For historians of slave societies in the Americas, the nineteenth century assumes a particular importance. Although slave labor had been virtually synonymous with European settlement in much of the Americas, it was only in the few decades before the end of slavery that the collection of economic and demographic statistics increased in both scope and reliability. Usable trade statistics, census counts, and business records are much more abundant for the nineteenth than for earlier centuries. For the two societies where the slave trade lasted longest, Brazil and Cuba, there are now available reasonably complete annual data on the volume and prices of most plantation crops. Less has survived on the size and costs of the slave labor forces which produced those crops, but some good slave price series for both Brazil and Cuba have appeared recently. In addition, census data for Cuba have been subjected to more rigorous examination. For Brazil, the pre-1872 era remains a relative wasteland as far as population counts go, but an innovative national income series and a price index are now available for much of the nineteenth century. For both societies, a continued influx of slaves was the basis of their rise to economic prominence in the nineteenth-century Atlantic economy. Decadal totals for this influx of slaves were estimated nearly 20 years ago by Philip Curtin. These have been modified subsequently, but Brazil and the French Caribbean (another important nineteenth-century importing zone) in particular still lack the annual data which together with other nineteenth-century social and economic series should permit new insights into Afro-American societies.1 The present essay attempts to fill this gap by generating annual estimates of slave imports in the last six decades of the slave trade. These incorporate research carried out since Curtin’s pioneering study. The essay revises and considerably extends earlier annual estimates published in the late 1970s.

Most of the sources for the nineteenth-century slave trade are British. Indeed, estimates issued by the Brazilian government and used by Brazilian historians were probably British in origin.2 This despite the fact that, after 1808, the British Americas received very few slaves from Africa and few British ships were involved in the shipping of slaves to the rest of the Americas. The explanation for this apparent anomaly is simply that after suppressing their own slave trade, the British devoted extensive resources to suppressing all slave trafficking across the Atlantic. The motives for such actions were deeply rooted in the process of economic growth in Britain and have been explored elsewhere,3 but the effect was that the British negotiated treaties with governments of importing and exporting regions which, on paper at least, put an end to the trade. Thus, the Portuguese trade was confined by an 1815 Anglo-Portuguese convention to regions south of the equator, and in 1826 the Brazilians, by now independent, agreed to prohibit all slave imports north and south of the line, a measure which took effect in 1830. The equivalent measure for Cuba was the 1817 Anglo-Spanish convention which was to take effect in 1820.4 France refused to enter into a similar agreement and eventually suppressed her own slave trade in 1830-31. By 1850, however, a vast British-centered network of slave trade treaties existed incorporating almost every country which had an Atlantic seaboard and many others which did not. Nevertheless, slaves from Africa continued to arrive in the Americas until 1867.

To organize and monitor this network, the British Foreign Office created a separate Slave Trade Department which grew from one clerk in 1819 to five full-time officials by the 1840s. This was the center of an extensive reporting system. British consuls in Havana, Santiago de Cuba, Rio de Janeiro, Bahia, Pernambuco, Maranhão, and Pará, and the British legation at Rio de Janeiro supplied the department with reports on the state of the slave trade, in addition to their regular consular duties. Consuls on the eastern side of the Atlantic in Nantes, Lisbon, Cádiz, Barcelona, Santiago in the Cape Verde Islands, and Santa Cruz were also expected to pass on information that might come their way. Most of these men were substantial merchants, familiar with local commerce and in a good position to report on ship movements and details of the organization of the trade. Some of them had their own informers paid, as in the case of the legation in Rio, with Foreign Office funds. Even more fertile sources of information were the judges and arbitrators of the Courts of Mixed Commission. Treaties between Britain on the one hand, and Spain, Portugal, Holland, and eventually many more countries on the other, resulted in the establishment of tribunals with the power to condemn slave ships belonging to the contracting parties captured by authorized cruisers. Between 1819 and 1867, Mixed Commission Courts sat in Rio de Janeiro, Havana, Paramaribo, Luanda, Kingston (Jamaica), New York, Capetown, and Sierra Leone, and the British commissioners sent back volumes of reports on the slaving voyages that came to their attention. The Admiralty completed the system by relaying to the Foreign Office extracts from the dispatches of commanders in the Cape of Good Hope, Brazil, and Jamaica stations. The antislave-trade squadron off West Africa varied between 5 and 36 ships, and regularly visited slave marts on the African coast and boarded slavers. Many other cruisers outside the squadron, however, carried Admiralty instructions which enabled them to search suspected vessels. Thus, at any given point between 1811 and 1867, the British had dozens of well-placed and well-qualified observers collecting data on the transatlantic traffic.5

The figures presented here are based partly on a data set of 5,378 voyages, and partly on the estimates of contemporary observers. The major source of both the data set and observations is the Foreign Office 84 series at the British Public Record Office (PRO)—essentially the records of the Slave Trade Department—but Foreign Office 72, 63, Admiralty 1, 123 and various High Court of Admiralty series in the same location have also yielded valuable information.6 About half the data set was printed in the parliamentary papers in 1845,7 though additional information has been uncovered for many of the voyages included in that publication. The expansion of the data has been greatly facilitated by three scholars in particular who have given generously of their own researches. Herbert Klein made available data taken from the Rio de Janeiro newspapers between 1825 and 1830 which largely confirm the reports sent to the Foreign Office by British officials there. Joseph Miller contributed data culled from the Foreign Office 63 series at the PRO for the pre-1820 Brazilian traffic. Finally, Serge Daget has given much from his own exhaustive work on the French traffic, though this branch remains the most poorly represented of all the major national participants.8

Information from ships and observers is sufficiently detailed to allow a breakdown by major importing and exporting regions. Here we concentrate on slave arrivals in the Americas, and divide our discussion according to the well-known political boundaries for the Americas. The main purpose here is to develop an annual series. Nevertheless, an aggregation of the data is provided at the end of the article first for each year and then for each major importing region for the entire 1811-67 period. The reliability of the estimates varies. Readers are warned that the series developed here is no more than the best it is possible to assemble from currently available sources rather than a definitive record of slaves actually traded across the Atlantic.

## Brazil

Estimates have been developed for three separate regions in Brazil. Table I shows three separate import series for each region. The first is simply slave arrivals counted directly from the slave ship data set. The second adds to this series an allowance for ships which disembarked an unknown number of slaves. This allowance is based on cargoes of known size which comprise 62 percent of Bahian arrivals, 73 percent of Rio arrivals, and 14 percent of those arriving in North Brazil. These known cargoes are spread evenly over the 42-year period. It is thus possible to derive an average cargo figure for 5, 6, or 7-year periods for each importing region and assign it to those landings of unknown size.

The third series is the least reliable of the three and requires detailed explanation. It is composed of estimates, some by the group of officials described above and others by modern writers. For many regions and periods, all three series have the same values, for example, southern Brazil in 1821. Here it is assumed that the British officials recorded almost every slave that arrived. For most regions and periods, however, recorded arrivals were less than actual arrivals, particularly for the period after 1830, when the slave trade was illegal and slaves were likely to be landed on a wider range of coastline.

### Brazil South of Bahia Province

Records of slaves entering Rio de Janeiro between 1811 and 1830 appear to be relatively complete as a result of the work of British officials, and are probably based on daily shipping reports.9 Such reports were carried by the Rio de Janeiro newspapers between 1825 and 1830, and these fill in the gaps in the British data for those years. In 1814 estimates for missing months have been added on the basis of the monthly distribution of arrivals between 1795 and 1811.10 It is questionable if all slaves carried into southern Brazil before 1830 were channeled through Rio de Janeiro. This issue arises because about 20 percent of the slave ships leaving Luanda in the 1820s gave “southern Brazil” as their destination rather than Rio de Janeiro.11 However, it is likely that southern Brazil and Rio de Janeiro were interchangeable as far as the Angolan records are concerned, and that Santos, Campos, and other centers received few slaves directly from Africa at this period. First, there is no suggestion in the British reports of arrivals at locations other than Rio de Janeiro. Second, the commercial dominance of Rio was such that almost all overseas trade with southern Brazil was conducted through the city. Finally, until the coffee boom spread to the western Paraíba valley in the 1840s, the economic base of regions south and west of Rio was simply not strong enough to absorb slave imports in the lots of 400 to 500 which transatlantic slave ships normally carried.

Under the terms of the Anglo-Brazilian treaty of 1826, slaves could no longer be brought into Brazil after September 1830, and Brazilian legislation to this effect was passed in the following year. There were undoubtedly few arrivals in 1831 and 1832, partly because of the glutted market created by massive inflows in 1829 and 1830, and partly because the slave dealers did not know how strictly abolition would be enforced. But as the British navy reported trading activity in Luanda, the major Brazilian supply port,12 some landings must have occurred. Early in 1833, at least one large estate owner was importing “considerable numbers,”13 and by the end of the year slave arrivals were “increasing rapidly. ”11 In 1834, “32 vessels sailed . . . from this port [Rio de Janeiro]” for the coast of Africa.15 The commissioners reported a “continued” and “uninterrupted” traffic in mid-1835, and approximately 45 departures in the second half of the year.16 In the following year, there were between 50 and 60 departures in the first six months.17 All but a few would have returned with slaves; those that did not were counterbalanced by ships which landed slaves and returned to Africa without clearing from Rio. Thus, official British correspondence indicates a steep upward trend in imports from 1833 to 1836. The actual figures in the third series from 1831 to 1833 are the result of a linear projection from 1834, when some data are available, to a very low figure, arbitrarily chosen at 1,000 for 1831. The 1834 estimate is the product of ship departures from Rio for Africa and the average cargo landed between 1831 and 1843. The same technique produces a figure for the first three quarters of 1836, and to this may be added the inferred total for the second series, all of which arrived in the last quarter. The 1835 estimate is again a linear projection from the years 1834 to 1836. Rough confirmation of this last estimate is provided by a clerk in an English merchant house which dealt with slave traders, who wrote of 19,000 arrivals in the first nine months of the year.18

After 1836, British officials recorded individual landings which came to their attention and occasionally provided estimates for those which did not. Thus for 1837, 1838, and 1843 we have estimates of 46,000, 42,970, and “between 30,000 and 40,000,”19 respectively, with the low end of the range for the latter year being accepted here. Between 1839 and 1842, Foreign Office personnel in Brazil declined to make estimates of imports into Rio, and noted only that actual imports greatly exceeded the arrivals listed in their dispatches. However, British officials in Rio, as elsewhere, did not always cooperate with each other. Relations between Sir George Jackson, the commissary judge, and W. G. Ouseley, the chargé d’affaires, were particularly strained in these years. The commissioners, the British minister at the Brazilian court, and the consul each appear to have used separate sources of information on the slave trade. As a result, the import series derived from the data set is greater than the annual figures supplied by any one official in Rio.20 Thus, the second series in Table 1 for the years 1839-42 is probably a very large sample, though certainly less than 100 percent. In the absence of better information, however, I assume that British officials caught all arrivals in their information net. In 1844, details of slave ships disembarking 16,118 slaves were returned, but Robert Hesketh and Frederick Grigg, the British commissioners in Rio de Janeiro, admitted missing many landings, and the Foreign Office apparently made allowances for these in their estimate of 19,500 incorporated in a return made for Parliament in 1852.21 For 1845 Hesketh and Grigg, and for 1846 and 1849 Hesketh alone, estimated total arrivals, with the latter noting that this 1849 figure may well have been too low. For other years there are estimates by the British minister to Brazil, backed up by returns from the British consul at Bio for some six-month periods.22 Unlike the early 1840s when British officials in Bio worked independently of each other, there is evidence in the last five years of the trade, particularly after the closing of the British-Brazilian mixed commission court, that the consular and diplomatic staff produced joint estimates of the traffic.23

### Bahia

The Foreign Office was not as well represented in Salvador da Bahia and Brazilian ports farther north: officials were both fewer in number and less diligent in their researches than their Rio counterparts. For Bahia, they returned neither detailed reports of landings nor estimates for the years 1818 to 1821, 1828, 1831 to 1839, the first quarter of 1811, the last half of 1814, the last quarter of 1826, and the first and third quarters of 1827. Other sources fill the gaps, however. Francisco Marques Goes Calmon published a complete series for Bahia for the years 1812 to 1830 which correlates well with the British consular series for the 11 years in which the two overlap.24 Goes Calmon’s figures are accordingly accepted for all years between 1812 and 1830 for which gaps exist in the British data. A rough check on this procedure, as well as an estimate for 1811, can be obtained by adjusting the British annual data with missing quarterly reports according to the proportions landed in these quarters during the years for which the data are complete. A comparison of the results for the three years where this is possible with the Goes Calmon figures further confirms the similarity of the series.25

Neither Goes Calmon nor the Bahian shipping data which Pierre Verger used can provide much help for the other problem years, and other methods must be used. As in Rio de Janeiro, imports would have been very low in 1831 and a figure of 1,000 is again arbitrarily assigned for this year. Between 1832 and 1836 and in 1839, the third series is based on the returns of ships entering Bahia in ballast from the coast of Africa compiled by the British consul. There is a strong presumption that these ships landed their slave cargoes outside Bahia. For the years 1837 and 1838, there is again little indication of the volume of imports, but for figures in these years and for support for other years in the series after 1830, we may rely on two independent contemporary sources. In 1835, a member of the British legation at Rio investigated the Bahian traffic in person, and reported imports at 12 to 15 shiploads a year26 or 4,000 to 5,000 slaves; eight years later the British consul at Bahia stated that “no regular copies of the returns were kept at this consulate up to the end of 1839 . . . but from the best information I have been able to obtain [imports] averaged about 4,000 annually.”27 From 1840, the series incorporates the lists of individual landings which the consul and the navy once more began to supply. As with the Rio figures for 1839-42 we assume that all arrivals were recorded. The slave trade to Bahia, though illegal, appears to have been carried on more openly than in any other part of the Americas at this time. Perhaps because of this, all shipments into the province were landed relatively close to the capital. There was a note of confidence in slave trade reports of the British consul for Bahia which was not always present in those of his counterparts in Havana and Rio de Janeiro.28

### Brazil North of Bahia Province

Information on arrivals in the provinces north of Bahia is more scattered than for other Brazilian regions. Maranhão, Pernambuco, Pará, and for the last part of the period, Paraíba, each had British consuls, but the reports of the latter were infrequent. The Pernambuco consul listed arrivals in 1829, reported slaving activity in 1831, 1832, 1835, and 1836, but began listing individual landings again only from 1837. In the later 1830s, both the consul and the commander of the naval packet boat which served Pernambuco stated that the traffic rivaled that of Bahia in size.29 While their belief is supported by the number of reported arrivals and the fact that Pernambuco acted as entrepôt for Paraíba and other northern ports,30 the traffic dropped off quickly in the 1840s. For the years before 1829, British data on imports into Pernambuco simply do not exist. Maranhão was the other significant northern importing center. It received 8,446 slaves between 1821 and 1828,31 with the annual total declining steadily from the high of 1,761 in 1821. After 1828 there were occasional landings only.32 Ceará, Pará, and Parnaíba received very few slaves directly from Africa, with the last two receiving no further shipments after 1835.33

For the years up to 1825 there are Correia Lopes’s data for the three main ports in the region for 1812, 1813, 1815-19, and 1823-25.34 These data are for slaves from Angola only, however, and an addition of 25 percent is made as an allowance for imports from other African regions. This step is justified by data from the years 1829-51, when some evidence of imports from all African regions exists. The figure for 1811 is a mean of imports between 1809 and 1813, while that for 1814 is similarly derived from the 1812-19 figures. After 1819, imports, particularly to Pernambuco, declined markedly. The figure for 1820 is a mean of those years between 1816 and 1825 for which data exist, but because of the pronounced fall in Pernambucan imports and the availability of data for Maranhão from the Foreign Office records, the estimates for 1821-22 and 1826-28 are part interpolation and part observation. In these years, the Pernambuco figure is based on the trend line from 1819 to 1823 and from 1825 to 1829. To the resulting figures are added the Foreign Office data for Maranhão and an allowance of 1,000 for each year for Pará. For 1829 and 1830, we have the Foreign Office figures for Pernambuco to which is added 500 for all other ports in the region. During 1831-38 imports into Pernambuco arc assumed to have been the same as those into Bahia, a step justified by the fact the economies of both regions were based on sugar. The estimate for Maranhão and Pará in the 1831-35 period is held at 500 per year. After 1835 in Maranhão and the other ports and after 1838 in Pernambuco, recorded arrivals are accepted as actual arrivals. For the period 1825-51, recorded arrivals amount to perhaps one third of actual imports.

## Cuba

The most frequently used import series for the last years of the legal slave trade (up to 1820) is based on Havana custom house returns. It was published first by Alexander von Humboldt in 1826 and again with minor variation by José Antonio Saco six years later. Recently, Herbert S. Klein has published fragments of the original data on which these sources were based.35 This series ignored legal shipments into other Cuban ports opened to the slave trade by the 1789 and 1791 cédulas, as well as smuggling into all parts of the island. Saco and Humboldt believed that a further 25 percent should be added to the Havana figures to arrive at imports for Cuba as a whole.36

A rough check on these estimates is possible using census data. The Cuban slave population appears to have had a rate of natural decrease during 1817-46 of about 2.2 percent a year.37 It is arguable that a similar rate held for the period 1792-1817, for which census data are available. Between these years, a small population base and a large influx of slaves no doubt distorted the age and sex distribution of the slave population more than in the later period, but on the other hand, sugar cultivation with its higher mortality rates was not yet as dominant as it was to become. If we assume a seasoning death rate of 0.1 within the first year of arrival—a rate for which there is solid evidence from other Caribbean regions—then the formula

$M=P1−P0−(P0r)1−S(2+r)2$
• where M = imports

• P1 = population at end of period

• P0 = population at beginning of period

• r = rate of natural change in population

• S = seasoning death rate38

yields imports of 246,800, compared to the Havana custom house series of 156,400, or, if we allow for 25 percent smuggling and arrivals at other ports, 195,000. Given the length of the period covered and the difficulties with the census data, these two independent estimates should be taken as mutually reinforcing in a broad sense. There is, however, the further piece of evidence that after 1820, slave arrivals at the Cuban outports averaged 20 percent of those at Havana.39 If this ratio held before 1820, and if the Saco-Humboldt correction is accepted, smuggling must have been only the difference between 20 and 25 percent. Perhaps there are grounds then for raising their 150-year-old adjustment from one-fourth to one-third of the Havana custom house figures, to allow for smuggling in excess of the 5-percent discrepancy. Table II incorporates this judgment for the years 1811-20.

For the period after 1820, we can return to British sources. The sample of slave imports to Cuba in the Foreign Office records is large, and we can therefore use the same approach in Table II as in Table I. The first column is a count of slaves, the second includes an allowance for cargoes of unknown size, and the third is an estimated series based partly on the first two columns and partly on the assessments of British officials.

The British commissioners at the Havana Court of Mixed Commission sent returns on arrivals in the Havana area throughout the period. From 1833, David Tolmé, the British consul in Havana also reported on slave imports relying in part on sources of information separate from those of the commissioners. In addition, the British consul at Santiago, John Hardy, sent occasional reports on imports in the South after his appointment in 1832. The Havana officials were very thorough; in 1842, the British commissioners gained access to the books of the Havana slave traders for the period 1835-42 and found in them few landings about which they did not already know.40 It is likely that in every year from 1821 to 1867 they recorded over 90 percent of all the landings made by slavers which had any connections with the Havana area. Connections in this sense would include those slavers owned by outport residents or disembarking slaves in the outports, but in both cases calling at Havana for refit or for outward cargo. Inevitably, however, they missed many landings at Santiago and Trinidad de Cuba. There are three indications of the scale of these missing landings. Between 1832 and 1835, John Hardy reported eight or nine slavers a year landing 2-3,000 slaves in the Santiago area.41 Unfortunately, he gave details of the landings in only a few cases, and it is not possible to check the data against the Havana lists to eliminate those ships which subsequently entered Havana and were spotted by the British officials there. It is thus not possible to consolidate the two sources. A second indication is Tolmé’s estimate that 20 percent should be added to his returns to take into account unrecorded arrivals in the island as a whole.42 Finally, the British commissioners at Sierra Leone reported all the Cuban slavers which came to their attention. About 30 percent of these intended to land slaves in the outports.43 Allowing for the fact that some of the ships disembarking near the outports became known to British officials in Havana, the Sierra Leone figures tend to support Tolmé’s suggestion for an addition of 20 percent to the Havana figures. As the present body of data is larger than Tolmé’s, it is reasonable to claim for it a sample size of 95 percent for the period 1832-43. For the years before Tolmé’s appointment, a sample size of 80 percent seems likely, with the exception of the two years 1824 and 1825, when British cruisers reported on arrivals in the Santiago area and increased the sample to perhaps 95 percent. The third series in Table II incorporates these judgments.

From the mid-1840s, disembarkation of slaves in Cuba began to occur over a much wider range of the coast than before. This was partly in response to the efforts of Captain General Gerónimo Valdés to suppress the trade in 1842 and 1843 and partly as a consequence of the Spanish penal law on the slave trade of 1845. From 1843 on, British officials adjusted to the new tactics by drawing up annual estimates which summed the shipments of known size and then added to the total a further third to allow for landings which had escaped their information network.44 This allowance, however, was also intended to cover landings which had come to their attention but for which no data on numbers of slaves disembarked were available. Assigning an average figure for each of these cargoes of unknown size results in slightly higher estimates than British officials reported for the years 1845 and 1850.45 In 1849, “[t]hree or four” cargoes among those landed came from Brazil and because we are concerned here only with transatlantic shipments, the British commissioners estimate for that year has been reduced from 8,700 to 7,400. Imports from Brazil may also have occurred in 1850, but the commissioners comments on the subject are ambiguous and no adjustments have been made.46 It might be noted here that total imports into Cuba in 1850 were low, and any bias from this decision on the aggregate series developed below will be small. For 1852 the British consul felt that the addition of one-third to the known arrivals was excessive and the estimate of imports has been arbitrarily reduced from 7,924 to 7,000.47 For all other years, the third series for Cuba in Table II follows the estimates of the Havana Foreign Office representative, although it should be noted that the figures for the 1860s in Table II are for the year January 1 to December 31, and not October 1 to September 30 as reported in the Havana returns.

Naval and diplomatic activities on both sides of the Atlantic meant that after 1850 the slave trade was confined to the Bight of Benin, the Congo region, and parts of southeast Africa. The British navy was now able to establish a blockade which, if it did not immediately suppress the trade, at least gave the naval officers and commissioners in Africa a good basis for estimating its extent. From the mid-1850s, it is possible to compare these reports with the Havana-based estimates of imports, because by this time Cuba was the only major American region still importing slaves. It was in these years that the British officials in Havana frequently noted that the addition of one-third to their count of slave arrivals to allow for unknowns was inadequate. The volume of the traffic increased markedly they said, and landings were taking place on ever more remote parts of the Cuban coast. In fact, the reports from Africa suggest that their concerns were unjustified, and that for some years even the addition of one-third may have been excessive. Thus, in 1856, naval officers off Africa reported 11 ship departures, or about 6,000 slaves,48 compared to the Havana commissioner’s import figure of 5,478, or 7,304 with the one-third addition. As African-based officials made no assessment of the completeness of their report, the higher of the two Cuban figures is accepted here. For 1857, the British commodore on the west coast reported shipments totaling 7,400. To these might be added six cargoes from the east coast of Africa reported by the British commissioners at the Cape of Good Hope. An allowance of 750 a cargo yields a total African export figure of 11,900, which after adjusting for deaths at sea is not inconsistent with the 10,436 imports estimated in Havana.49 In the following year, separate estimates by the Sierra Leone commissioners and West African commodore suggest 10–11,000 exports from West Africa, to which 3,000 might be added for the east coast. Crawford’s estimate of 16,992 imports into Cuba thus looks high, despite his comment that the one-third allowance for unknowns incorporated into this figure was inadequate; perhaps 15,000 might be a more appropriate figure.50 In 1859, the African sources indicate 20,000 departures. However, for both 1859 and 1860, the peak years of the post-Brazilian traffic, the African reports are noticeably less confident in their estimate of total exports. Nevertheless, Crawford’s Havana-based estimate of 30,473 imports would mean that the British West Coast squadron knew of only three out of every five ships escaping with slaves. Given the blockading system and the fact that each slaver spent some weeks off the African coast, this seems implausible. These considerations were probably the basis of W. Wylde’s (the head of the Slave Trade Department) revision of the 1859 figure to 25,000, and Wyldes figure is accepted here.51

In 1860, the efficiency of the West African squadron declined noticeably, and the ratio of slavers met on the coast to slave ships on which information had been received from all sources was at its lowest point in many years in the period 1857-67.52 The export figure of 10,450 is thus much below the import estimates of 18,671 and 24,895 (including the one-third addition), and cannot really be used to modify the latter.53 If, however, the Havana commissioner overestimated imports during 1860 to the same extent as in the rest of the period 1858 to 1865, then actual imports were probably much closer to 18,671 than 24,895, and a figure of 21,000 is arbitrarily selected.

For the remainder of the period, the Havana commissioners’ count of actual slave landings probably came close to being a complete record of imports, and we can ignore their addition of one-third to allow for unknowns. From 1861 on, the African blockade was tightened and information accordingly become more reliable. It seems unlikely that more than one or two ships a year escaped without the knowledge of the naval and consular network on the African coast. For 1861, we have recorded imports of 13,812 to set against recorded exports from Africa of 13,000. In 1862, the equivalent figures are 10,271 and 11,110, and in the following year, 3,799 and 2,610, although the Cape commissioners acknowledged that in 1863 shipments may have escaped from the east coast without information reaching them. In 1864, the Havana commissioner reported 2,353 imports as against 2,100 exports reported from the African side. Again, however, the west coast naval commodore acknowledged the possibility of two other shipments, details of which he could not communicate. In 1865, 745 definite arrivals in Cuba were reported, while from the African side the escape of 600 slaves was noted independently by the Luanda commissioner and the west coast commodore, and one other shipment may have left the east coast. In 1866, one shipment was noted from both sides of the Atlantic, and a figure of 700 imports into Cuba seems reasonably secure. There can be no certainty as to whether the final transaction took place in 1866 or 1867. While reports from Africa indicate one shipment and those from Cuba suggest two arrivals in 1867 totaling 1,000 slaves, the dates of the reports from the two sides do not coincide. This could indicate three or (if the Cuban reports of arrivals were unfounded) zero expeditions, but the Havana figure of 1,000 is accepted here.34

The reiterated doubts during the final years of the trade of several different Havana commissioners in the completeness of their import estimates would thus seem to have had little basis in reality. After allowing for mortality and voyage time (which meant that annual import and export data would not be for exactly the same period) the coincidence for most years between the two sets of independent estimates is comforting. The possibility of a slave ship escaping both information networks was not very high. For the same reason, it is unlikely that imports continued into the 1870s as Juan Pérez de la Riva has suggested.55 The east coast from which such imports were supposed to have originated was in fact blockaded along with the west coast for several years after 1867. Moreover, emancipation in the United States and the Moret law of 1870 may have reduced expected returns from all slaves.

Finally, it should be noted that none of the Cuban series allows for reexports, of which there were two kinds. After December 1832, the Havana Court of Mixed Commission emancipated several thousand new arrivals and shipped some of them to British Caribbean colonies, mainly Trinidad and Demerara. A second branch of the trade was reported operating sporadically between Cuba and Texas from 1835 to the 1840s. While numbers cannot be estimated, slave prices in Cuba were said to have dropped sharply when General Santa Anna proclaimed the abolition of Texan slavery in 1836.56

As in the pre-1821 period, Cuban population data may be used to provide a rough check on the accuracy of the import estimates. Solving Equation 1 for the rate of natural change (r) gives

$r=P1−P0−M(1−S)M(1−S)2+P0$

Substitution of the slave population data from the censuses of 1827 (P0) and 1846 (P1), among the more reliable of the nineteenth-century Cuban population counts,57 while using the import data for 1826-47 (M) from Table II and accepting a “seasoning” mortality rate of o.1, yields an annual rate of natural population change of −2.1 percent. This of course assumes an insignificant number of manumissions and re-exports, which for this period appears plausible. Such a rate is consistent with the experience of other sugar colonies in the Caribbean which imported large numbers of slaves relative to the size of their slave populations.

The same calculation for 1846-61 using the census of 1861 and again slave imports from Table II for 1846-61 yields an annual rate of natural population decrease of 1.5 percent. A reduction in the rate from the previous period is plausible if we keep in mind first, the experience of other sugar colonies and second, the low level of slave imports in this period both in absolute terms and relative to the total slave population. The distorting effects of the slave trade on the Cuban population pyramid and thus the rate of natural population change were likely to have been reduced in the 15 years or so before imports became significant once more in the late 1850s.

## The French Americas

Imports into the Americas other than Brazil and Cuba are underrepresented in the Foreign Office records. Information on the French Caribbean is particularly sparse because the Foreign Office had no representatives in any of the French American possessions, although it is clear that imports during wartime and after 1831 were small. Because of this lack of hard data, the census approach becomes the primary method of calculating imports instead of just a test for the accuracy of import figures.

Using Moreau de Jonnès’s population figures, Curtin found that the rate of natural decrease for Martinique, Guadeloupe, and French Guiana in the years immediately after the slave trade was finally suppressed around 1831-32 was 1.3, 1.1, and 2.3 percent respectively.58 Substituting these rates, the appropriate slave population data, and a seasoning rate of 0.1 into Equation 1 yields imports of 24,250 into Martinique for 1816-31, 37,547 for Guadeloupe for 1816-32, and 14,500 for French Guiana for 1814-31. The differences in the time periods here are explained by the availability of the census data. For Martinique and Guadeloupe, estimates for 1814 and 1815 are necessary (imports into Guiana and Martinique in 1831-32 were small enough to be ignored), and based on the shipping data discussed below, the totals are estimated at 680 and 1,040 respectively. This brings the totals for Martinique and Guadeloupe for 181.4-32 to 24,930 and 38,600, respectively, and for the French Caribbean as a whole to 78,00°59 or about 4,600 per year although, as we shall see, most of these arrived in the first half of the 1820s.

An exhaustive set of data on suspected French slave ships collected by Serge Daget provides broad support for these estimates, and permits an annual breakdown of the totals. Daget has incomplete information on 714 slave voyages in this period, only 50 of which can be found in the Foreign Office records.60 For most of these 714 the precise embarkation point is not known, but the relative completeness of the Foreign Office sample of voyages ending at Brazilian ports and Havana suggests that very few French ships could have gone to Brazil or anywhere except the outports of Cuba, Puerto Rico, or the French West Indies. Indeed, both the French sources and British naval reports indicate that French slave traders dominated the traffic to Santiago and other outports far removed from the knowledge of the French consul in Havana. Indications of intended destinations exist for about half of these suspected ships, and lend strength to this assessment. Of 355 ships, about half were going to the French West Indies and a third to Cuba, mainly Santiago. The rest were headed for the Mascarenes and, in seven cases only, Surinam. Applying these proportions to all 714 ships in Daget’s list yields 357 ships to the French West Indies and 238 to Cuba and Puerto Rico between 1814 and 1832. Not all these ships landed slaves; at least 40 were definitely prevented from disembarking their cargoes by the British or French authorities. On the other hand, Daget does not believe that his list includes all French slave ships in this period. The assumption that ships omitted from the list were equal to those that failed to complete their mission successfully, permits an estimate of imports independent of the population-based figures.

The derivation of mean cargo figures for the conversion of these shipping data into slave imports presents some difficulties. Daget’s averages of actual cargoes are 210 for the French West Indies (63 observed cases) and 250 for Cuba (46),61 means which are well below those recorded for ships arriving in other American regions at this time. While French slavers of this period, particularly those from the French West Indies, tended to be smaller than those from other jurisdictions, these means seem too far below average cargoes disembarked under, say, the Spanish flag (324.0 for 1821-30, 149 observed cases) and they probably reflect the inclusion of several cargoes which were incomplete at the time of capture. Perhaps mean cargoes of 250.0 for French ships arriving in the French West Indies and 300.0 for those going to Cuba would be more reasonable. The product of these and the shipping data yield imports of 89,250 for the French West Indies and 71,400 for Cuba in the years 1814 to 1832. Given the significant re-exports from Martinique to Puerto Rico noted by Scarano62 (perhaps 1,000 a year for most of the 1820s), and given too that imports by non-French slave ships into the French West Indies at this time would have been negligible, then it would appear that the first of these figures confirms the population-based import estimates for net slave arrivals in the French Americas. The second figure, for French imports into Cuba, does nothing to disturb the earlier estimates for the island, and in fact does much to fill in the smuggling void before 1821 and the blind spot in the British reporting system after that year. Perhaps one-third to one-half of the 71,000 arrived before 1821 and a further one-fifth were included in the British counts of arrivals after that year. This leaves about 20,000 to be distributed between Puerto Rico and the Cuban outports. Imports to the latter perhaps made up the difference between the arrivals recorded by the British and the total of actual Cuban imports between 1821 and 1831 (the difference between columns 2 and 3 in Table II in fact).

Such calculations are, of course, speculative and do nothing to replace the population data as the foundation of any estimate of slave imports into the French West Indies between 1814 and 1832. However, we can break down the aggregates for the period into annual imports if we assume that the annual distribution of ships in Daget’s list was similar to that of slave arrivals. Column 1 of Table III shows the annual percentage distribution of the 538 ships in the Daget list which were clearly not intending to ship slaves to Cuba or to the Mascarene Islands. Columns 2, 3, and 4 are the product of these percentages and the total imports calculated for each of the three French possessions from the population data.63Table III also shows arrivals of engagés in these territories between 1854 and 1862. Almost all of these were purchased as slaves and given their nominal freedom before embarkation but in no sense did they make a voluntary decision to go to the Americas.

## All Other American Regions

The remaining importing regions are grouped together in Table IV. They include the Dutch Americas, Puerto Rico, British colonies in the Americas, and Montevideo. The British judge at the Court of Mixed Commission at Paramaribo in Surinam was the most underworked of all the Foreign Office officials in the Americas. Slave imports from 1814 on were very few, although Daget has found seven ships intending to disembark there. The Dutch islands were never major importing regions. Slaves arrived at Puerto Rico down to the early 1840s64 although recorded arrivals are few. On the other hand, reports from the Bahamas, Jamaica, Dominica and other British-held territories probably included all arrivals. British cruisers or bad weather brought more than 20,000 slaves to these areas, mainly to the Bahamas during the 1830s. Except for a few hundred, none of these were sent on to Sierra Leone, and none were en route to the United States as Curtin hypothesized.65 Slaves captured by U.S. cruisers and taken to ports in the United States have not been included here because their stay in the United States was usually brief. They were quickly subjected to a second transatlantic crossing before disembarking in Liberia. Montevideo received a few thousand imports in the mid-1830s. For all regions except Puerto Rico between 1811 and 1843 and the United States between 1811 and 1822, the assumption is made that reported arrivals equaled actual arrivals. For Puerto Rico, there were probably few slaves before 1830 who arrived direct from Africa, but as the estimates for the French islands, the other region of provenance for Puerto Rico, are net of re-exports, a separate estimate for the Spanish island is necessary. Imports into Puerto Rico are set at 10 percent of Cuba’s following the ratio of the islands slave population to that of Cuba in this period. Slave imports between 1811 and 1845 are thus taken as 46,870, and these are distributed over the period 1811 to 1843 according to Scarano’s distribution of imports into the sugar-growing municipality of Ponce.66Table IV adds the results of this calculation to imports into the other areas as observed by Foreign Office representatives. Finally, there are no reports of slaves being imported directly into the United States from Africa in this period, although before the Piracy Act was passed, there were probably some. One thousand has been added to the import estimates for each year from 1811 to 1822 in Table IV to cover such arrivals, and there was a documented arrival of 338 in Georgia in 1858.67

## Aggregate Imports into the Americas

We can now address the issue of aggregate imports. Table V shows imports of slaves into all American regions from Africa by decade from 1811 to 1870. The decadal figures are summarized from Tables I-IV. Both the Cuban and Brazilian estimates for these years are larger than those previously made. In the Brazilian case, Leslie Bethell cites estimates by Foreign Office representatives in Brazil for the post-1830 period which are frequently lower than those given in the third series and indeed the first and second series for some years.68 Bethell, however, did not go behind the estimates and compare the shipping lists which these officials often returned. As long as the consuls and commissioners used different sources of information, one was always likely to discover landings unknown to the others. Thus for years such as 1839 and 1840, the Foreign Office clerks consolidated the different Rio lists and obtained annual figures greater than the estimates of any one official. In the Cuban case, only Murray has made a modern attempt at annual figures for the post-1821 period, and he tends to underestimate average cargo sizes and the number of slaves landed in Cuba outside Havana.69

Given the foregoing comments, it is not surprising that the import series for the Americas as a whole in Table V exceeds Curtin’s estimates for the same period by 23 percent. The current figures are in fact larger than any made for the 1811 to 1867 period apart from the series which James Bandinel, a former head of the Slave Trade Department of the Foreign Office, presented to the British House of Commons select committee on the slave trade in 1848, and the estimates of Thomas Fowell Buxton, the British abolitionist leader.70 The aggregate estimates for the whole period of the trade are, however, much less affected by these new data. To Curtin’s original aggregate of 9.6 million slaves imported into the Americas and the Atlantic basin during the whole four and a half centuries, the new figures represent an increase of almost 9 percent. Accepting Lovejoy’s reworking of Curtins pre-1781 figure, however, the new total is only 5 percent greater than the 1982 “Synthesis” figure.71 In the light of the high figures that were accepted as the likely volume of the trade in the pre-Census era, this essay must be taken as confirming rather than undermining Curtins projections.

Yet the major purpose of the work presented here is not to revise aggregate totals, but rather to develop more refined data in the form of annual estimates. What, if anything, do these new data tell us about the societies which received the slaves? Space constraints prevent an adequate answer to this question, and indeed the issue has been taken up elsewhere. However, one or two salient points need to be stressed. In the first place, there was a weak correlation between annual as opposed to decadal slave arrivals and plantation output—whether of coffee or sugar—in any of the importing societies. For example, annual slave imports into south central Brazil climbed steadily during the 1830s, yet remained below those of the 1820s, while coffee exports almost tripled between 1826-30 and 1836-40. In Cuba, annual sugar output nearly doubled during the 1840s, at the very time that annual slave arrivals followed a strong downward trend.72 Except for the temporary distorting effects of war, there are no similar anomalies in earlier centuries: produce exports and slave arrivals tended to move in tandem. The explanation for the nineteenth-century trend is of course suppression. Fluctuations in slave arrivals were influenced by the activities of the British antislave-trade naval squadron, and, more importantly, the attitudes of the authorities of the slave importing zone—whether in Brazil, Cuba, or the French Americas. The behavior of slave prices and the distribution of the slave labor force among major crops makes this clear. Whenever suppression occurred, prices of slaves increased and slaves became more concentrated in the sector producing the most valuable export crop. Indeed a major domestic slave trade developed in Brazil and the southern United States in the wake of suppression of the African traffic.

Regions which could attract alternative sources of labor as the price of slaves increased were able to reduce the impact of suppression. In Cuba, the arrival of indentured Chinese labor at midcentury occurred after the first serious domestic restrictions on arrivals of slaves from Africa and a strong rise in slave prices. In Brazil, attempts to attract European migrants intensified after the ending of the transatlantic slave trade, but the major source of additional labor in the southern Brazilian coffee sector was neither Europe nor China, but other regions of Brazil itself. Restrictions on slave imports and higher slave prices forced the cotton growers of the northeast to switch to free labor. Cotton exports stagnated from the early 1820s. On the other hand, the sugar-producing sectors of Bahia and Pernambuco did very well as long as the African traffic remained open. Sugar exports tripled between 1821-25 and 1851-55 as we might expect from the import series developed above. The decline of northeastern sugar production dates only from the 1850s, and even then “decline” must be defined as failure to expand rather than in terms of absolute decreases in sugar exports. In relative terms, the coffee fazendeiros could cope best with the post-1850 labor situation. But the major point is that the switch to other forms of labor was always reluctant. Clearly, suppression of the slave trade—like the later suppression of slavery itself—was imposed. In no sense did the slave trade stop because slaves were no longer wanted in the Americas. The potential for slave imports into the nineteenth-century Americas was much greater than in earlier centuries, and the effect of restricting the flow was to reduce the income-earning potential of Brazil and Cuba at least in the short run. The annual series presented here should allow a further exploration of this somewhat disturbing implication, as well as throw light on other salient trends in the nineteenth-century economic history of the slave-importing Americas.

1

Basic statistical material for nineteenth-century Brazil may be found in Instituto Brasileiro de Geografía e Estatística, Anuário estatístico do Brasil, Ano III (Rio de Janeiro, 1937); ibid., Ano V (Rio de Janeiro, 1939-40); Mircea Bueseu, 300 anos de inflação (Rio de Janeiro, 1973), in particular p. 223; Robert W. Sienes, “The Demography and Economics of Brazilian Slavery: 1850-1888” (Ph.D. diss., Stanford University, 1976); Nathaniel H. Leff, Underdevelopment and Development in Brazil, 2 vols. (New York, 1982), I, Economic Structure and Change, 1822-1947. For slave prices, see Pedro Carvalho de Mello, “The Economics of Labor in Brazilian Coffee Plantations, 1850-1888” (Ph.D. diss., University of Chicago, 1977); Kátia M. de Queirós Mattoso, Herbert S. Klein, and Stanley L. Engerman, “Trends and Patterns in the Prices of Manumitted Slaves: Bahia, 1819-1888,” Slavery and Abolition, 7:1 (May 1986), 59-67. For Cuba, see Manuel Moreno Fraginals, El ingenio: Complejo económico social cubano del azúcar, 3 vols. (Havana, 1978), in particular III; idem, Klein and Engerman, “The Level and Structure of Slave Prices on Cuban Plantations in the Mid-Nineteenth Century: Some Comparative Perspectives,” American Historical Review, 88:5 (Dec. 1983), 1201-1218; Kenneth F. Kiple, Blacks in Colonial Cuba, 1774-1899 (Gainesville, 1976). Summaries of research on the volume of the nineteenth-century slave trade may be found in Philip D. Curtin, The Atlantic Slave Trade: A Census (Madison, 1969); Paul E. Lovejoy, “The Volume of the Atlantic Slave Trade: A Synthesis,” Journal of African History, 23:4 (1982), 473-501.

2

Leslie M. Bethell, The Abolition of the Brazilian Slave Trade; Britain, Brazil and the Slave Trade Question, 1807-1869 (Cambridge, 1970), 388-390.

3

David Eltis, Economic Growth and the Ending of the Transatlantic Slave Trade (New York, 1987), chaps. 1 and 2.

4

See Bethell, Abolition of the Brazilian Slave Trade, 1–73 and David R. Murray, Odious Commerce: Britain, Spain and the Abolition of the Cuban Slave Trade (Cambridge, MA 1980), 50-84 for a review of these measures.

5

For the operation of the Courts of Mixed Commission, see Bethell, “The Mixed Commissions for the Suppression of the Transatlantic Slave Trade in the Nineteenth Century,” Journal of African History, 7:1 (1966), 79-93. For the British antislave-trade network, see Eltis, Economic Growth, chap. 6.

6

See Eltis, “The Export of Slaves from Africa, 1821-1843,” Journal of Economic History, 37:2 (June 1977), 409-433; “The Direction and Fluctuation of the Transatlantic Slave Trade, 1821-43, ” 273-301 for a presentation of an earlier version of these data.

7

Great Britain, Parliamentary Papers (hereafter PP), 1845, LVIX, 593-633. The complete data set on which this study is based (slave ship records) is on deposit in machine-readable form at the National Archives, Ottawa, Canada.

8

For descriptions of and references to these data, see Klein and Engerman, “Shipping Patterns and Mortality in the African Slave Trade to Bio de Janeiro, 1825-1830,” Cahiers (d’Études Africaines, 59 (1975), 381-398; Joseph C. Miller, Way of Death: Merchant Capitalism and the Angolan Slave Trade, 1730-1830 (forthcoming); Serge Daget, “Long cours et négriers nantais du trafic illégal (1814-1831),” Revue Française d’Histoire d’Outre Mer, 62:226-227 (1975), 90-134; “La navigation nantaise pour le ‘commerce légitime’ à la côte occidental d’Afrique, 1833-1872,” Études et Documents, 6 (Nantes, 1981), 87-88.

9

Robert Walsh published a contemporary series of imports for Rio de Janeiro 1820–29, Notices of Brazil in 1828–1829 (London, 1830), 322, which is about 10 percent higher than the one given here. His source is not apparent, however, and he provided no details of individual arrivals.

10

Klein, The Middle Passage: Comparative Studies in the Atlantic Slave Trade (Princeton, 1978), 57, 73-94.

11

Miller, “Legal Portuguese Slaving from Angola. Some Preliminary Indications of Volume and Direction, 1760-1830,” Revue Française d’Histoire d’Outre Mer, 62:226-227 (1975), 151.

12

Admiralty to Palmerston, Dec. 12, 1831, sub. enc. Commander Harrison to Admiral Hayes, Sept. 23, 1831, British Public Record Office, Foreign Office, Slave Trade Series (hereafter FO 84), vol. 126. Cf. Robert Conrad, “The Contraband Slave Trade to Brazil, 1831-1845,” HAHR, 49:4 (Nov. 1969), 618-621.

13

W. G. Ouseley, Rio de Janeiro, to Palmerston, June 5, 1833, private, Broadlands mss. (National Register of Archives, London), EC/OU/26.

14

G. Jackson and F. Grigg to Palmerston, Dec. 27, 1833, FO 84/138. See also the Brazilian Foreign Ministry report of 1833 cited in Conrad, “Contraband Slave Trade to Brazil,” 622.

15

Jackson and Grigg to Palmerston, Mar. 23, 1835, FO 84/174.

16

Jackson and Grigg to Palmerston, Sept. 30, 1836, FO 84/199.

17

Ibid.

18

Henry S. P. Eyre to his father, Aug. 6, 1836, Rhodes House Library, Oxford, Buxton papers, 29/fol. 168.

19

Gordon to Palmerston, Feb. 28, 1838, FO 84/252; J. Samo and Grigg to Palmerston, July 17, 1843, enc. “Report on 1838-42,” FO 84/454; Samo and Grigg to Palmerston, Feb. 20, 1844, enc. “Report on 1843,” FO 84/510.

20

See, for example, Bethell’s survey of the reports of individual officials in Rio, Abolition of the Brazilian Slave Trade, 391-395.

21

Hesketh and Grigg to Aberdeen, Mar. 21, 1845, FO 84/563; PP 1852, LV, 337.

22

Full references for each of the estimates for the closing years of the southern Brazil traffic may be found in Bethell, Abolition of the Brazilian Slave Trade, 393-395.

23

Thus for 1846 and 1847, the Rio officials sent no lists of individual shipments to London, whereas in other years such as 1839 and 1840, the Foreign Office received two different listings for each year.

24

Francisco Marques Goes Calmon, Ensaio de retrospecto 0 commercio e a vida econômica e commercial da Bahia de 1823 a 1900 (Bahia, 1925), cited in Amaral, História geral da agricultura brasileira, 3 vols. (São Paulo, 1939-40), I, 324.

25

Calmon’s data are accepted for 1814, 1818-21, and 1826-28.

26

H. S. Fox to Palmerston, Nov. 8, 1835, enc. W. G. Ouseley, “Notes on the subject of the slave trade in the province and city of Bahia, Sept. 1835, FO 84/179.

27

E. Porter to Aberdeen, July 30, 1843, FO 84/470.

28

Full references for the Bahia estimates for 1844-51 may be found in Bethell, Abolition of the Brazilian Slave Trade, 393-395. For some years, the count of slave and ship arrivals exceeds the annual estimates of the British consul, although the only major discrepancy is for 1849 where the latter reported 8,081 arrivals as against 9,921 derived from the consular and naval correspondence during the year. As the details of arrivals are such as to reduce the possibility of double counting, the higher figure is taken as the preferred series.

29

Admiralty to Palmerston, May 25, 1837, FO 84/228.

30

B. Newcomen to Palmerston, July 24, 1843, FO 84/470; H. Hamilton to Palmerston, Dec. 24, 1841, FO 84/367.

31

R. Hesketh to Palmerston, Aug. 3, 1831, enc. “Report of the State of the Slave Trade in the Northern Provinces of Brazil,” FO 84/122.

32

Ibid.; John Moon to Palmerston, May 27, 1837, FO 84/223; William Wilson to Palmerston, June 20, Oct. 5, 1839, FO 84/289; Vice-Admiral Colpoys to Admiralty, May 22, 1832, Public Record Office, Admiralty (hereafter Adm.) 1/228.

33

R. Ryan to Palmerston, Nov. 11, 1843; Newcomen to Palmerston, July 24, 1843, FO 84/470.

34

Edmundo Armenio Correia Lopes, A escravatura (subsidios para a sua historia) (Lisbon, 1944), 139-142.

35

Alexander von Humboldt, Ensayo político sobre la isla de Cuba (Havana, 1960 edition), p. 191; José Antonio Saco, “Análisis de una obra sobre el Brasil,” in Colección de papeles científicos, históricos, políticos y de otros ramos sobre la isla de Cuba, 3 vols. (Havana, 1960), II, 74-75; Klein, The Middle Passage, 214.

36

See the discussion in Murray, Odious Commerce, 16-19.

37

Below,

38

See David Galenson, White Servitude in Colonial America: An Economic Analysis (New York, 1981), 212-213, and Henry A. Gemery, “Emigration from the British Isles to the New World, 1630-1700: Inferences from Colonial Populations,” Research in Economic History, 5 (1980), 170-231 for the derivation of this formula and the “seasoning” death rate. The formula used by Curtin makes no separate allowance for seasoning deaths and in addition uses the population present at the start of the period as a base against which to calculate rates of migration and crude rates of population change (The Atlantic Slave Trade, 31-33). His implicit assumption is that all those who died between census counts were present at the start of the period, whereas in fact some who died were bozales who had already undergone seasoning. The Gemery formula makes a crude adjustment for these by calculating the rate of change against the average of starting and finishing populations. Rates of change thus come out lower than those calculated by Curtin. For the census data used here for 1792 and 1817, see Kiple, Blacks in Colonial Cuba, 28-38.

39

See below. Fragmentary archival sources for slave arrivals at Santiago de Cuba for eight months during 1789-90, and almost the whole of the year 1794, show import levels comparable to those of Havana (Klein, “North American Competition and the Characteristics of the African Slave Trade to Cuba, 1790 to 1794, ” William and Mary Quarterly, 28:1 [Jan. 1971], 91, n. 15). The relative development of the hinterlands of the two ports indicates that parity of slave imports did not, however, persist for very long. Nevertheless an adjustment of the official figures is clearly necessary.

40

J. Hardy to Palmerston, Mar. 1, 1833, FO 72/415; J. Kennedy and Campbell J. Dalrymple to Palmerston, Jan. 1, 1842, FO 84/395.

41

Hardy to Palmerston, Mar. 1, 1833, FO 72/415, Apr. 1, 1834, FO 72/431, Apr. 15, 1836, FO 72/468; Clarke to Palmerston, July 10, 1835, FO 72/449.

42

C. D. Tolmé to Palmerston, Sept. 17, 1839, FO 84/280; cf. Kennedy and Dalrymple to Palmerston, Jan. 1, 1841, FO 84/348.

43

The actual numbers for the years 1819 to 1837 are 40 to the outports and 98 to Havana. These 138 ships are about 15 percent of all ships which arrived in Africa between these years with the intention of shipping slaves to Cuba.

44

See Murray, Odious Commerce, 243-248 and “Statistics of the Slave Trade to Cuba, 1790-1867,” Journal of Latin American Studies, 3:2 (Nov. 1971), 139-148 for a discussion of the British estimates of the Cuban traffic in this period.

45

The estimate for 1854 has been revised upwards despite the opinion contained in an internal Foreign Office memo that the original total probably included all landings. The evidence of extra landings seems incontrovertible. (Memorandum by T. L. Ward, Jan. 20, 1855 attached to Backhouse to Clarendon, Jan. 1, 1855, FO 84/959.)

46

Kennedy to Palmerston, Jan. 1, 1850, FO 84/789, Jan. 1, 1851, FO 84/832.

47

Joseph T. Crawford to Malmesbury, Jan. 1, 1853; to Russell, June 3, 1853, FO 84/899.

48

Admiralty to Clarendon, Sept. 14, 1857, enc. Commander Hope, Feb. 17, May 23, 1857, FO 84/1040; George Jackson and Edmund Gabriel to Clarendon, Aug. 18, 1856, FO 84/985. Crawford and F. Lousada to Clarendon, Jan. 31, 1857, FO 84/1012.

49

Commodore C. Wise to Rear-Admiral Grey, Aug. 6, 1858, Adm. 123/175; Frere and Surtees to Clarendon, Jan. 20, 1858, FO 84/1044; Crawford and Lousada to Clarendon, Jan. 26, 1858, FO 84/1042.

50

G. Skelton to Malmesbury, Mar. 2, 1859, FO 84/1072; Commodore Wise to Rear-Admiral Grey, Jan. 20, 1859, Adm. 123/175; Frere and Surtees to Malmesbury, Jan. 23, 1859, FO 84/1078; Crawford to Malmesbury, Jan. 25, 1859, FO 84/1073.

51

The Luanda commissioner wrote of 17 shipments and 12,750 slaves in the first eight months (Gabriel to Russell, Sept. 20, 1859, FO 84/1076). Commodore Wise claimed 17 shipments in the first 9 months with 11,000 slaves and thought 20,000 reasonable for the year (to Rear-Admiral Grey, Oct. 22, 1859, Adm. 123/175). His successor noted 7 successful departures in the last three months (Commodore Edmonstone to Rear-Admiral Walker, July 25, 1860, Adm. 123/180). If to these we add 4 shipments from the east coast, allowing 750 slaves a shipment (Frere and Surtees to Russell, Jan. 16, 1860, FO 84/1170), the total known exports would amount to about 20,000. The Havana commissioners’ estimates are in Crawford and Ryder to Russell, Dec. 31, 1859, FO 84/1073.

52

Between Oct. 1, 1859 and Dec. 31, 1860, the West African squadron met with only 47 out of 90 ships supposed by the Foreign Office to be engaged in the traffic (Edmonstone to Walker, July 25, 1860, Mar. 24, 1861, Adm. 123/180). While a similar ratio emerged in 1861 and 1862, the commodore was able to state confidently that most of the suspected ships had never arrived off Africa. Though the size of the squadron was not reduced in 1860, Edmonstone’s complaints in the above dispatches indicate that fewer ships were available for blockading duties relative to the size of the trade.

53

The export figure for West Africa is derived from the dispatch cited in the previous note. While the Cape commissioners reported the slave trade from East Africa with Cuba as being carried on to “a great extent” (Frere and Surtees, Sept. 29, 1860, FO 84/1107) the only precise estimate is three shipments in the last six months of 1860 (Admiralty to Russell, Mar. 8, 1861, enc. Commander Oldfield, Dec. 31, 1860, FO/1148). The 10,450 is thus made up of 8,200 from the west coast and 3 × 750 or 2,250 from the east coast. The import estimates are in Crawford to Russell, Feb. 5, 1861, FO 84/1135.

54

For exports, see Commodore Edmonstone to Rear-Admiral Walker, Nov. 7, 1861, Oct. 22, 1862, Adm. 123/180; Admiralty to Russell, Apr. 29, 1864, enc. Commodore Wilmot, Dec. 31, 1863, FO 84/1228; Admiralty to Russell, Jan. 1, 1865, enc. Commodore Wilmot, Dec. 1, 1864, FO 84/1252; Admiralty to Clarendon, Jan. 12, 1866, enc. Commodore Wilmot, Dec. 19, 1865, FO 84/1267.

55

Juan Pérez de la Riva, “Cuando llegaron a Cuba los últimos bozales?” Revista de la Biblioteca Nacional José Martí, 16:2(1974), 174-179.

56

E. Schenley to Palmerston, June 22, 1936, FO 84/195.

57

Kiple, Blacks in Colonial Cuba, 39-58.

58

Curtin, The Atlantic Slave Trade, 81-82.

59

The percentage breakdown of imports into the French West Indies by colony produced by this method is similar to the percentage breakdown of suspected French slave ships by intended Caribbean destination drawn from Professor Daget’s data discussed below. The total of 78,000 is not much different from Curtins figure despite the separate allowance for seasoning deaths provided for by the formula used here (The Atlantic Slave Trade, 78-84, 234).

60

Much of the hard information in the following discussion, though not the use to which it is put, is based on a personal communication from Serge Daget. I am indebted to Daget for the opportunity to examine his data before the completion of his doctorat d’état. A partial list of these ships has appeared in Gemery and Hogendorn, Uncommon Market, 299-301.

61

Personal communication from Daget.

62

Francisco A. Scarano, Sugar and Slavery in Puerto Rico: The Plantation Economy of Ponce, 1800-1850 (Madison, 1984); 127-130.

63

The Fourteenth Report of the Directors of the African Institution (London, 1820), 70–72 reported 17 cargoes disembarked at Guadeloupe in 1819 and early 1820. Most of these were small. The data for engagés is taken from François Renault, Libération d’esclaves et nouvelle servitude (Abidjan, 1976), 175-177, 189.

64

For the Dutch trade, see Pieter C. Emmer, “Surinam and the Decline of the Dutch Slave Trade,” Revue Française d’Histoire d’Outre Mer, 62:226-227 (1975), 245-251 and “The Abolition of the Abolished: The Illegal Dutch Slave Trade and the Mixed Courts” in Eltis and Walvin, The Abolition of the Atlantic Slave Trade (Madison, 1981), 177-190. For the small number of arrivals in Puerto Rico after 1843, see John Lindegren to Aberdeen, Sept. 11, 1844, FO 84/520; to Palmerston, June 11, 1845, FO 84/578, Mar. 27, 1847, FO 84/674; Charles Lindegren to Clarendon, May 26, 1856, FO 84/988.

65

Curtin, The Atlantic Slave Trade, 236.

66

Scarano, Sugar and Slavery in Puerto Rico, 136-137. The 46,870 figure for the whole island may be on the high side. Scarano estimates that imports into Ponce were in the 4-6,000 range in the nineteenth century, and that Ponce produced about one-fifth of the island’s sugar output (ibid, xxiv, 137). Slaves were also used, though less extensively, in coffee cultivation. Until the mid-1830s, coffee exports were comparable in value to sugar exports.

67

These are the only slaves which the present study accepts as being imported into the United States after 1822. The case is discussed in Warren S. Howard, American Slavers and the Federal Law, 1837-1862 (Westport, CT, 1963), 145-146. There may well have been a few hundred others after 1822, but the silence of the British reporting system on the subject tends to support the conclusions of Howard and Kiple (The Case Against a Nineteenth Century Cuba-Florida Slave Trade,” Florida Historical Quarterly, 49:4 [Apr. 1971], 346-355) among others on the insignificance of slave imports into the United States after 1820. The Spanish-agent-turned-British-informer who supplied the British consul in New York with detailed data on slave ship movements wrote in 1859, “I question if besides the ‘Wanderer’s’ cargo more than one or at most two cargoes have been landed . . . The newspaper accounts of arrivals are due to political motives . . .” He did refer to expected arrivals in Texas, however (E. M. Archibald to Russell, Oct. 4, 1859 [ene.], FO 84/1086). The real counter to W. E. B. DuBois’s estimate of 250,000 arrivals after 1808 in his The Suppression of the African Slave-Trade to the United States of America, 1638-1870 (New York, 1896) is the absence of evidence of African-born individuals in any of the postbellum census material, Bureau of Freedom records, and the extensive W.P.A. Writers’ Project interviews.

68

Bethell, The Abolition of the Brazilian Slave Trade, 388-395.

69

Murray, “Statistics of the Slave Trade to Cuba, 1790-1867, 131-149.

70

Sir Thomas Fowell Buxton, The African Slave Trade and Its Remedy (London, 1839—40), 195-202. For the estimates of James Bandinel, see PP 1847-488, XXII, 8.

71

Curtin, The Atlantic Slave Trade, 216, 234. See Lovejoy, “Volume of the Atlantic Slave Trade,” for a comprehensive revision and discussion of the work of Curtin and subsequent scholars.

72

For a fuller presentation of the discussion which follows and the statistics on which it is based, see Eltis, Economic Growth, chap. 11 and Appendix F.

## Author notes

*

The author would like to thank Serge Daget, Stanley Engerman, Herbert S. Klein, Paul Lovejoy, Joseph Miller, Robert Paquette, and three anonymous referees for comments on earlier drafts of this essay. Parts of it appeared in David Eltis, “The Direction and Fluctuation of the Transatlantic Slave Trade, 1821–43: A Revision of the 1845 Parliamentary Paper,” in The Uncommon Market: Essays in the Economic History of the Atlantic Slave Trade, Henry A. Gemeryand Jan S. Hogendorn, eds. (New York, 1979), 273–301. I wish to thank Academic Press for permission to reprint these insertions here.