I wish to thank Michael Edelstein, Stanley Engerman, Nathaniel Leff and Philip Curtin for their invaluable comments and suggestions.

¹ Curtin has estimated Portuguese slave exports from the West African coast at 611,000 between 1701 and 1800. Portuguese exports from Angola during this same period, were over double that, or 1,414,500, thus representing 70% of the total slaves shipped. As for Angola's relationship to the total volume of the slave trade, its 1.4 million represented 26% of total African exports of all nations in the 18th century. These estimates are taken from Curtin, Philip D., The Atlantic Slave Trade, A Census (Madison: University of Wisconsin Press, 1969), p. 211, table 63Google Scholar.

² While the port of Luanda was not colonized by the Portuguese until 1575, slave trading in the region was already a fully developed system in the first half of the century. With effective occupation, and the keeping of more exact records, as well as formal contracts, it would appear that exports for the next century may have gone from 5,000 to 10,000 per annum, though the reliability of these estimates is questionable. Lopes, Edmundo Correia, A Escravatura (subsidios para a sua historia) (Lisbon: agência geral das colónias, 1944), pp. 85–87Google Scholar.

³ While port registers for either city are unavailable for either Benguela or Luanda in any number in the 1790's (see Appendix Tables I and II) a rough idea of exports can be obtained from Rio de Janeiro port registers. Since Rio de Janeiro accounted for over half of the exports in this period, its figures can be considered a reasonable reflection of trade conditions. Thus in the period 1795–1800, Benguela exports to Rio de Janeiro accounted for 38,990 slaves out of a total of 90,329 shipped from Angola to that port, which represented 43% of the total exports. See Klein, Herbert S. “The Trade in African Slaves to Rio de Janeiro, 1795–1811; estimates of mortality and patterns of voyages,” Journal of African History, X, no. 4 (1969), p. 545Google Scholar. It would seem that by the first decade of the 19th century, Benguela probably came close to supplying half of the total exports from the region, since its total share of the imports into Rio de Janeiro from 1795 to 1811 was 50%; ibid., p. 540.

⁴ The Portuguese tried to settle Cabinda permanently several times, and even succeeded in establishing a fort there in 1783, but this was destroyed by the French a year later. For the details of the Portuguese expedition of the 1780's, see the initial exploration report of Antonio Maximo de Souza Marques in Arquivo Historico Ultramarino [Hereafter cited as AHU], Angola, caixa 37, report dated 16 March 1780. On the setting up of the fort, see the letters of the Governor at Luanda to the Crown in 1783 in AHU, Angola, cx. 38, letters dated 10 and 11 August 1783. This apparently was not the first such settlement attempt by the Portuguese, according to Grandpré, and the French when they destroyed the fort built in 1783 did not replace it with one of their own. Grandpré, Luis O'HierVoyage à la côte occidental d'Afrique fait dans les annees 1786 et 1787 (2 vols.; Paris: Dentu, 1801), I, 31–32Google Scholar. By the second decade of the 19th century, however, the Portuguese must have more firmly established themselves at Cabinda, for in both 1814–1815, and 1817–1818 the port registers of Luanda list a sizeable number of vessels which temporarily stopped at Luanda on their way from Brazil to Cabinda. The figure was 9 ships in 1814, 3 in 1815, 4 in 1817 and 11 in 1818. AHU, Angola, cx. 62, 31 January 1815; and cx. 64, 31 January 1816; cx. 65, 14 January 1818; cx. 66, n.d. for 1818 report.

⁵ Klein, “The Trade in African Slaves”, p. 540. This was the average for the 162,498 slaves shipped from both Benguela arid Luanda in the 1795–1811 period.

⁶ One of the few systematic recordings of length of voyages was given by the British consuls in Rio de Janeiro. They reported on the voyage time of all slavers that arrived in the port of Rio de Janeiro from 2 January to 23 December 1821. This showed that the 22 ships that arrived from Luanda averaged 36.3 days in the crossing. The five ships that sailed from Benguela averaged 36.0 days, and finally the 7 ships that came from Cabinda averaged 35.8 days. Great Britain, Parliamentary Papers (1845), XLIX, “Slave Trade-Slave Vessels” report printed 25 February 1845, pp. 9–10. The major importing months at the end of the century for Rio de Janeiro were the southern equatorial seasons of spring and summer. Klein, “The Trade in African Slaves”, p. 539.

⁷ In the total slave exports for the eight years of 1812, 1815, 1817, 1822–26, for which surviving records provide ports of destination, Pernambuco, Maranh˜o, and Pará increased their share to 37% of exports. Rio de Janeiro continued to dominate with 54% of the total trade, with Bahia suffering a decline to only 6%. The rest was taken by southern Brazilian ports. These figures are taken from AHU, Angola, maço 16 and from Manuel Dos Anjos da Silva Rebelo, Relações entre Angola e Brasil, 1808–1830 (Lisbon: Agência-Geral do Ultramar, 1970), quadro no. 2, after p. 81. The Silva Rebelo table has several errors in it, including arithmetic ones, which I have corrected by referring to the originals of several years in the AHU. For 1822 he has the Ceará total incorrectly listed; it should be 465. In 1826 the original report lists no shipments to Maranh˜o and a year-end total of only 11,658. This makes the final total for his table 102,441, instead of 104,630, with 2 slaves shipped to Cabinda in 1822 and therefore not listed.

⁸ For the very special trade that developed between Bahia and the Gold and Slave Coasts, see the major study by Verger, PierreFlux et reflux de la traite des nègres entre le golfe de Bénin et Bahia de todos os santos, du dix-septième au dixneuvième siècle (Paris: Mouton, 1968)Google Scholar, especially chapters vi and following. In the 19th century, twice as many ships went to the Guinean coast from Bahia as went to the Angolan area, ibid., pp. 655–57. On the Dutch response to this special trade see Postma, Johannes “The Dutch Participation in the African Slave Trade: Slaving on the Guinea Coast, 1675–1795” (Ph.D. dissertation, Department of History, Michigan State University, 1970), pp. 90–92, 108, 124–26Google Scholar.

⁹ Between 1795–1811, 96% of the slaves exported from Africa to Rio de Janeiro came from the ports of Angola. Klein, “The Trade,” p. 540.

¹⁰ Ibid., p. 543.

¹¹ Average cargo capacity for the 166 sailings of the Curveta was 351 adults per vessel; for the Galera, with 137 sailings, the average was 457 adults. The Navio (translated in the 18th century as ship) was third in importance, with 75 sailings or an average of 427 adults per vessel. The Bergantim (or brig), however, only accounted for 17 sailings, or 3% of the total, and carried a low average of 288 adults per sailing.

¹² According to the 1684 decree, which was printed and is to be found in several collections of documents in the AHU, the rule was that (according to chapter vi) “If it is a decked ship in which there are portholes through which the Negroes can easily receive the necessary fresh air, then capacity below decks should be 7 adults (cabeças) for every two tons; not having said portholes, the capacity should be only 5 slaves per two tons below decks.” It also allowed that for either type of ship, they could carry another 5 slaves per ton above deck. The law also provided detailed listings of the type and quantity of food and water to be carried by the slaver in order to provide three meals per day per slave. Provisions for housing sick slaves apart from the rest and detailed registration by port officials for all of these provisions were also required. See Lei 3, 18 March 1684, in AHU, Angola, cx. 10. This ship's capacity law remained in effect throughout the 18th century, and was only changed by the decree (alvará) of November 24, 1813, which changed the tonnage to a uniform listing of 5 slaves per two tons. See the typical post-1813 slave ship's registration papers in AHU, Angola, maço 14 “Passaportes.”

¹³ It has been estimated that the average tonnage of all English shipping engaged in the Atlantic trade from Europe to North America in the mid-18th century was less than 150 tons per vessel. North, Douglass C. “Sources of Productivity Change in Ocean Shipping, 1600–1850,” The Journal of Political Economy, LXXVL 5 (Sept./ Oct. 1968), 958Google Scholar.

¹⁴ AHU, Angola, cx. 17, letter of Provedor de Fazenda Real [Superintendent of the Royal Treasury] of Luanda to the Crown dated 19 February 1728.

¹⁵ In some 56 expeditions financed by the Dutch West Indies Company between 1681 and 1751, the percentage of children under fifteen years of age was only 8%. Postma, “The Dutch Slave Trade,” pp. 177–78. Postma noted that West India Company officers advised slave captains to buy children only at the last minute before sailing and only if space was left. Postma, ibid., p. 179.

¹⁶ The records of the Companhia Geral de Pernambuco e Paraiba show that of the 49,344 slaves shipped by the Company from Africa between 1761 and 1786, only 508 were listed as children, or only 1%. These figures probably do not include infants at the breast (crias de peito), which would probably have accounted for another 250 or so children. Even so, this would only bring the total up to 2%. Carreira, AntónioAs companhias pombalinas de navegac˜o, comercio e tráfico de escravos entre a costa africana e o nordeste brasileiro (Porto: N.P. 1969), p. 261Google Scholar.

¹⁷ These figures are calculated from previously unanalyzed data in the Arquivo Nacional, Rio de Janeiro, Policia, Codice 242.

¹⁸ These accounts are found in Biblioteca National de Lisboa [Hereafter cited as BNL], Colecç˜o Pombalina, Codice 617.

¹⁹ AHU, Bahia, cx. 37, document dated 18 December 1727.

²⁰ BNL, Colecç˜o Pombalina, Codice 617, folio 222.

²¹ To give some idea of other costs to Captain Proença e Sylva, his records note that in October of 1759 he was charged for export taxes, maintenance, passage, branding and registration fees for shipping one slave to Recife the sum of 16$480, with 8$000 of that figure due to transportation charges (ibid., folio 174). This must have been a fairly standard set of charges, for another male adult slave he sent in February of that year to Recife, also cost him 16$480 to ship to his factor in that city. In addition, the slave apparently sickened on his arrival and medical costs added another 2$690 in expenses before he could be sold. The final sale price was 70$000, leaving a gross profit of 50$830 (ibid, folio 114). Also in 1759 he shipped to Pernambuco another 5 slaves. Of these, one died at sea and another two were shipped as his personal property to a relative in Lisbon. This left two young female slaves whom he sold for 120$000. His total costs for shipping the five slaves and maintaining them until their sale or shipment to Portugal was 78$955, which still left him with a gross profit of 41$045. The charge for maintaining a slave in Recife while waiting for shipment to Portugal was 60 reis per diem, (ibid, folio 116v). In 1761 and 1762 he shipped nine slaves to Rio de Janeiro, of which lot about five died. The rest sold at a total profit of 254$898 (ibid., folio 193). In this same account with his factor in Rio de Janeiro, it is noted that he has bought and sold letters of credit to Brazil over the past several years.

²² In one report to the Crown, (AHU, Angola, cx. 21, n.d.), Angolan officials noted that the total cost of 9 varieties of woolens and textiles, which would go to make up the cost of one slave, were priced in Lisbon at 26$400. These same types and quantities of goods in Luanda cost 48$300. Their cost at delivery point in the interior was 80$000. What is not clear from the table is whether this is the price they valued the goods at for the African purchaser, or whether this 80$000 represented the base cost at Luanda plus shipping costs to the interior. My own impression is that the former is more probable. If this is so, then taking 48$300 as the cost to the Luanda merchant of European goods used to purchase slaves, and another 16$480 as his costs for shipping him to Recife (see note 21), then the total costs would be 64$780, bringing his net profit on the sale of the slave in Recife to 5$220. This would give him a profit of only 8% on his investment for this slave. This would seem too small a return, especially given the fact that he would have to expect in normal operations the total loss of investment on several slaves. Also, using this same purchase price of 48$300 per slave for the seven slaves shipped by him to Rio de Janeiro in 1762 (Table 8), and assuming the three young children were adults and also sold at 90$000 (total gross income being 360$000), thus equalizing their sale prices with the assumed adult purchase price of 48$300 for each one, the net loss on this voyage was 109$111. On the basis of this very limited evidence, I assume that the report exaggerated the costs of the textiles in order to obtain benefits in reduced prices from the Crown.

²³ In 1795, for example, of the 337 Portuguese vessels which left the harbor of Lisbon (out of a total of 958 ships), only 14 went to Africa, as opposed to 51 that sailed for Brazil. At the same time, of the 348 Portuguese vessels that entered port in that year (out of a total of 1,036 ships), only two were listed as coming from Africa, and these two were from the Cape Verde Islands. At the same time, 98 were listed as arriving from Brazil. Arquivo Geral da Marinha (Lisbon), caixa “Entradas e saidas de Navios, Registro do Porto de Lisboa, 1741–1800”, in undated mappa. In the same collection of port registers, there is a listing of sailing times for all ships from port of departure. These show that the average length of voyage from Brazil was quite long. In the registers for 1797, for example, of four ships which came from Bahia in April to June of that year, the average sailing time was 81 days; the two ships coming from Recife, Pernambuco averaged 73 days; and the two ships coming from Rio de Janeiro to Lisbon averaged 109 days.

²⁴ Luanda in 1773 had a population of 612 troops and 1,519 civilians. Of the troops, 140 were Mulattoes and 72 were Negroes, while the civilian town population consisted of the following breakdowns:

These population statistics are found in AHU, Angola, cx. 36 documents numbered 19 and 25 and both dated 27 March 1773.

²⁵ In the wheat flour sold in Luanda in 1800 and 1801, for example, Brazilian imported sacks accounted for 22% and 19% respectively of all wheat flour sold, the rest being locally produced. AHU, Angola cx. 51, 1 April 1801 and cx. 53, 16 June 1802.

²⁶ In the mid-1780's, for example, the Angolans were importing 1,340 barrels of Gerebita per annum, with the bulk of this production (or 76%) coming from Rio de Janeiro. For three-year importation figures (1782–1784) see AHU, Angola, maço 13, 5 January 1785.

²⁷ Birmingham, Trade and Conflict, p. 138.

²⁸ Though evidence on this from Luanda is not presently available, data from Benguela would seem to support this assertion. Thus when the Curveta “Nossa Senhora de Agua de Lupe, e Bom JESUS dos Navigantes” sailed for Rio de Janeiro on 22 January 1763 it carried slaves which had been purchased on its account as far back as the preceding October, with the bulk being purchased in December. AHU, Angola, cx. 30, document no. 8.

²⁹ Birmingham, Trade and Conflict, pp. 138–39.

³⁰ For a typical example of such a tax contract, see the printed contract given to Jacinto Dias Braga for the collection of the novo imposto (a tax of l$200 per slave exported) in all the ports of Angola for a six year period (the typical term of all tax contracts) to begin on 5 January 1742. AHU, Angola, cx. 23, dated 7 September 1740.

³¹ These taxes were the direito real, direito novo, novo imposto and the preferencia.

³² For these decrees see Arquivo Nacional, Torre de Tombo, Manuscritos Miscellaneos, no. 926, folio 252.

³³ See, for example, the contract given to Domingos Dias de Silva and his associates for the collection of the new unified head tax which was granted for a six year term beginning on 5 January 1766. The cost to the tax farmer and his associates was the annual payment to the crown of the sum of 88:300$000. Given the estimated 12,000 slaves snipped per annum in the 1760's, this would give the tax farmer a gross income of 104:400$000, for a net profit of 16:100$000. A copy of the contract is in AHU, Angola, cx. 30, dated 6 September 1765. For an explanation of Portuguese currency in the 18th century, see the Note to Table 8.

³⁴ The founding charters of the two companies will be found in Carreira, As companhias pombalinas, appendix, documents “E” (pp. 313–336) and “H” (pp. 347–372). A contemporary copy of the original charter for the Maranh˜o Company can be found in Arquivo Geral da Alfandega de Lisboa, Codice 51, libro 1, folios 8–21v. A detailed study of all aspects of the Maranh˜o company, in which the slave trade assumed only a minor role, is found in Dias, Manuel Nunes “Fomento ultramarino e mercantilismo: a Companhia Geral do Gr˜o-Pará e Maranh˜o (1775–1788),” Revista de Historia (São Paulo) XXXII, no. 66 (Abril-Junho, 1966), 359–428Google Scholar; this is Part I of a study which extends over 10 parts in as many issues.

³⁵ Carreira, As companhias pombalinas, p. 261.

³⁶ Ibid., p. 91.

³⁷ The lower averages are possibly due to the shortness of the periods being considered, and also to the poor quality of recording. Too many ships had identical names, identical ship type identifications, and therefore were impossible to follow accurately in the subsequent listings. Since captains usually sailed for only one trip, it was impossible to distinguish several ships. An example of this was the listing under Corvetes with the name of “Nossa Senhora da Conceipcão, Santo Antonio e Almas.” Between 1762 and 1767 there were three different ships with that name and vessel type designation, all of which operated in Luanda. In other cases, when names were the same and ships' types were different, thus facilitating identification, later listings often changed the ship's designation, especially from the generic navio to a more precise term.

³⁸ Klein, “The Trade”, p. 544 and Verger, Flux et reflux, p. 658.

³⁹ Carreira, As companhias pombalinas, pp. 254, 261.

⁴⁰ Ibid., pp. 52–53.

⁴¹ In data reconstructed from the Maranhão Company records, Carreira has been able to determine the sex of some 20,141 slaves who were carried by Company vessels between 1756–1788, with the resulting breakdown being 7,572 females and 12,569 males, which means a 38% contingent of females. Ibid. pp. 94–95. Postma, in calculations made for 56 West India Company vessels that carried slaves from Africa in 1681–1751, found 8,629 females (or 29%) out of a total of 29,532 slaves who were transported. Postma, “The Dutch … Slave Trade,” pp. 177–178. He also notes that “as a rule, WIC [West India Company] captains had instructions to purchase a slave cargo consisting of two-thirds men and one-third women slaves.” Ibid, p. 179. Unger in his study of a Dutch free trading company found a rather high ratio of 41% (or 10,249) women, out of the total of 25,051 slaves shipped by the company vessels, Unger, “Bijdragen tot de geschiedenis …,” Finally, one of the few breakdowns by sex available in New World importation records can be found in the Havana port registers. Here between 1790 and 1794, 25,959 slaves were imported, of which 25% (or 6,535) were women. Klein, Herbert S. “North American Competition and the Characteristics of the African Slave Trade to Cuba, 1790 to 1794,” William & Mary Quarterly, 3rd series, XXVIII, 1 (Jan. 1971), 98, table VIGoogle Scholar. In the period from August 1815 to September 1818, (with the exception of a missing register for April 1817), some 61,851 slaves were introduced into the port of Havana. Of this number, only 19,988 (or 32% ) were women. These registers are to be found in the Archivo General de Indias (Seville), Audiencia de Santo Domingo, legajo 2207.

⁴² Postma notes that the Dutch slave trade was virtually wiped out by the fourth Anglo-Dutch war of 1780–1784 (Postma, “The Dutch … Slave Trade, pp. 160ff). Also, the French slave trade equally suffered a disastrous decline as a result of international conflicts. The trade was temporarily halted as a result of the Seven Years War, and then was totally suppressed as a result of the wars of the French Revolution which stopped French slavers from operating after 1792. For the impact of the Seven Years War see Gaston-Martin, Nantes au xviiie siècle: L'ère des negriers (1714–1774) (Paris: Félix Alcan, 1931), ch. viGoogle Scholar; and for the developments in the 1790's see Meyer, J. “Le commerce negrier nantais (1774–1792),” Annales: économies, sociétés, civilisations, XV (1960), 120–29CrossRefGoogle Scholar.

⁴³ A brief glimpse of Bahian investors in slave ships is given in the studies of Alves, Marieta “O comércio marítimo e alguns armadores do século xviii na Bahia,” Revista de Historia (São Paulo), XXXIV, 69 (1967), 91–98Google Scholar.

⁴⁴ By the early 19th century, average cargo size was rapidly going up in the exports from Angolan ports. Thus in the five year period from 1812, 1814 to 1817, average cargo size was 464 slaves per ship. Silva Rebelo, Angola e Brasil, quadro no. 1, after p. 81. As for total volume, this rose to an annual of 14,038 in the 13 years from 1812, 1814 to 1820 and 1822 to 1826. Ibid.

⁴⁵ Nevertheless, some 19th-century data give the impression of a rather thriving merchant community in the early years of the century. Thus the Luandan merchant Francisco Luís Vieira (who was also a colonel in the local militia) claimed in official documents before the Crown that he had exported 11,074 adult slaves and 13 children to Brazil in the period 1822 to 1825. Another militia officer and local merchant, José Severino de Sousa, officially swore to the Crown that he had shipped 20,018 adult slaves to Brazil between 1811 and 1816; and a third sought admission to the Order of Christ on the basis of having exported 4,825 slaves to Brazil between 31 October and 31 January 1805. Ibid., p. 94.

⁴⁶ In 1815, for example, Luanda imported 996.9 million reis worth of goods from Brazil and Portugal (the latter accounting for only 11 million of the total), and exported only 937.5 million reis to Brazil in slaves, wax and ivory. Of the Brazilian imports, 765.9 million reis were accounted for by Rio de Janeiro, 202.7 million by Pernambuco and 17.1 million by Bahia. Of the exports, 884.8 million reis (or 94%) was accounted for by slave exports, with wax being second at 45.5 million and ivory taking just 7.1 million reis, AHU, Angola, cx. 63, report dated 31 January 1816. By value, the most important products imported were textiles, followed by Gerebita, foodstuffs and tobacco, and gunpowder (which since 1808 was manufactured in the new Royal Powder Factory in Rio de Janeiro). For a detailed evaluation of goods, see Silva Rebelo, Angola e Brasil, quadro 3 following p. 179. On the Rio de Janeiro gunpowder factory and its exports to Angola see Ibid. pp. 155ff.