salience of Yoruba captives in the last stages of the slave trade to those regions, why did the "Congo" of St. Domingue not leave a stronger imprint on Haitian culture? Were the colonists of the eighteenth century wrong to associate what they called "vaudou" with "Arada" slaves and especially the west province of St. Domingue? Given that captives from the Bight of Benin made up barely one-fourth of the colony's imported slaves, though nearly one-half of the Africans taken to Martinique, one might have expected the latter island to have more association with voodoo.
29
Several factors may have impeded the preservation of Bantu traits in the French colonies. The age and sex ratios of the West Central Africans were less favorable to the preservation of native culture than those of peoples from the Bight of Benin (see Table I). The latter's numerical predominance early in St. Domingue's history, as Sidney Mintz and Richard Price argued, perhaps gave them a critical edge in determining the shape of Creole culture. Table II shows their clear dominance in the French slave trade in the period 1701–1725, when sugar cultivation burgeoned in the colony. Finally, as Roger Bastide suggested, the association of Ewe-Fon and Yoruba with powerful African states may have won them converts on the plantations, whereas the Bantu emphasis on ancestor worship rather than pantheons of deities may have been more easily disrupted by the slave trade.50
30
A necessary corollary to these answers is to point out that the cultural impact of West Central Africans was rather greater than often has been allowed. Just as scholars in Cuba and Brazil have in recent years criticized an over-concentration on the Yoruba presence in their countries, students of Haitian voodoo have been progressively uncovering unsuspected Kongo influence in the religion, notably in the now salient Petro cult.51 One can understand why French colonists associated voodoo with western St. Domingue, given the greater presence there of slaves from the Bight of Benin (see Table VII). It is also significant that the two surviving vaudou chants from the colonial period are now known to be in the Kikongo language and that they were recorded by colonists from the north and south provinces, where West Central Africans were most numerous.52 Their overwhelming predominance in the mountains of the North Province perhaps explains why George E. Simpson in the 1940s found Petro to be fully integrated into voodoo there, instead of being the semi-separate cult it is elsewhere in Haiti.53 In colonial times it presumably had been the cult of the majority, not of "newcomers." Colonial demography may similarly explain why in northern Haiti and certain mountainous areas of the west Petro is sometimes called "Lemba," after the north Kongo healing cult from which some believe it came.54 The use of different terms in northern and western Haiti, such as adjipopo and makakri (maleficent charm), appears also to reflect the slave trade's uneven distribution of speakers of Bantu and West African languages.55 A Kongo influence on Haitian voodoo can thus be detected from colonial times to today, though surely much more remains to be discovered.
31
The place of Kongo and Ewe-Fon cultures in Haiti is paralleled in the scholarly literature by that of the Igbo and Akan in the British Caribbean. Scholars have generally neglected the Igbo, despite their prominence in the British slave trade, but lately greater cultural impact has been attributed to them.56 With this in mind, it would be interesting to know what traces Igbo may have left in Guadeloupe, where in the period 1770–1789 they constituted 37 percent of the enslaved African population.57 The local prominence they gained from the Bight of Biafra's position in the French slave trade to the island was evidently bolstered by the British slave ships that also supplied Guadeloupe's plantations (see Table V). The Igbos salience among African migrants can only have increased in the nineteenth century, and, with an unusually balanced sex ratio, they had an above-average chance of intermarrying and preserving aspects of their culture in the Caribbean (see Table II).
32
One example of survival against extreme odds can be found toward the western end of Haiti's northern plain. There the self-styled "Mandingo Society of King Muhammad" numbered about 1,000 members at the end of the 1960s. Most had inherited their membership. They practiced ceremonies with seemingly Islamic features they considered more authentically African than voodoo. Their burial customs were distinctive, resembling those of the Bambara.58 Muslims can have been only a tiny minority of the Africans brought to the French Caribbean, and the slave trade data show that northern St. Domingue was anything but a favored location for the disembarkation of Senegambians. Mandingues accounted for no more than 3 percent of the African slaves of the northern plain at the end of the colonial period. Yet this proportion was considerably more than anywhere else in the colony, and, in the plains at least, Bambara slaves were also more numerous in the north. Thus some historical basis can be found for understanding this curiosity in the Haitian cultural landscape.59
33
To date, little work has been done on regional variations in Francophone Caribbean culture. It is hoped that the knowledge gained from the slave trade data and other sources about the African forebears of the French West Indian population will help shape future inquiry into the specifics of regional cultures that are fast eroding with the incursion of roads and radio and the disruption of outmigration.
34


David Geggus is a professor of history at the University of Florida. He thanks Paul Lachance for helpful comments on a draft of the article and Beth Komski for additional help.

Notes

1 Jean Mettas, Répertoire des expéditions négrières françaises au XVIIIe siècle, 2 vols., ed. Serge Daget and Michèle Daget (Paris, 1978, 1984); Serge Daget, Répertoire Des expéditions négrières françaises à la traite illégale (1814 1850) (Nantes, 1988). The latter's data are slightly modified in Daget, La répression de la traite Des Noirs au XIXe siècle: l'action Des croisières françaises sur les côtes occidentales de l'Afrique, 1817–1850 (Paris, 1997), 138, 166–73.
2 Eltis, "The Volume, Age/Sex Ratios, and African Impact of the Slave Trade: Some Refinements of Paul Lovejoy's Review of the Literature," Journal of African History, 31 (1990), 485–92.
3 Stein, The French Slave Trade in the Eighteenth Century: An Old Regime Business (Madison, 1979); Eltis, Behrendt, and Richardson, "The Transatlantic Slave Trade, 1562–1867: A Database Prepared at the W.E.B. Du Bois Center, Harvard University," unpublished paper,, 6.
4 Munford, The Black Ordeal of Slavery and Slave Trading in the French West Indies, 1625–1715, 3 vols. (Lewiston, N. Y., 1991); Donnan, ed., Documents Illustrative of the History of the Slave Trade to America, 4 vols. Washington, D. C., 1930 1935); Ly, La Compagnie du Sénégal (Paris, 1958); Barbot, A Description of the Coasts of North and South-Guinea (London, 1746).
5 Eight of the first 9 sailings of slave ships from Bordeaux listed in Saugera, Bordeaux, Port Négrier: chronologie, économie, idéologie, XVIIe-XIXe siècles (Paris, 1995), 351, are not in the dataset; he further suggests a 17th-century total of 53 known French slaving voyages, as compared to 42 in the dataset (p. 201). For the period 1800–1805, the dataset lists 14 sailings, whereas in "Pour une histoire de la traite négrière française sous le Consulat et l'Empire," Revue Française d'Histoire d'Outre-Mer, 76 (1989), 203–29, Saugera claims 45 proven cases of slave trading, 16 probable, and 7 possible, of which only 15 involved Indian Ocean destinations. He also notes an unrecorded expedition of 1802 to Senegal fitted out in Cayenne. On the other hand, Saugera's total estimate of 508 separate voyages by ships departing Bordeaux approximates very closely the 415 in the dataset, given that more than 80 terminated in the Indian Ocean.
6 Archives Nationales, Paris, 505 Mi 85–86. A few extra details can also be found in Barault Roullon, Mémoire à consulter sur les dettes Des colons de Saint-Domingue (Paris, 1828). Paul Lachance, "The Demography of French Slave Colonies, part 1 (1700–1760)," paper presented to the Social Science History Association meeting, Chicago, Nov. 1998, makes a probing comparison for the early 18th century of the database, the colonial censuses, and the slave trade statistics generated in Philip Curtin, The Atlantic Slave Trade: A Census (Madison, 1969), and Stein, French Slave Trade in the Eighteenth Century.
7 Problems came to light regarding projections in the database concerning some expeditions whose slaves were sold in more than one port. The distortion resulting from these cases probably slightly exaggerates African arrivals in Martinique and understates arrivals in St. Domingue, especially Cap Français.
8 The latter figure (for the period Jan. 1, 1714–Aug. 22, 1721) is given in Kenneth J. Banks, "The Illicit Slave Trade of Martinique, 1716–1756," paper delivered at the Conference on the Medieval and Early Modern Atlantic Economy, Charleston, Oct. 1999. Port statistics reported in Jean Tarrade, Le commerce colonial de la France à la fin de l'ancien régime: l'évolution du régime de l'Exclusif de 1763 à 1789, 2 vols. (Paris, 1972), 2:520, suggest that possibly a few hundred slaves sold in the Windward Isles in the 1780s may have been missed.
9 Archives de France, Section d' Outre-mer, Aix-en-Provence (ANOM), Colonies, F 3/149, fol. 196 (1785); C 9 A/161, letter of Mar. 20, 1788 (1786, 1787); CC 9 A/4, letter of Mar. 10, 1790 (1788); C 9 B/39, "État" pour 1789 (1789).
10 Eltis and Stanley L. Engerman, "Fluctuations in Sex and Age Ratios in the Transatlantic Slave Trade, 1663–1864," Economic History Review, 46 (1993), 308–23; Eltis, "Volume, Age/Sex Ratios, and African Impact of the Slave Trade" 489.
11 Miller, Way of Death: Merchant Capitalism and the Angolan Slave Trade, 1730–1830 (Madison, 1988), 78, 220 28.
12 Only 4 expeditions are reported to have acquired slaves at Benguela, 3 at Luanda, and 24 at Ambriz, compared to 610 on the Loango coast and 438 from "Angole."
13 Disembarkation statistics show that the decline was real and not caused by seizures during the "illegal trade" of vessels before they were fully loaded.
14 Eltis and Richardson, "Productivity in the Transatlantic Slave Trade," Explorations in Economic History, 32 (1995), 465–84. The increasing concentration on St. Domingue rather than colonies farther east does not seem sufficient to account for the magnitude of change.
15 Imputed mortality rates per vessel correlated as follows: with crossing time to the Americas (R2 =. 152, p =. 000), and with region of embarkation (F = 9.764, sig 000).
16 On voyages from the Bight of Biafra, mortality correlated strongly with crossing time (R2 =. 39, p =.000).
17 On ethnic stereotypes see Geggus, "Sugar and Coffee Cultivation in Saint Domingue and the Shaping of the Slave Labor Force," in Ira Berlin and Philip D. Morgan, eds., Cultivation and Culture: Labor and the Shaping of Slave Life in the Americas (Charlottesville, 1993), 79–90; Geggus, "La traite Des esclaves aux Antilles françaises à la fin du 18 me siècle: Quelques aspects du marché local," in Silvia Marzagalli and Hubert Bonin, eds., Négoce, Ports, et Océans, XVIe-XXe Siècles: Mélanges Offerts à Paul Butel (Bordeaux, 2000), 235–45; and note 36 below. The basis of such stereotyping was a mixture of supposed physical and cultural attributes, among which average height was demonstrably important. All rice-eaters were seen as difficult to feed, adapting badly to Caribbean diets. Those of Sierra Leone, who were said easily to sicken and die, had a negative reputation; whereas their taller counterparts from Senegambia had a more ambiguous reputation.
18 Patrick Manning, Slavery and African Life: Occidental, Oriental, and African Slave Trades (Cambridge, 1990), 56; Michael Craton, James Walvin, and David Wright, eds., Slavery, Abolition, and Emancipation: Black Slaves and the British Empire (London, 1976), 88; Daniel C. Littlefield, Rice and Slaves: Ethnicity and the Slave Trade in Colonial South Carolina (Baton Rouge, 1981), 17.
19 Eltis, Behrendt, and Richardson, "Transatlantic Slave Trade," table II; Herbert S. Klein and Engerman, "Long-Term Trends in African Mortality in the Transatlantic Slave Trade," Slavery and Abolition, 18, No. 1 (1997), 36–48, table II.
20 Though this was clearly not the case at the turn of the 19th century, it may be that reduced American demand for slaves in that period helped reduce shipboard mortality, in the manner explained in Miller, Way of Death, 433.
21 Behrendt, "Crew Mortality in the Transatlantic Slave Trade in the Eighteenth Century," Slavery and Abolition, 18, No. 1 (1997), 49–71.
22 As with the British trade, this decline was partly owing to a shift away from high mortality areas toward West Central Africa; see ibid., 51, 61. For the French, unlike the British, Senegambia was a low mortality region, the two nations trading in different ecological zones.
23 Klein, The Middle Passage: Comparative Studies in the Atlantic Slave Trade (Princeton, 1978), 197.
24 I have included here 6 expeditions (0.2% of cases) to Tobago during the brief time it was a French colony.
25 See above, note 7. More than 2,500 slaves were imported into the zone occupied by British troops during the years 1793–1798. In the period 1800–1820, the ex-slave governor Toussaint Louverture and head of state Henry Christophe also purchased Africans from slave ships for employment on plantations and as police. In addition, the dataset includes one vessel coming directly from Africa in 1817 that landed its 521 slaves, not in Christophe's northern kingdom, but in Alexandre Pétion's Republic of Haiti.
26 Previous estimates, using Mettas, Répertoire Des expéditions négrières françaises, suggested slightly lower figures (nearly 18,000); others, using the Affiches Américaines port statistics, rather higher figures (more than 22,000). See Geggus, "The Haitian Revolution: New Approaches and Old," in James Pritchard, ed., Proceedings of the 19th Meeting of the French Colonial Historical Society (Cleveland, 1994), 143 n. 8, and Geggus, "The Sugar Plantation Zones of Saint Domingue and the Revolution of 1791–1793," Slavery and Abolition, 20, No. 2 (1999), 45 n. 30; Jean Fouchard, Les marrons de la liberté (Paris, 1972), 217–18.
27 The Du Bois Institute database makes a valuable contribution in this respect. According to Eltis, it shows some 90,000 Africans disembarked in the French Caribbean from foreign vessels, of which 86% were British. Martinique received 45% of the captives; Guadeloupe, 37%; St. Domingue, only 14%. These figures do not include slaves re-exported from foreign colonies to their French neighbors.
28 Tarrade, Commerce colonial de la France à la fin de l'ancien régime, 2:517–24. The latter authorization was the work of local officials only. Bounties of 200 livres were also paid on slaves landed in southern St. Domingue and Cayenne and bounties of 160 livres on those in Martinique and Guadeloupe.
29 Liliane Crété, La traite Des nègres sous l'ancien régime: le nègre, le sucre et la toile (Paris, 1989), 204–05, 208, 233; Geggus, "La traite Des esclaves," 235– 45.
30 Geggus, "The Major Port Towns of Saint Domingue in the Late Eighteenth Century," in Franklin W. Knight and Peggy K. Liss, eds., Atlantic Port Cities: Economy, Culture, and Society in the Atlantic World, 1650–1850 (Knoxville, 1991), 87–116. More neglected than Cayenne, in fact, was St. Lucia, which attracted no French slave ships at all, supposedly because of its lack of specie; Tarrade, Commerce colonial de la France à la fin de l'ancien régime, 2:521 n.
31 Tables IV and V use the imputed figures in the dataset. The age/sex ratios in Table IV are averages of the proportions per vessel. Calculations using the aggregate numbers disembarked per colony (as in Table VI, for St. Domingue) produce somewhat higher ratios but without changing the ranking. Using this method, the proportion of children among Africans disembarked in Cayenne, for example, was 33.5% (N = 1,280). On ethnic stereotyping, see note 17 above.
32 Geggus, "The Demographic Composition of the French Slave Trade," in Philip Boucher, ed., Proceedings of the 13th/14th Meetings of the French Colonial Historical Society, Natchez, 1988 (Lanham, Md., 1990), 14–30, and "Sex Ratio, Age and Ethnicity in the Atlantic Slave Trade: Data From French Shipping and Plantation Records," J. African Hist., 30 (1989), 23–44. These articles used only 18th-century expeditions for which sex and age ratios could be calculated.
33 Archives Nationales, Paris, 505 MI 85–86, correspondence of the Cap Français merchant house Foäche, Morange & Hardivilliers, 1785–1791.
34 This generalization masks the rapid growth of Guadeloupe's slave population between 1750 and 1780, when it overtook Martinique's, after being boosted by importations during the British occupation of 1759–1763. Neither population grew very much during 1783–1790, the height of the French traffic.
35 Geggus, "Sugar and Coffee Cultivation in Saint Domingue," 80.
36 Ibid. Not all colonists agreed in these matters. For a sugar planter who thought highly of Igbo workers and much additional detail on ethnic stereotypes, see Bernard Foubert, Les habitations Laborde à Saint-Domingue dans la seconde moitié du XVIIIe siècle, contribution à l'histoire d'Haïti (Plaine Des Cayes), 2 vols. (Lille, 1990), 458–64, 545. The Islamic names of some of these slaves, however, suggests the colonist confused Igbo and Hausa, who probably arrived in the same ships.
37 Geggus, "Indigo and Slavery in Saint Domingue," Plantation Society in the Americas, 5 (1998), 189–204.
38 Tarrade, Commerce colonial de la France à la fin de l'ancien régime, 2:519. In 1720, African slaves sold for 900 livres in St. Domingue and 700 livres in Martinique. See Crété, La traite Des negrès sous l'ancien régime, 226.
39 These tentative conclusions, based on a limited quantity of price data in plantation papers, port records, and merchant correspondence, are drawn from Geggus, "La traite Des esclaves aux Antilles françaises," table III. See also the 1784 slave import statistics reprinted in Fouchard, Les marrons de la liberté, 120, which show average prices of 1,709 livres for northern St. Domingue and 1,907 livres for the rest of the colony. This average price of 1,795 livres rose in the following years to 1,968 livres (1786), 1,963 livres (1787), 2,099 livres (1788), 2,134 livres (1789). See ANOM, C9A/161, letter of Mar. 20, 1788; C9A/163, letter of Apr. 23, 1789; C9B/39, letter of June 25, 1788; and "État pour 1789"; CC9A/4, letter of Mar. 10, 1790.
40 See above, note 32.
41 Bean, The British Trans-Atlantic Slave Trade, 1650–1775 (New York, 1975); Eltis, The Rise of African Slavery in the Americas (Cambridge, 2000).
42 The average time in port at Cap Français fell from 6.5 months before 1730 to 4 months in the 1740s, 3.5 months in the 1750s, and just over 3 months in 1785–1790.
43 Vanony-Frisch, "Les esclaves de la Guadeloupe à la fin de l'ancien régime," Bulletin de la Société de la Guadeloupe, 63–64 (1985), 39. Compare Bernard David, "La population de Rivière-Pilote (Martinique), 1802–1829," Rev. Franç. d'Hist. d'Outre-mer, 60 (1973), 352, and Myriam Cottias, "Mortalité et créolisation sur les habitations martiniquaises du XVIIIe au XIXe siècle," Population, 44 (1989), 55–84.
44 See above, p. 126. Even though the rate of importation declined by one quarter in the first half of 1791, more than 8,000 Africans disembarked in Cap Français between Jan. and Aug. of that year.
45 Geggus, "Sugar Plantation Zones of Saint Domingue," 31–46.
46 Only rare inventories use subregional terms (Congo-Mayaca, Congo-Moussondi, and so on), as do some prison lists of recaptured fugitives. Personal names might offer a few clues, if their regional usage could be precisely delimited. Captives exported from West Central Africa by the French were assuredly more culturally homogeneous than those taken to Brazil in the 19th century.
47 Morgan, "The Cultural Implications of the Atlantic Slave Trade: African Regional Origins, American Destinations, and New World Developments," Slavery and Abolition, 18, No. 1 (1997), 122–45.
48 The following discussion comes from Geggus, "Haitian Voodoo in the Eighteenth Century: Language, Culture, Resistance," Jahrbuch für Geschichte von Staat, Wirtschaft, und Gesellschaft Lateinamerikas, 28 (1991), 21–51.
49 Debien et al., "Les origines Des esclaves Des Antilles," Bulletin de l'Insitut de l'Afrique Noire, ser. B (1961–1967), 23–29.
50 Mintz and Price, An Anthropological Approach to the Afro-American Past: A Caribbean Perspective (New Haven, 1976), 25–26; Bastide, African Civilizations in the New World (London, 1971), 105–06.
51 Suzanne Comhaire-Sylvain, "Survivances africaines dans le vocabulaire religieux d'Haïti," Études Dahoméennes, 14 (1955); Joseph Cornet and Robert Farris Thompson, The Four Moments of the Sun: Kongo Art in Two Worlds (Washington, D. C., 1981); John M. Janzen, Lemba, 1650–1930: A Drum of Affliction in Africa and the New World (New York, 1982).
52 See note 48 above. In the 1950s, too, the Ewe-Fon Rada cult was the main cult in much of the western plains, while the Congo cult was dominant in the southern coffee-growing region of Jacmel. See Odette Mennesson-Rigaud, "Le rôle du vaudou dans la guerre d'indépendance d' Haïti, Présence Africaine,, 18/19 (1958), 43–67.
53 Simpson, Religious Cults of the Caribbean (Rio Piedras, 1970), 238.
54 Jean Price Mars, "Lemba-Petro, un culte secret," Revue de la Société haïtienne d'histoire et de géographie, 28 (1938), 24–25; Janzen, Lemba, 273 292; Comhaire-Sylvain, Les montagnards de la région de Kenscoff (Rép. d'Haïti): une société Kongo au-delà Des mers (Kinshasa, 1984), 169.
55 Michelson Paul Hyppolite, Les origines Des variations du créole haïtien (Port-au-Prince, 1949). Note also the Fon word marassa (twins) is associated with the west but not the north. See Glenn R. Smucker, "The Social Character of Religion in Rural Haiti," in Charles R. Foster and Albert Valdman, eds., Haiti: Today and Tomorrow (Lanham, Md., 1984), 56. Similarly, northerners tend to describe voodoo deities as zanges (angels), rather than use the West African word lwa. See Gerson Alexis, Lecture en anthropologie haïtienne (Port-au-Prince, 1970), 178.
56 Douglas B. Chambers, " 'My Own Nation': Igbo Exiles in the Diaspora," Slavery and Abolition, 18, No. 1 (1997), 72–97.
57 Vanony-Frisch, "Esclaves de la Guadeloupe," 32.
58 Alexis, Lecture en anthropologie haïtienne, 173–85.
59 See the statistics in Geggus, "Sugar and Coffee Cultivation in Saint Domingue," 81, and "Sugar Plantation Zones," 39.