You are on page 1of 30

The Volume of the Atlantic Slave Trade: A Synthesis

Author(s): Paul E. Lovejoy
Source: The Journal of African History, Vol. 23, No. 4 (1982), pp. 473-501
Published by: Cambridge University Press
Stable URL: http://www.jstor.org/stable/182037
Accessed: 01-01-2017 18:51 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Cambridge University Press is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of African History

This content downloaded from 73.110.34.75 on Sun, 01 Jan 2017 18:51:06 UTC
All use subject to http://about.jstor.org/terms

Journal of African History, 23 (I982), pp. 473-501 473
Printed in Great Britain

THE VOLUME OF THE ATLANTIC SLAVE
TRADE: A SYNTHESIS'
BY PAUL E. LOVEJOY

SINCE its publication in I969, Philip D. Curtin's The Atlantic Slave
Trade: A Census has been the subject of a lively debate. On the basis of published material, Curtin estimated that 9,566,IOO slaves were imported into

the Americas and other parts of the Atlantic basin from I 45 I to I 870.2 Curtin
intended his study as a 'point of departure that will be modified in time as

new research produces new data, and harder data worthy of more sophisticated
forms of calculation. It will have served its purpose if it challenges others to
correct and complete its findings. '3 Scholars responded quickly to Curtin's
challenge. Individuals working on specific sectors of the trade have used new
data, gleaned from archives the world over, to modify Curtin's estimates for
various portions of the trade. These revisions have even led one scholar J. E. Inikori - to reject the validity of Curtin's entire census. Indeed Inikori
states categorically that 'there is now some consensus among specialists that

Curtin underestimated the volume of Atlantic exports'.4 Are Curtin's 'global
figures . . . much too low', as Inikori claims ?5 This article is an attempt to
synthesize the various revisions into a new global estimate. As such it provides
both a review of the literature since i 969 and presents a revised set of tables
on the volume and direction of the slave trade across the Atlantic. On the
basis of these revisions, it is clear that Inikori is premature in claiming the
emergence of a consensus. Curtin's initial projection of total imports appears
to have been remarkably accurate, despite numerous modifications in his
partial figures.

The significance of these efforts to quantify the slave trade hardly needs
to be mentioned. Ever since J. D. Fage first used Curtin's I969 figures in his
seminal article on the impact of the slave trade on West Africa,6 scholars have

1 An earlier version of this article was presented at the African Studies Association
annual meeting, Bloomington, Indiana, 198I. I wish to thank David Eltis, Stanley
Engerman, Joseph C. Miller, Henry Gemery, P. C. Emmer, Patrick Manning, Robert
Stein, Philip D. Curtin, and Russell Chace for their comments and assistance at various
stages of this synthesis. While I am solely responsible for any remaining errors, their efforts
were crucial in catching more than a few mistakes. I do not wish to hold them responsible
for my conclusions, but in more ways than is usually the case, this synthesis was a group
effort.

2 Philip D. Curtin, The Atlantic Slave Trade: A Census (Madison, Wisconsin, I969),
268. 3 Curtin, Census, xviii.
4 J. E. Inikori, 'Introduction', in Forced Migration: The Impact of the Export Slave
Trade on African Societies (London, I 98I ), 20. Also see Inikori, 'Measuring the Atlantic
slave trade: an assessment of Curtin and Anstey', Journal of African History, xvii, ii
(I976), I97-223; Inikori, 'Measuring the Atlantic slave trade: a rejoinder', Journal of
African History, XVII, iv (I976), 607-27; and Inikori, 'The origin of the Diaspora: the
slave trade from Africa', Tarikh, v, iv (1978), 8.
5 Inikori, 'Introduction', I9.

6 J. D. Fage, 'Slavery and the slave trade in the context of West African history',
Journal of African History, x, iii (I969), 393-407.

This content downloaded from 73.110.34.75 on Sun, 01 Jan 2017 18:51:06 UTC
All use subject to http://about.jstor.org/terms

474

PAUL

E.

LOVEJOY

attempted to correspond export figures with political and economic
developments in Africa. This literature in turn has spawned its own
controversies.7 Curtin's Census lowered the estimate for the scale of the slave
trade - which some scholars have taken to mean that the impact on Africa
was less than previously thought. For example, Fage has argued that

demographically, at least, the trade had a minimal impact, although politically and economically the trade had profound repercussions.8 Other scholars
- notably Manning and Thornton - have shown that the demographic effects
were significant, no matter what the absolute totals. Manning has postulated
a demographic model which distinguishes between slave-depleted areas,

slave-importing areas, and slave-trading areas,9 while Thornton has uncovered census data for late eighteenth-century Angola that confirm the sexual
imbalance in local societies which matches the sex ratios in the trans-Atlantic
trade.10 Finally, Inikori -using figures considerably higher than Curtin's
Census - has argued that the demographic impact of slave exports had a
retardative impact on African development."I Hence the debate over the slave
trade is far from a quibble over numbers; the debate ultimately relates to a
major theme in the history of the Atlantic coastal basin and, also, southeastern
Africa.'2
A number of Curtin's I969 estimates have been revised upwards, some
partial figures by substantial amounts, and consequently it is easy to see why
Curtin's global estimate for the total volume of the trans-Atlantic trade has
been challenged. Robert Stein, for example, has presented a figure for the
eighteenth-century French trade which is 2I 4 per cent higher than Curtin's
figure.13 Roger Anstey reached a total for the British trade from I76I to I807
which was IO-3 per cent above Curtin's figure; subsequent revisions of
Anstey's calculations increased the total still further - to a level i 8'3 per cent
I See, for example, Inikori, Forced Migration, which contains an excellent sample of
articles and sections of books relating to the debate; Roger Anstey, The Atlantic Slave
Trade and British Abolition, I760-I8i0 (London, 1975), 58-88; and my own study:
Transformations in Slavery: A History of Slavery in Africa (Cambridge, forthcoming).

8 J. D. Fage, A History of Africa (London, I978), 244-88.
9 Patrick Manning, 'The enslavement of Africans: a demographic model', Canadian
Journal of African Studies, xv, iii (I98I), 499-526; and Slavery, Colonialism and Economic
Growth in Dahomey, I640-I960 (Cambridge, I982). It should be noted that Manning is
working on a book which will examine the demographic impact of the slave trade on Africa;
see his preliminary paper, 'The political economy of African slavery', presented at the
Johns Hopkins University, i December I98I.
10 John Thornton, 'The slave trade in eighteenth century Angola: effects on demographic structures', Canadian Journal of African Studies, xiv, iii (I980), 417-27.
Inikori, 'Introduction', 20-59.
12 For example, Anstey (Atlantic Slave Trade, 79-88) argues that there was no net
population loss in the interior of the Bight of Biafra. David Northrup. Trade without
Rulers: Pre-Colonial Economic Development in South-Eastern Nigeria (Oxford, I 978), 81-2,
argues that the region only sustained the loss of population because of higher natural
growth rates than Anstey allows.

13 Robert Stein, 'Measuring the French slave trade, 1713-1792/3', Journal of African
History, xix, iv (1978), 515-21. Also see Stein, The French Slave Trade in the Eighteenth
Century: An Old Regime Business (Madison, 1979). Stein notes that 'It must be recognized
that the estimates advanced by Curtin for French slave exports were too low . . . due to
the quality of the available published data'. ('Measuring', 520.)

This content downloaded from 73.110.34.75 on Sun, 01 Jan 2017 18:51:06 UTC
All use subject to http://about.jstor.org/terms

1790-I867'. Rout considered Curtin's projection of 925.15 David Eltis examined the nineteenth-century trade with a similar result: Curtin's data were 30 per cent too low for the period from I82z-43.. however. 273-301. 244. In my calculations. I rely on Eltis's research for the period after I820. 18 Curtin. Secondly. I8. so must the global figure.. see especially page 29I. The African Experience in Spanish America: 1502 to the Present Day (Cambridge. if the partial totals required upward revision. They responded to Curtin's original challenge and generally adopted the same statistical methodology as Curtin in attempting to arrive at more accurate figures. 131-49). Journal of Economic History. II. xxxvii (I977). R. none of these scholars attempted a reassessment of Curtin's global figure for the slave trade. I base my estimates on the volume of slaves carried by the ships of different countries and not by import area for 1790-I8I0. It should be noted that D. It should be noted also that the totals in Murray's article have been corrected in his book: Odious Commerce: Britain. Murray demonstrated that imports into Cuba were greater than Curtin allowed. and Hispanoamerica y el comercio de esclavos: Los Asientos Portugueses (Seville. 1971. 176I-I807'. in Vera Rubin and Arthur Tuden. 15 Enriqueta Vila Vilar. Spain and the Abolition of the Cuban Slave Trade (Cambridge. 01 Jan 2017 18:51:06 UTC All use subject to http://about.. eds. new data showed that the figure Curtin used was wrong by I35. 17 Leslie B. these same individuals praised Curtin's efforts and accorded him the honour of providing the scholarly community with sets of figures which could be tested. 3-13. 1977).18 Rout favoured an educated guess to the detailed efforts at quantification represented in Curtin's study. The impression which gradually emerged was that Curtin must be wrong on that score too. Rout Jr. 16 David Eltis. 65. I980). Firstly. I2.34. III.000 slaves in a forty-five-year period. 4I0-I5.17 Rout's claims are reminiscent of the projections of those scholars whom Curtin criticized in his discussion of the historiography of the slave trade. 'The large-scale introduction of Africans into Veracruz and Cartagena'. and 'The direction and fluctuation of the Transatlantic slave trade.110. I977). I82I-I843: a revision of the I845 Parliamentary Paper'. Despite the upward revisions of different portions of the trade. 14 Roger Anstey.75 on Sun.Curtin's most serious critic . eds.VOLUME OF THE ATLANTIC SLAVE TRADE 475 higher than in the I969 census. This content downloaded from 73. In 1973 he questioned 'the credibility of Curtin's computations' for the volume of the slave trade to Spanish America before i8Io. His figure is also important because Inikori . after all. Rout. Murray's comments are not relevant for two reasons. Jr. Leslie B. 'The volume and profitability of the British slave trade. 1975). 1973). Murray was one of the first to question the accuracy of Curtin's estimates ('Statistics of the slave trade to Cuba.jstor.16 In all fairness to these and other scholars who could be cited. Journal of Latin American Studies. I82I-1843'. The Uncommon Market: Essays in the Economic History of the Atlantic Slave Trade (New York. 1979). in Stanley Engerman and Eugene Genovese. eds. Race and Slavery in the Western Hemisphere: Quantitative Studies (Princeton. ii. which supersedes Murray's work.4 Enriqueta Vila Vilar took Curtin to task for his treatment of the early trade to Spanish America. instead he estimated a figure of -5-I'6 million slaves.. 267-8o. 112. Census. was one of the first scholars to suggest that errors in the calculation of partial totals might indicate that Curtin's entire census was suspect.org/terms .000 slave imports off by 62 per cent. in Henry Gemery and Jan Hogendorn. Anstey's estimate is 'based on material not known to Curtin's authorities and therefore supersedes his results'.relies on it. 8o. Comparative Perspectives on Slavery in Newv World Plantation Societies (New York. 'The export of slaves from Africa. although he did not add any new data of his own.

Rawley demonstrated that the London share of the eighteenth-century British trade was much greater than previously known and indicated that upward revision in the British trade was necessary to allow for his reassessment. The method is dubious because Inikori transformed Curtin's estimate for imports into the Americas into an export figure .he estimates total slave imports into the Americas.org/terms .I00). Journal of African History.which Curtin did not do .000.using a percentage of approximately I 3 per cent to allow for losses in transit. 428. 'Measuring the Atlantic slave trade once again'.000 for other slaves which he claimed were exported to the Americas in the sixteenth and seventeenth centuries or otherwise were unrecorded in Curtin's estimates for the Portuguese trade. 'The British slave trade 175I-I807: a comment'. iv. and Seymour Drescher. He did not present alternative figures at the time.4 per cent higher than Curtin's estimate). Curtin. 595-605. 'Assessment of Curtin and Anstey'.205-13. 428). 19 Inikori.400. and the results can be used as an alternative to the method employed in this article in reaching a global estimate.110. obtaining a figure of I 5. The Transatlantic Slave Trade: A History (New York. Econocide: British Slavery in the Era of Abolition (Pittsburgh. the revised paper was subsequently published in the Journal of African History in I 976 and resulted in a series of exchanges involving Inikori. Finally. 20 Inikori. African Economic History. this kind of manipulation of statistics cannot be accepted. (1976). LOVEJOY Inikori's challenge came at a conference in I975. 277-8 fn. Inikori thereby arrived at an estimate for exports from Africa of I I. sources and a reappraisal'.392.000 slaves exported from Africa and I 3. IX (I980).000 compared to Curtin's figure of 9. with Eltis's revision for the period of I82I-43 (34. xvii. Rawley. and Robert Stein's reassessment of the eighteenth-century French trade (2 I *4 per cent higher than Curtin's figure). Anstey and Seymour Drescher. 1977). however. makes the same mistake of mixing data which overlap.000. This content downloaded from 73.OOO slaves.jstor. iv (1976). 20-I.19 Inikori used a straightforward (but dubious) method of adjusting Curtin's global figure upwards by 4.8oo.21 Rawley's method is more reliable than Inikori's. which he effectively attributes to Curtin. xviI. 606-7. Inikori then added his own projection for the eighteenth-century British trade (49-2 per cent higher than Curtin's estimate for British exports). See 'The port of London and the eighteenth century slave trade: historians. His computation of a global figure for the slave trade is also considerably higher than Curtin's estimate for imports into the Americas (I I.476 PAUL E. even though Rout's figure was only a 'guesstimate'. Roger Anstey. Inikori also converted Rout's import estimate into an export figure by allowing for losses in transit of 13 per cent. 197-223. 85-I00.ooo more slaves. Philip D.400.000 slaves (Slave Trade.' 19-21. Journal of African History.75 on Sun. 1981). He also accepted Rout's 'correction' of the Spanish-American import figure. or so Inikori claimed. In an earlier article. 'Introduction'. 01 Jan 2017 18:51:06 UTC All use subject to http://about. Presumably.345. 'Rejoinder'.ooo slaves. following a different procedure. The revision suggested is an additional 75.000. Inikori allowed a correction of almost 2.34. 21 James A. Curtin.000 slaves imported into the Americas. if the errors are corrected. Inikori makes a basic mistake in mixing different sets of data without accounting for the duplication resulting from such a mixture.20 As will be demonstrated below. 607-27: 'Introduction. Rawley basically employs the same procedure as Curtin did in I969 . These modifications accounted for an additional I.566. thereby accounting for approximately 66o. James Rawley. for a summary of Rawley's calculations.

my estimates are based on shipping data as analysed by David Eltis. Economic History Review. Rawley accepts Curtin's calculations for imports into the British Caribbean (pp. which Curtin did not record. but Eltis raises Curtin's total for I821-43 by I92. There is another error in the nineteenth-century calculation: Rawley refers to Eltis ('Transatlantic slave trade.1 (Slave Trade. I972).000 slaves. It also should be noted that Rawley refers to Table 14.ooo slaves exported fromn Africa between 1450 and I900 (Table i). which I calculate as 9. For most of the nineteenth century. and the Cape Verdes in Seventeenth Century Commerce and Navigation (Chicago. The details of my calculations are discussed below. basing this revision on the estimated 8o.23 An adjustment in Curtin's global figure for this trade would bring Rawley's and Curtin's estimates even more into line. 1979. these partial totals are based on the different sectors of the carrying trade.698.000 slaves to compensate for what he claims is an allowance for the previously underestimated London trade. for the period 145I-I600). he did not remain true to this procedure in estimating the volume of imports into the British Caribbean. p. Inikori and Rawley.I as a source for his estimate for imports into British North America.II0. ii. I 65) for British exports. I82I-i843'. 273-30I) in modifying Curtin's figure.jstor. This total estimate for the export trade is then used to derive a figure for the number of slaves imported into the Americas and other parts of the Atlantic basin between I450 and I867 (excluding the offshore islands after i6oo). although table 7.with minor adjustments . since these imports are also included in his totals for the Old World and Spanish America (see table 2. other than the Americas after i 6oo.3 does not record all the slaves tabulated by Curtin for the British Caribbean (see Census. British-transported slaves and slaves imported into British territory are simply not the same.000 as Rawley has it. not the number of slaves imported into the British Caribbean. Atlantic Islands.110. but since there is no such table it is not possible to examine his calculations. These calculations produce an estimate for the volume of British slave exports from Africa. Rawley also makes allowance for 122. 289). 1690-I 807.ooo for imports into the British Caribbean.VOLUME OF THE ATLANTIC SLAVE TRADE 477 attempting to bypass shipping data.22 These errors alone account for an excess of I. Rawley also makes the mistake of adding an estimate for slaves carried on British ships after i 807.100 (p. 2I0. I.000. Census. I have essentially used Curtin's findings .778.ooo slaves calculated by David Eltis ('The British contribution to the nineteenth-century Transatlantic slave trade'. which is derived from Curtin. but it should be pointed out that when the revisions of the scholars who have worked on the various sectors of the trade are analysed. For I 70I-I8IO. 226). Rawley again mixes exports by national carrier with imports by receiving area. Rawley provides no explanation for the discrepancy. The volume of the Atlantic slave trade is here estimated at i I. Bentley Duncan. the Azores.75 on Sun. Madeira. which produces an error for Brazil of 224. it becomes clear that the claims of Inikori. The combined error is 332. The procedure I have followed differs from the methods employed by Curtin. 25. which is well within range of Curtin's original projection. Rawley has counted all Portuguese slave imports as Brazilian imports.before I 700 and have used the partial totals of a variety of scholars for the eighteenth and nineteenth centuries. and adds 75. Rout and Rawley are indeed extreme.3). I40). rather than on estimates for imports into different colonies.900 slaves. so that an adjustment for these sectors alone reduces Rawley's global estimate to IO. i64-7 and table 7. although there is invariably some overlap between all approaches. xxxII. not 300.34. ii6. Unfortunately. This content downloaded from 73.org/terms . and he made a serious mathematical error in computing the Brazilian sector.800 for Brazil and 778.235. Furthermore. 000 slaves shipped to other parts of the Atlantic basin. 01 Jan 2017 18:51:06 UTC All use subject to http://about. For imports into Brazil before i6oo. 23 Rawley relies on T.500 22 Rawley refers to table 7.

THE VOLUME OF THE EXPORT TRADE There have been only a few revisions of Curtin's figures for the period before I700.000 slaves. Census.110.900 in total imports.ooo 6.698.566. especially once the relatively minor changes are made. I5I9-I8IO (Mexico. and it is shown that my estimate is remarkably close to Curtin's original global estimate of 9. Neither presents convincing proof that Curtin's figures are wrong. whose This content downloaded from 73.ooo 100-0 Source: Tables 2-4. Rawley and other scholars. I35. 1946). LOVEJOY Table i.000. which translates into 2I6.I00 exports. 275). 25 In her initial tabulation.24 Exports from I450 to I700 are calculated at 2. The revisions come from two studies. Also see Curtin. in Gemery and Hogendorn. Vila Vilar examined import data for Spanish America from I595 to I640 and calculated that Aguirre Beltran's figure was too low by I35.34. Rio de la Plata. 206-9). 367). 'The Dutch participation in the Atlantic slave trade. although both Inikori and Rout claim that more extensive revision is called for. rounding off to the nearest hundred.1 i6-o 5 24 28-5 Total 11. 1970].477 slaves) (Ernst van den Boogaart and Pieter C. but she later revised these figures as follows: Cartagena. The volume of exports in the period before I700 (Tables 2 and 3) has been reached by allowing for losses at sea on the order of I 5.org/terms .600 slaves. ig. Uncommon Market. La poblaci6n negra de Mexico. 220.75 on Sun.000 3. 276-7. Slave exports from Africa: the Atlantic trade Period Volume Per cent 1450-I600 I60o-1700 170I-I800 I80I-I900 367. Dutch exports suffered to the extent of 17-9 per cent losses in transit between i637 and i645 (sample of 27. Inikori.235. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 44.560.664.000 I.478 PAUL E. 6. Hence an estimate for exports from Africa in this period depends largely on Curtin's import calculations. which are primarily based on estimates of the slave population in the Americas and the islands of the Atlantic basin. and the Caribbean islands. 69. but I have added these figures as 268. 20 and 25 per cent .jstor. My import figure is then compared with the import estimates of Curtin. 292). Emmer. Her total is 268.000.133. Vera Cruz. The Royal African Company [New York. G. I00. Vila Vilar's figures replace the earlier estimates of Gonzalo Aguirre Beltran. Vila Vilar estimated that 220.8oo slaves were imported through Cartagena and Vera Cruz alone between I 595 and i 640 ('Veracruz and Cartagena'. The modifications in Curtin's total for this period require an upward revision of I72. I 596-I 650'.25 Robert Fogel and Stanley Engerman 24 The Royal African Company recorded losses in transit of 23-5 per cent between i 68o and i688 (K.000 3. slaves (Table 8). which amounted to an estimated I91I per cent of total Atlantic shipments during the whole period of the slave trade.868.200.the preferred series is based on 20 per cent losses.330. Davies. 7.664 (Comercio de esclavos.

480 slaves. I68o-I700. including the decades before i640 when his estimates are now shown to be too low. Because Rout has claimed . nor did it allow for the system of calculating the trade on the basis of piezas . 15. 21-5. illegal trade and licenses was an attempt to adjust for the inevitable overestimation of the asiento figures. Ioo slaves were imported into Spanish America between I64I and 1773. Fogel and Engerman. that my figures for the whole period from 1575 to i65o are calculated as follows: I used Curtin's average for each quarter century to account for the years not covered by Vila Vilar. He used the asiento as a capacity figure which represented the estimated labour requirements of the Spanish regime. indicates a similar level. however. For the whole period. as the legal trade. there has been no substantial evidence presented which would require a further upward revision for the period before I700. It should be noted. as well as scattered information on various contracts to deliver slaves.jstor.. see Slaves of the White God: Blacks in Mexico. so that it is only necessary to consider the last sixty years of the seventeenth century. 26-8. 1976). 1443 slaves. and Curtin is higher than a simple addition of the revisions to Curtin's original figure.and Inikori has accepted the claim .rather than too low .000. total. 01 Jan 2017 18:51:06 UTC All use subject to http://about.as a guide to the volume of imports into Spanish America.500 slaves on the basis that Curtin had underestimated imports into North America. which seems too high.VOLUME OF THE ATLANTIC SLAVE TRADE 479 added another 20. 1570-I650 (Cambridge. 30. Rout published before the work of Vila Vilar appeared. 20. Inikori does not refer to Vila Vilar's work. and indeed for the first seventy years of the eighteenth century. Engerman. 1974). Curtin estimated that 5 I6. did not include contraband traffic and private licences. it is necessary to examine the basis of Rout's arguments. Palmer has challenged Curtin's estimate for Spanish American imports. but because my recalculation of the totals based on Vila Vilar. that Palmer's estimate is not based on as strong evidence as the other two revisions. she accounts for the period from I595 to I640.an idealized standard less than the actual number of slaves imported.) I have rounded off the figures to the nearest hundred and divided the estimates into quarter centuries.000. 1432 slaves.300) (Census. The second. Asiento contracts were rarely filled and often fell well short of the number of slaves specified. I have not adjusted for Palmer's suggestion.600 was used by Curtin in calculating the imports into Spanish America in this period (Census. 28). Census. Mass. 27 Curtin. I670-80. Colin A. Curtin may well be too high . 26 Robert Fogel and Stanley Engerman. Curtin was well aware that the asiento.in his estimate for the last sixty years of the seventeenth century. which seems more likely (p. for I570. and as a result. Nonetheless. for 1553. Time on the Cross: Evidence and Methods (Boston. but Vila Vilar's revisions are hardly on the order of Rout's 'guesstimate'.659 slaves.75 on Sun. I946 slaves. hence his decision to disregard the pieza.26 Otherwise. It should be noted. for an average of 3. moreover. (Personal communication.org/terms . 23). This content downloaded from 73. He relies on two population estimates for the number of slaves in Spanish America. Fogel and Engerman calculated the following time pattern: I62o-50.880 slaves per year. For the period before 1595. indicates a slave population of 20.27 Although it is not possible at present to substitute new data for Curtin's figures for Spanish imports between I64I and I700 -as Vila Vilar did for the period 1595-I640 -it is possible to use the data assembled by Jorge figure of 132. 25) to 73. I650-70. The first population estimate.that the volume of imports into Spanish America should be adjusted upwards substantially. raising Curtin's figure (51.the legal contracts issued by the Spanish government to different countries .110. Curtin used the figures for the various asientos .34. S.

34.480 PAUL E.0 00 -00 o c.0 N vE tl_ 0 _ ) -4 tf) tn tt) tn.v 000 irgi. in in Ogg O . E .jstor. ~ ~ ~ ~ ~ ~~~r 0. S This content downloaded from 73.00 .e * 0 0 ~~~ 00 0 00 0 0 0 0000800 00 0 0 0 0 X~~ ~~ u 0.t . ob+ . 4. LOVEJOY 0 4*~0. 01 Jan 2017 18:51:06 UTC All use subject to http://about.org/terms ..: 0 0 0 0 c ino in OC 0 n X ON in OC o m o M . iz 0N.~0 0000 0 0 0~~~~~~~~~~ .75 on Sun.110.

VOLUME OF THE ATLANTIC SLAVE TRADE 48i z 0 0 00 00 oc 0 0 0 0 0 0 0 0 r.110. -.E--4 4-1 .- 00 This content downloaded from 73. E--4 ql) ."C en en r. v m x CZ Cs 0. N 00 tj 0 0 0 0 0 0 0 0 0 0 en en 00 Cs 0 0 0 0 0 0 0 0 0 0 0 "C N t.75 on Sun.en 11.34.4u 06 N tn 0 tn 0 N tn t.0 0 0 N tn t.00 0 x 0 0 0 0 0 0 0 0 0 0 0 't t- 00 0 X in 00 "C en 00 N 00 tn Z tz v uu 4-i "" v t. tn r cn 1-. o tz cn 0 0 0 0 0 0 0 0 0 u 0 00 "C N C.jstor.en 00 N 0 "C t. 01 Jan 2017 18:51:06 UTC All use subject to http://about.org/terms . 0.

29 Preciado. an average of I...31 For I 685-6. 2.699 in i699. LXII (I975).000 slaves or 52-4 per cent of total trans-Atlantic shipments in all periods. Chace notes that the volume of imports probably was on the order of 5oo-6oo slaves per year in the I640s. 99. 32. Postma bases his study on the records of the Dutch West India Company. Data for i663-74. The scholars who have presented partial revisions include Johannes Postma. 35 Certainly the imports into Buenos Aires.110. 30-I.840 slaves per year. 30 Curtin. has been calculated on the basis of numerous revisions. La Trata de Negros por Cartagena de Indias (Tunja. Race and Slavery. only 490 slaves were legally imported. 2. 34 Ibid. Inikori 28 Jorge Palacios Preciado. legal imports were only I50-200 slaves per year. Stein.D. but in the second half of the century. could not have accounted for the levels necessary to sustain the claims of Rout and Inikori. but for the period I640-I700 as a whole. Census. 1973).34 If illegal imports and the legal imports into other Spanish American ports were known. 32 Ibid. imports totalled 2. supports Curtin's belief that even when allowance is made for the contraband traffic and the other factors the asiento figures are still too high. 39 fn. Cartagena was the major port for Spanish America during much of the seventeenth and eighteenth centuries. French Slave Trade. 49. 'The origin of African slaves: the Dutch activities on the Guinea Coast.000 slaves) were legally imported into Spanish America. 36 Johannes Postma.886 in i698. 'French slave trade'. 37 Stein. 29. 25. I 1. thesis.207. I675-1795'. This content downloaded from 73. the volume of legal imports into Cartagena increased dramatically. who has done similar archival work on the French trade . 5 i 9. 237.32 Again. and Stein.37 Drescher. Under the Portuguese asiento.75 on Sun. 36-41. which amounted to 6. Preciado estimates that legal imports into Cartagena probably did not exceed 400 slaves per year.880.I33. Revuefranfaise d'histoire d'outre-mer. 01 Jan 2017 18:51:06 UTC All use subject to http://about.28 Preciado's information on imports into Cartagena and other parts of Spanish America. see Russell Chace. trade was unstable at best. were small.34. LOVEJOY Palacios Preciado to check the validity of Curtin's work. when the Spanish colonies suffered an economic recession. indeed. which was only a minor centre of the trade in this period. for example.500 per vear.org/terms .36 who has calculated the Dutch trade from archival sources. Trata de Negros. Stein bases his study on the records of the Amiraute which kept detailed accounts of ships and cargoes. For the period from I650 to i697.853 piezas (perhaps 22. unpublished.668 in I70I.600 in I700 and i. 'The African impact on colonial Argentina (Ph.000 slaves per year. yet for i676-9. it is likely that total imports were greater than Curtin's average for these years. the illegal traffic.35 The eighteenth century trade. These are incorporated into Tables 4 and 5 and include material directly based on the volume of slave exports from Africa.29 a time when Curtin estimated that annual imports were on the order of 3. while important. show that I7..30 Licenses and contraband traffic could have accounted for an additional 2. the average was probably less than Curtin estimated. and Postma 'The Dutch slave trade: a quantitative assessment'. Santa Barbara. Trata de Negros.482 PAUL E. University of California. 31 Preciado. in Engerman and Genovese.jstor. 31. 33 Ibid.33 In the last years of the seventeenth century. but the range of imports was still on the order of Curtin's estimate. but hardly the volume necessary to suppQrt the claims of Rout and Inikori. despite the fact that the asiento allowed for an import of I . 1969).

Curtin. 1978). 267. 28.VOLUME OF THE ATLANTIC SLAVE TRADE 483 Table 4. 'The Dutch slave trade: a quantitative assessment'. for 176I-I800. Econocide.900 Sources: English trade: for 1701-50. 519. 170I-60. 237. for 176I-I800. Census.200 Danish 73. LXII (1975). 01 Jan 2017 18:51:06 UTC All use subject to http://about. Census. 'British slave trade'. 1978). Revue franf aise d'histoire d'outre-mer. 'The origin of African slaves: the Dutch activities on the Guinea Coast.110.201. I760-I8io'. Revuefranfaise d'histoire d'outre-mer. 'The history of the Danish Negro slave trade.900 North American I94. The Middle Passage: Comparative Studies in the Atlantic Slave Trade (Princeton. North American trade: Roger Anstey.I80. Dutch trade: Johannes Postma. Also see Anstey. French trade: Curtin.ooo to the total for the French trade in order to allow for unrecorded vessels between 1713 and 1728 (Stein. 2I3-4. I have preferred this estimate to the one in Johannes Postma. for West Africa. for Angola. 151. 26I. I have adjusted for Anstey's total by allowing a figure of 245. 201. personal communication). 27. in Stanley L. Atlantic Slave Trade. I701-1800 Carrier Total English 2. and Anstey.300 350.38 Herbert Klein. who has done comparable archival work on the Portuguese trade from west-central Africa from I7IO to i8oo. Klein.jstor.900 Other (Swedish. for 1701-10. 1975). 13. 17I3-I792/4. see Seymour Drescher..796. Danish trade: for I733-60. for 1751-1800. This content downloaded from 73. eds. 176I-I8I0'.000 for I8oI-7.39 and 38 Drescher. 'Measuring the French slave trade. Atlantic slave trade. The Middle Passage: Comparative Studies in the Atlantic Trade (Princeton. Journal of African Histor.75 on Sun. XVII. I have also added io. Roger Anstey. 607. 'Assessment of Curtin and Anstey'. 'Continentalpowers'. in Engerman and Genovese. andAnstey. for 1733-60.org/terms . 1977). 49. Portuguese trade: Herbert S. I70I-60. 27. 39 Herbert Klein. iv (1976). Roger Anstey. 'The British slave trade 175I-I807: a comment'. I733-I807'.y. Curtin. ii (1977).300 French Dutch I. 'The slave trade of the continental powers.34. Revue franfaise d'histoire d'outre-mer. 267.300 Portuguese I. 28. Journal of African History. Engerman and Eugene D. I3. for I76I-I800. Genovese. 21 I. 176I-I807'. while Anstey used clearance lists for British ships for 176I-I807. whose combined archival work on the British trade covers the years I75i-I8oo.532. xix. and Inikori. 207. xxx. Anstey 'Continental Powers'. Race and Slavery in the Western Hemisphere: Quantitative Studies (Princeton. I675-1795'. Econocide: British Slavery in the Era of Abolition (Pittsburgh. LXII(1975). Klein used shipping records from Rio de Janeiro and Angola for the Portuguese trade.132. Svend Erik Green-Pedersen.000 Total 6. The Economic History Review. Race and Slavery. 'The volume and profitability of the British slave trade. and Robert Stein. Drescher and Inikori used Customs series 3 and 17 in the Public Record Office for 1777-I807. iv (1978). Brandenburger) 5. 'The volume of the North American slave-carrying trade from Africa. Anstey. LXII (1975).

LOVEJOY Svend Erik Green-Pedersen. Revue fran.jstor. ed.ooo too low for the period I760-I8I0 (Time on the Cross: Evidence and Methods. see 'The slave trade of the continental powers. This content downloaded from 73. In his reconstruction. I 26. The two tables on the eighteenth-century trade analyse the volume of exports according to carrier and export region.75 on Sun. Scheuss de Studer. ii. or by area of import in the Americas.474 slaves imported into Buenos Aires under the French asiento. therefore. I958).600 slaves are not accounted for in Table 5. 'The volume of the North American slave-carrying trade from Africa. LXII (I975). and the French trade for I70I-I0. Revue franfaise d'histoire d'outre-mer. Some of these slaves came from South-east Africa and Madagascar. however. I 760. Stein has no data for the early years of the eighteenth century. however. may have traded in I7II or I7I2. La trata de negros en el Rio de la Plata durante el siglo XVIII (Buenos Aires. Green-Pedersen and Curtin. this has not been done here. Charleston Customs records.574 were imported between I703 and I7I0. Postma.aise d'histoire d'outre-mer.760 slaves. which was Curtin's method for reaching a total for the century. exports by coastal region. While it is possible to assign these slaves to the different regions according to the known distribution. 42 Stein. 'African Impact on Argentina'. Documents Relating to the Slave Trade (Washington. the Portuguese trade from West Africa. There is one problem. 'The history of the Danish Negro slave trade. For purposes of calculating total exports for the century.42 Elena F. The Economic History Review. I am assuming that these 33 ships contributed to the portion of the total French trade assigned to I70I-I0.484 PAUL E.ooo slaves on board. LXII (I 975). 30-I). Stein does not state when these ships sailed.000 slaves per year to Spanish America but were not able to achieve this figure. 26I. hence the 33 documented ships may have had as many as io. 43 Elena F. 01 Jan 2017 18:51:06 UTC All use subject to http://about. I70I-I0. I803-7). xxx. 41 For Anstey's synthesis of the continental trade. see Chace. Green-Pedersen relies primarily on the records of the Great Negro Trade Commission of Denmark. (I 977). when Europe was at war and French participation in the trade is difficult to assess. 1930-5). I733-I807'. Only in the years before I 76 I are any of Curtin's original projections used: these include the British trade from I70I to I750. Ships carried an average of 306 slaves. 47-5 I .000 slaves for other minor carriers in the eighteenth century.g. For the North American trade. Except for 33 ships in the years before I 7 I 3. 20I. which includes some archival material (e. Anstey relies on the work of Herbert Klein. Stein's I978 calculations raise Curtin's I969 estimate for eighteenthcentury French exports by I84.org/terms . Fogel and Engerman have provided some verification of Anstey's revision: they calculate that Curtin's estimate for total imports into the United States was I48. I76I-I8I0'. who has calculated the relatively small Danish trade from primary sources. 65. this revision is an important one.43 Preciado accounts for another I3. if not most. For a fuller discussion of the Argentine trade. especially the material in Elizabeth Donnan.110. It should be noted that 6 I 9.34.207. and some. I I . but the coastal origin of the vast majority is simply not known. Since Stein's work is based on shipping data previously not used.40 Anstey has also revised the figures for the North American and various European carriers from I76I to i800. The French were supposed to deliver 4. the summation of the revisions of the different carriers is clearly superior to the summation of exports by coastal origin. see Anstey. For my purposes. Scheuss de Studer has recorded 3.41 I am also making an allowance of 5. of which 2. French Slave Trade.734 slaves imported into the other parts of Spanish America by the 40 Svend Erik Green-Pedersen. Anstey uses widely scattered data in his synthesis of the North American trade..I 8 I 0'.

N _Es.. Co2 Q ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~u.~~~~~~~~ ~ ~en~~t-I.4 -c . 0 *_+.org/terms voO .4~ _ ^ X=.jstor.CZ . > .VOLUME OF THE ATLANTIC SLAVE TRADE 485 cn~~~~~~~~~~~~~~t en 08 -. &I e +mbn 0 0 4 - ^Q X~~~~~~~~~~~~~~~~.C %) Z U^o V ? v .. == > _t o.X:S. - o t-X t tme)=Nts tn'* S _ 1 t1 t Ch > m b > C A: Q ~~~~~~~~~~~~~~~~.0D== F^O?C= 4 ' ? _ _ ~ ~ ~ ~N 0C _.o This content downloaded from 73. C/ (z m-0 ^ 0 00 000.0 X *:< b Q rs>a~..--r0s4 ml= X >l+xX^=O -3= O _ .0Oq0g :w==' Er Xw~~~~~~~~~~~~~~~~~. > 0.. t .. >Z 0/ x=?=QX<. 01 Jan 2017 18:51:06 UTC All use subject to http://about.b tr ^ O 0. N t. I_0 N0 7t X >.X cn D 'S t~~~~~~~.110.00 00 0.34.3CZ|. / .75 on Sun.

iio-ii. 47 Op.but not the others. Anstey presented an initial revision in I975. Although the volume of the eighteenth-century trade is calculated here on the basis of national carrier and not importing region. 32-3. LOVEJOY French.000-15. 'African impact'. 49 Chace. Before I 720.jstor. I65. A second volume of material collected by Mettas is forthcoming. This estimate for the unrecorded vessels is partially confirmed by John C. 45 Stein would allow for one or two ships per year (personal communication).000 higher that Curtin's figures for these years. 13. and by combining their studies it is possible to revise Curtin's figure for the British trade in the period from I 75I to I807.44 Hence the French delivered at least I7.3.000 higher than Stein's figure for I7I3-92/3. Palmer. with illicit trade another one-third to one-half . see especially p. At Cartagena. I 1 I 5-39. 3-31. 40 fn. but until one is done.000 were exported to French colonial possessions. 48 Op. Colin A. It seems likely that at least another 9. Between I747 and I753.486 PAUL E. 312. cit.75 on Sun.880) for the period i641-1773.for the years 1728 and 1729 . I wish to thank Professor Stein for discussing this problem with me. the speculations of Rout and Inikori cannot be taken as a contribution to sound historical reconstruction.057 slaves. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 338. 29).45 Hence my total for the French trade is 40. It should be noted that Preciado's data on imports into Buenos Aires differs from Scheuss de Studer's figure. Stein's figures are also too low for I7I3-28. 'British slave trade'. imports averaged 286 per year. 50 Anstey. which would allow the possibility of 2. it is necessary to return to the problem of imports into Spanish America in order to continue the assessment of Rout's and Inikori's criticism of Curtin on this point.49 Anstey.208 slaves. for example. as 74.org/terms .ooo slaves.300 slaves imported illegally. a more detailed study of the trade to Spanish America is warranted. Serge Daget. Preciado accounts for 3. Stein appears to have included a few of these ships .46 Between I730 and I736. Certainly. although the size of the ships and the number of slaves purchased are not known. another 4. exports were probably on the order of 2I. his estimate is not altered significantly. Human Cargoes: The British Slave Trade to Spanish America (Chicago. But as Curtin used an annual average (3. imports increased to 789 per year.50 but the real break44 Preciado. if at all. in the I720S.110. average annual imports into Cartagena were 263 slaves. 1981]. perhaps by another io. As the information on the French asiento between I 703 and I 7 I 3 makes clear.760. ed. because data are missing for some French ports for some years.000 slaves. Clark (La Rochelle and the Atlantic Economy during the Eighteenth Century [Baltimore.34. 53-64. legal imports were still on an order compatible with Curtin's estimate: I. Trata de Negros. Nantes. shows British imports into Spanish America. Drescher and Inikori have used new data for the British trade in the last decades of the British trade.48 It would have taken a considerable number of illegal imports both at Cartagena and elsewhere to produce a volume on the order claimed by Curtin's critics. cit. 46 Preciado.47 and the level of legal imports was scarcely any higher in the middle decade of the century. who records eight ships leaving for Africa between 17io and 17i9 and eleven ships between 1720 and 1729. This content downloaded from 73. the scattered returns for the next several decades do not suggest a higher level of imports. Trata de Negros. 1978).986 slaves are recorded. at least not for those years when data are available. I have not attempted to compare Stein's material with the data on the French trade collected by Jean Mettas (Repertoire des expeditions negrieres franfaises au XVIIIe siecle. which averages 7I2 per year. i98i).564 slaves per year.

5 Anstey's revision for 175I-I807 is 1. series 3 and I7.380 slaves. 214 for estimates of 2. 'Rejoinder'. 'Assessment of Curtin and Anstey'. 624. The revised figure for the British trade from I75I to I807 accepted here is I.53 For the period in which both Inikori and Drescher used the same archival material . and 2.307. (b) an increase ranging from 394. who reached similar results working independently. however.307.099.959. I defer to Anstey and Drescher. 'Rejoinder'.300.913. Inikori. the problems of interpretation centre on the number of slaves per ship. Econocide.600 to 563. 28. (c) for both .300.476. 'Comment'. taking no account of unrecorded trade.365.380 slaves ('Comment'. 142).000 slaves for I80I-I807 is in Econocide.500.800) on the basis of new statistical evidence.VOLUME OF THE ATLANTIC SLAVE TRADE 487 through in the study of these decades was the discovery of Customs records. working independently.52 Since Inikori and Drescher counted the actual ships involved in the slave trade. while Curtin's figure for 1701-50 is 863. Econocide.777. ranging from 2.51 Their tabulation of the raw data was slightly different. or (c) an increase of 922. at the Public Record Office. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 56 Inikori. Anstey joined the debate in defence of his own calculation.56 His argument is based on an assessment of commercial patterns and an allowance for error in existing figures. This content downloaded from 73. that even if Inikori's calculations on the British trade are accepted .699.75 on Sun. 51 Inikori. Inikori used a much greater ratio.48I. and I have used Curtin's estimate for the first half of the century (863. making allowance for non-slave carrying vessels.9I0. 54 Drescher. and now there are better results. It should be noted. 52 Anstey. The other problem with the eighteenth century relates to an assessment of the minor slaving countries. and he reassessed the data himself. In this case.900 (Census. (b) for the period after I750. and 2. Dutch and Danish ships. but far more significantly.014. assuming unrecorded ships more than made up for non-slave carriers. thereby resulting in a greater estimated volume. incorporating Inikori's findings on the number of ships involved in the British trade. 607).(a) for the period before I 750. but Inikori has not accounted for his difference with Curtin (4i8. presents different estimates: 2. examined these documents.959. 'Assessment of Curtin and Anstey'. and Drescher.300 greater than the Curtin-Anstey figure used here.34. 605-6. which is I8-4 per cent higher than Curtin's original estimate for British exports in this period.4I6.476. Curtin made a rough estimate for North American. 28.55 It should be noted that Inikori's estimate for the British trade from I70I to I807 (3. Inikori calculated the British trade variously. Drescher and Anstey used similar ratios. although Drescher did not attempt an estimate for the period before I777. 53 Inikori. Inikori and Drescher came to far different conclusions in their interpretation of these data.572) is 922. Drescher's calculation is 10-4 per cent higher than Curtin's original estimate.913. taking into account unrecorded trade.my global estimate for exports would be affected as follows: (a) an increase of 4I8.800. 624.org/terms .900). 28. 214. Drescher reached a total of I.300 slaves. while Inikori's figure was I.986 to 2. Inikori and Drescher. It is possible that Curtin's figure for the first half of the century is low. and hence their totals are approximately the same.54 In order to reach a figure for I70I-I807 I have accepted Anstey's figure for I75I-I807. for a total of 2. while Inikori's figure is 48-8 per cent higher.986.I777-I807 Inikori was also considerably higher in his calculation.jstor.110. Inikori does not indicate how he arrived at these conflicting figures.600. Drescher's estimate of 245.

49.62 A discussion of the nineteenth century. 'Danish Negro slave Trade'. 64 Curtin. Miller. unpublished paper presented at the African Studies Association annual meeting. and Anstey estimates that a few slaves .170.D.470 slaves. 63 Anstey. Green-Pedersen. 62 Eltis. again basing his calculations on archival data. 'Slave trade of the continental powers'. ships handled another I00. which is here estimated at 3. David Eltis has led the way in a reassessment of the nineteenth-century trade. 5 Curtin had allowed 173. I844-I867'.488 PAUL E.34. 'Slave trade of the continental powers'. 65.000 slaves carried by British ships.57 Postma used archival material to reach a figure of i I6. Drescher. and Eltis. Eltis's major contribution is his study of the period from I821 to I867. 273-301. 20I.were on French ships.60 The Danish trade has now been estimated at 72. I8i I-2o. adjustments in the partial figures of these carriers. Daget accounts for approximately i i 8. 419-42. Econocide. 'The direction and fluctuation of the transatlantic trade. 'Origin of African slaves'.ooo slaves exported on French ships before I845. Anstey has estimated that U. 28. i6i. 'Continental powers'. 26 I. Anstey.59 Postma also accounts for 234. and Anstey. Anstey. 'Legal Portuguese slaving from Angola. see 'Transatlantic slave trade. Census. 61 Green-Pedersen. Brandenburger and others probably took only a few thousand slaves in the course of the whole century . Some preliminary indications of volume and direction'. Census. but a preliminary discussion is provided in 'British repression of the illegal French slave trade: some considerations'. 59 Curtin. 'North American slave-carrying trade'. 1783-I815' (Ph. The Portuguese trade is based on Miller's data for Angola. I82 I-I843'.600 as a figure for Dutch exports in this period.000 slaves for the Danish trade. 'North American slave-carrying trade'. Indiana. and Anstey. 65. see Robert Stein. It should be noted that Eltis's revision of I844-67 supersedes the calculations of David R. Curtin's original estimates for the rest of the Portuguese trade largely stand. Bloomington. Drescher. therefore. 49.63 Drescher accounts for 245. Census. 212. I980). 210-20. 60 Postma. Uncommon Market. hardly affect the overall total for the slave trade. or 28&5 per cient of the total slave trade (see Tables 6 and 7). I now allowed for the thirteen French ships which sailed for Africa between I804-6 from Nantes. 21 I. 58 Postma. I have not seen Serge Daget's thesis on the nineteenth century French trade. 287-8.i IO . while the North American trade after I76I was I94. York This content downloaded from 73. 'Origin of African slaves'. Since all these carriers only transported about io per cent of total slave exports in the eighteenth century. while important for an analysis of individual countries.000. 'Transatlantic slave trade. thesis.110. Spain and the Abolition of the Cuban Slave Trade (Cambridge.330. 01 Jan 2017 18:51:06 UTC All use subject to http://about. in Gemery and Hogendorn. 20I. I981. Odious Commerce: Britain.472 slaves handled by the Dutch from I70I to I760. 267. Murray.jstor. LXII (1975). 'The Nantes slave traders. 'Danish Negro slave trade'.org/terms . Revue franfaise d'histoire d'outre-mer. 299-301. Joseph C. The calculation for the first decade of the nineteenth century is based on the revisions of Anstey. Miller and Green-Pedersen for most of the export volume.4I6 slaves for the Dutch trade from 176I to I794.75 on Sun. I82I-I843'.here estimated at 5. primarily involves the first two decades and the period after I867. Green-Pedersen allows 2. 26I.61 and the Swedish. although his calculations may be low. Eltis has allowed for the illegal French trade.000 slaves. LOVEJOY although the total for all minor carriers is not significantly different from Curtin's initial estimate.S. although I have also relied on some of his estimates for the decade.000.64 Miller shows that west-central 57 Curtin.

1972).8oo which I have left unassigned. For I 817. 2I6. Miller has accounted for 22. and it is likely that several thousand slaves per year were being sent in the early years of the second decade. 6-50.000 slaves (personal communication).A Study of Quelimane District (London.e. Custom records for Quelimane. 'Sources and knowledge of the slave trade in the Southern Atlantic'.jstor. Curtin had estimated exports from southeastern Africa at 5. the reports show both the number of slaves embarked in Africa and the number delivered in Brazil. they do suggest that the volume of exports from southeastern Africa was significant and very likely spilled over into the Atlantic. I have assigned s.IOO unrecorded slaves to Cabinda.ooo slaves for Mozambique Island in I820. while in i 8I5 and i 8 i 6 only slaves delivered are recorded. Alpers has reported information on one Brazil-bound ship in I 802 with 620 slaves.239 slaves. 'Slave trade of the continental powers'.110. I980). west-central Africa. Anstey's summary of the I8oi-io Portuguese trade is used. the allowance here is 6o.600 slaves. and Miller.000 slaves were sent to Brazil.600 slaves. 'Portuguese slaving from Angola'.f65 The i 5. Miller. 92. which could have involved the transfer of a total of 5. Capitalism and Colonialism in Mozambique . Also see Joseph C. Ivory and Slaves. and s.272 in i821. 213. For the years i8 I '-14 and I8 I9. 01 Jan 2017 18:51:06 UTC All use subject to http://about. exclusive of Cabinda. I have allowed 9. Miller's data on the Portuguese trade in the decade I8I 1-20 are based on British consular reports from Brazil which were not included in the I 845 Parliamentary Paper. Custom records for Mozambique Island show that 8. and 12. Ivory and Slaves in East Central Africa (London. in I803 Portuguese exports from Mozambique Island alone were 5. 26I.10O slaves he lists as unrecorded is an allowance for Cabinda and southeastern Africa. and 4. 1976. i 6 i.000 may not be too high.300 to southeastern Africa. 67 Miller.69 The University.34. although again there is a gap in the data for the region north of the Zaire River and for imports into Pernambuco. 1975).i64 slaves were exported in i8i8. Otherwise. 69 Alpers. although most of these went to the Mascarenes and not Brazil. which combined with 5.ooo as a result of scattered data for Mozambique for I802. In I 809. Alpers. A.241 slaves were exported from i 814 to i 820. 66 E. I87-9.ooo. I6I.400 slaves. and i 8 i 8 and i 820 there were no returns. 68 Miller. I 500 to 2. Mozambique: The Africanization of a European Institution: The Zambezi Prazos.75 on Sun. perhaps to a level even greater than ioooo for the decade.66 While these figures barely affect the figure for total Portuguese exports to the Americas. Miller has shown that west-central Africa exported 246.700 slaves previously assigned to southeastern Africa brings its total to io. 'Portuguese slaving from Angola'. I61. i. E. A. 7.68 but other data suggest that this figure should be higher still. Leroy Vail and Landeg White.ooo although a figure of 75.VOLUME OF THE ATLANTIC SLAVE TRADE 489 Africa exported I88. 1750-I902 (Madison. and I have substituted this figure for the one used by Anstey. This content downloaded from 73. and Allen Isaacman. Stein suspects that there were at least as many more ships from other French ports in these years.org/terms . 'Portuguese slaving from Angola'.67 I have allowed io. but I have adjusted this upward to Io. the other major port in southeastern Africa.5oo exports from southeastern Africa. Furthermore.ooo per year for i8i6 and I817 .920 in i8i9. For the Portuguese sector I estimate a total of 385. unpublished paper read to the Pacific Coast Branch of the American Historical Association. The difference is 6.ooo of Anstey's I5.ooo slaves for these sectors. 65 Anstey.8oo. The calculation of the trade in the second decade of the nineteenth century has been divided into Portuguese and non-Portuguese sectors. I803 and I 809. 271-3. I975). show that 17.

CZ C 00 "C 0 %c 00 0c) v 0. W 0 0 0 0 0 0 0 0 0 u 0 0 0 0 0 0 4-J 4 4-0 0 -0 N u 0 OC) 0 OC) co u 0 0 0 0 0 0 0 u > 00 a. h t:3 > Qk) 'j- tt 00 0 0 r. O > tn en ON 0c) 0 44 0 -0 0 0 0 0 %) oo _w *4 0 (4 O I -00 0 W1-0 4 4 q') 00 00 1. N in x? " 6 4 lls $-4 0 ..3 E-m c.0 OC) 00 N en N.4 -4 >. u C) u U qj Z r.4 0 Co 00 0-4 N -.4 1-0 0 to 00 C.org/terms C) U co u .C) :z +5 0 to CZ 00 0.34.110.75 on Sun. 0 4) 44 4Z.0 114C." 0 0 N oo OC) 1-. = - 4- C) '. t- t- 'I- 0 64 N *u 0 -2 0 0 > 0 m u 0. v 0Q -z *4 .4 N CZ -0 .4 N N sc N 1- s-4 0 00 U 4-4 CO U C) C) c._ N t.4 U u V .4 > u 0 1-0 %) :.'t ! >. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 4-J C) 6.J X X 0 4' &_4 ccnl oe_ 'S E--' -"n W' This content downloaded from 73. o "C 00 0 in . 0 00 v -4 -4 4- 0 0 -15 C) 0 0 0 0 0 0 to > z p CZ U > 4-4 1-4 4) u z u 416 16 04 to N 0...0 0 -4 ..t-J t-J 6 N. v u CZ u0 0 -4 4. oi :-. 0 tn 1.4 0. M r . LOVEJOY It ON u 0 1- 0 tt 0 0 0 0 ....4 -4 0 0 0 0 4.490 PAUL E. 0 0.jstor. 4) +4 3 0 u v v C) 4-J 00 > oo CZ u -4 CZ . -4 . x 64 E-4 00 t- CN 00 0 4-4 v I U P-4 0 0 0 0 0 0 00 0 0 0 0 0 A 0 CZ 0 - 0 0 r x 0 u -4 tn cn - u cuz 0.4 0 0 0 0 0 0 0 E-4 0 0 0 0 0 0 C) N 4u-4 C) 4-01 0 C) 0. 0 "O CZ E __4 0 0 0 0 4-0 r. 0 00 0-4 00 SIC -v > U 00 64 qj U *4 %4* $.1-.4 U Co u 4-J CZ . v oc) 4-0 oo 00 b 0 60u in iq n m in 0 0 0 C u b 'j- 14) CZ 1-0 0 0 0. -...-4 0.

It should be noted.ooo for the early years of the decade. however.000 slaves. 655.325.700: 86. 'Transatlantic slave trade I82I-I843 '. These estimates can be checked against exports in the case of the French trade. 427. These voyages may have carried some 6i. 'Illegal French slave trade'. then the distributional sample used by Eltis contains a serious error which must be explained. is 52. for the volume of imports.110. I I7. therefore.500) iS probably close to the mark.e. and it seems reasonable to allow another 8.7' A reasonable mortality rate for the middle passage would be I 5 per cent. The combined figure for the Bights (I67. most of which would have been to French possessions. 73 David Northrup.75 on Sun.300) may be too high. Census. 'The compatibility of the slave and palm oil trades in the Bight of Biafra'. 667. if it is assumed that these ships carried an average of 3I0 slaves per ship and that virtually the whole of French imports was confined to French ships (Eltis. It may be. and Daget. Eltis's work on the period from I82I to I867 is so thorough that there 70 Patrick Manning. It may well be. then there should have been an economic set back.70 The non-Portuguese sector is estimated on the basis of Curtin's importbased data for the American receiving areas other than Brazil. that only half of the French ships stopped at the French islands between I8I4 and I820. Bight of Biafra 45-4 per cent. Manning derives his estimate from the shipping data presented in Pierre Verger.73 Possibly Biafran exports were higher than 68. Uncommon Market. i (1976). Curtin's estimate may be too low by half. 423.34. the Bights of Benin and Biafra are often confused in the sources. so that the total for the Bight of Benin (99.400 into the French Caribbean. so that exports probably were I 50. although the volume of the French trade in the decade i8i 1-20 presents a problem. Imports totalled about I27. It is hoped that Daget's work will resolve some of these problems.ooo into the United States. 429). J7ournal of African History. 234. however. 'The slave trade in the Bight of Benin. therefore.400 slaves imported into the French Caribbean (234). I37-8.300 into the Spanish Caribbean. in Gemer and Hogendorn. I640-I890'. 72 Eltis. Even so. Flux et reflux de la traite des negres entre le golfe de Benin et Bahia de Todos os Santos du I7e et i8e siecles (The Hague. 288).org/terms . Curtin's I969 estimate was 3I. that Northrup is correct. Bight of Benin 26-8 per cent. 01 Jan 2017 18:51:06 UTC All use subject to http://about. and additional research may lead to a reassessment. it is doubtful that an upward revision for the Bight of Biafra would attain the level Northrup has proposed. If he is.000 slaves. The size of slave cargoes has been determined from the calculations of Eltis for the French trade from I82I-33.ooo slaves probably went to Brazil from the far western shores of Africa. using these estimators.VOLUME OF THE ATLANTIC SLAVE TRADE 49I total. 71 Curtin. and io. If Eltis is correct that the trade was much smaller than Northrup claims. since then Serge Daget has located data on I93 slaving voyages between I8I4 and I820. i. that there appears to have been no serious economic depression in the Bight of Biafra in the decade after i8io.72 It should be noted that assignment of slaves according to Eltis's model results in a much smaller total for the Bight of Biafra (68. 3I. when he estimates 338 ships carried I05. xvii. but I have adopted Eltis's distributional breakdown for I82I-25.200. Exports to Bahia from the Bight of Benin account for another 59.200) than David Northrup's speculative revision of I50.jstor. Eltis believes that this distribution is more appropriate than the breakdown for the I 82os as a whole because the rush to buy slaves for the Brazilian market before abolition took effect influenced the pattern in the late I820S. 287. This content downloaded from 73. and another io.000 slaves to the French Caribbean (p. The distribution of these slaves presents a problem. if the pattern identified by Eltis for I82I-33 also applied to this earlier period. I968). 'Export of slaves from Africa'.ooo slaves.200. 358. Western Guinea 27-9 per cent.

unpublished. 4I5.173 I 887 I.ooo slaves. principally in the East Indies. La Torre.409 Total 55. but I am basing the figures used here on the information gathered from the archives of the Dutch recruitment agency in Kumasi (P. I967).740 I 892 1. but he did not include them in his total for the Atlantic slave trade. 98.D. 'Export of slaves from Africa'.o66 I 894 2. Mass. A Question of Slavery (Cambridge. This content downloaded from 73.500 slaves who were purchased as recruits for the Dutch colonial army.664 I 897 3. who were called libertos as a ploy. The modification is miniscule.34. including the 23I3 voyages contained in the I845 British Parliamentary Paper on the trans-Atlantic slave trade.492 PAUL E.535 I 893 2.000 records. 'Wealth surpasses everything: An economic history of Asante. is little to be said in the way of comment. which in turn was a compilation of information gathered by the British Foreign Office.75 on Sun. see Joseph R. C. La Torre records I. thesis. especially its Slave Trade Department. Curtin noted the libertos in his I969 Census. James Duffy tabulated the trade between I876 and I900 (Table 7).130 I 885 2.191 I900 4.69 I i I I 888 1.786 889 1.. 01 Jan 2017 18:51:06 UTC All use subject to http://about.76 There were probably more before I876.. 4I0-5. 77 Curtin. LOVEJOY Table 7.68i I 896 2.110. 273-30I. A Question of Slavery (Cambridge. I82I-I843'.I70 slaves. The traffic in Libertos.75 The final modification is in the period after i867. The only other modification is an allowance for Dutch exports from the Gold Coast between I836 and I842 and again from I858 to I862. Census. 76 James Duffy. Berkeley.77 Nor did 14 Eltis. I978).org/terms . 1967). Eltis accounted for the engag's. 98. however: the total is about 2. when 56. when some slaves were still purchased for use on Sao Thome and Principe. I750-i874' (Ph. 75 For the Dutch trade in some of these years. Emmer has kindly shared this information with me). he did not include libertos. 'Direction and fluctuation of the transatlantic trade.510 I89I 3. 25I. It should be noted that this trade was primarily to islands off the African coast.131 890 8. I844-I867'.889 Source: James Duffy.425 I 898 3.74 While it is possible that Serge Daget's work on the French trade may result in some upward revision of the totals for the I820s and i 83os. Eltis.jstor.o66 I 899 3. I876-I90O Year Volume Year Volume I 876-84 10. and the French were involved in a similar subterfuge . and Eltis. were exported. Eltis has already accounted for most of this trade. University of California. 'Transatlantic slave trade.223 i 886 1. He compiled a data set of over 5. Mass. but since he stopped in I 867.468 I 895 2.they called the slaves engages a temps.

Adam Jones and Marion Johnson have argued that the Windward Coast. 1760s and 1770s. In Anstey's revision of Curtin's estimate for the British trade. Hence the suggested distribution presented here should be taken as a challenge to scholars to undertake more detailed work on each decade and coastal region.80 I have also taken Anstey's formula for the 1790S and applied it to the revised volume of the British trade for that decade. In this way.jstor.81 Accordingly.764 from the Bight of Biafra. in contrast to 140.79 I have also followed Curtin's formula . but I have not been able to do so for the I750s. Another modification in the regional origin of slaves relates to the elimination of the Windward Coast as a category. the size of the error is unknown. which was used in Curtin's I 969 Census and employed by virtually every scholar since i 969. xxi. hence there is likely to be considerable error in my calculations. respectively. or for Sao Thome after I 700.but I have used Inikori's tabulation for the total number of British ships involved in the trade and Anstey's slave/ship ratio. Even then. 'Measuring'. is an elusive term which meant different things to the British and French. I have disaggregated the figures for the Windward Coast and reassigned the French and British portions to the Bight of Benin and Sierra Leone. as proposed in Anstey's revision of his own calculations.34. especially the Bight of Benin. so that all estimates of coastal origin contain the possibility of considerable error.630 slaves came from the Bight of Benin. Undoubtedly a few slaves did come from the area of Liberia and western Ivory Coast previously designated the Windward Coast. The 78 Anstey. '9 Curtin. 80 See Anstey. as reference to the decade of the 1780s makes clear.org/terms . i io-I I i. i (i980). Shipping data are often confused and incomplete with respect to African destination. 606-7. There are some specific problems relating to the distribution of slaves to different parts of the African coast. but until Jones and Johnson publish more detailed material. I7-34. but they do affect efforts to gauge the impact of the slave trade on Africa. I have adjusted for the upwar'd revision in the British trade for the 1780s and 1790s. and for those decades I have relied on Anstey's first estimates of the distributional origins of British slave purchases. 'Slaves from the Windward Coast'. while the French were referring to the area to the east.110. I wish to thank Professor Curtin for his advice o this matter.VOLUME OF THE ATLANTIC SLAVE TRADE 493 he estimate the volume of trade to the other Atlantic islands in the period after i 6oo.78 While Curtin subsequently accepted Anstey's revisions for the other decades between I76I and I8io. 'British slave trade'. 'Comment'. The Bights of Benin and Biafra represent one kind of problem in delimiting the origins of slaves. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 81 Adam Jones and Marion Johnson. it may not be possible to arrive at more satisfactory estimates. I have also excluded this trade from my analysis. he still relied on the contemporary account of Norris in devising a formula for the distribution of British purchases during the I78os. Anstey estimated that only I8.applying the percentage breakdown for the British trade devised in I 969 . except for a few thousand slaves whom the French probably obtained on the Gold Coast and the eastern Ivory Coast. 13.75 on Sun. Jones and Johnson contend that the British really meant Sierra Leone and the upper Guinea coast. Journal of African History. This content downloaded from 73. These problems do not affect the estimate for the total volume of the slave trade.

and Francois Renault. M. 69.ooo. exports from Madagascar represent a problem. although there has been no attempt to estimate its total volume. T. II7). Alpers. Armstrong accounts for 4. and slavers on their way to the Americas. D. 5i6. On the order of I 3 5. 'Madagascar and the slave trade. ii. Journal of African History.ooo slaves.org/terms . although J. 01 Jan 2017 18:51:06 UTC All use subject to http://about.000 slaves per year were exported to the Mascarenes between i8I5 and I830. 84 James C.83 At least 6. I976). This content downloaded from 73. The reassignment of slaves from the Windward Coast to the Bight of Benin accounts for the difference between my estimates and those of Manning ('Bight of Benin'. The Shaping of South African Society. in Ocean Indien et Mediterran&e: Travaux du Sixieme Colloque International d'Histoire Maritime et du Deuxieme Congres de l'Association Historique Internationale de l'Ocean Indien (Lisbon. 66 per cent of whom came from Madagascar. Newitt. Liberation d'esclaves et nouvelle servitude (Paris. where the engage trade is estimated at 45.000. Filliot.. 77-84. Hardyman. I58.jstor. LOVEJOY number of slaves reassigned to the Bight of Benin is I63. I8I0-I895. This trade continued after i8io. of which 52 per cent came from Madagascar and 46 per cent came from East Africa. 'The Madagascar slave-trade to the Americas (I632-I830)'. the rest came from West Africa. Also see Gwyn Campbell.75 on Sun. where it is noted that the trade between Mozambique and the Mascarenes continued in the I 8 I os and I 82oS. although Campbell does not make any projections for total exports. 22I. Hardyman has estimated that about I2. I98I. and since these islands were developed as European plantation colonies not unlike the American colonies. and perhaps as many as 20. even though the trade was not across the Atlantic. the slave trade to the Mascarenes should perhaps be included in a discussion of the total European trade. 83 J. D.34. T. Armstrong. many of the partial totals that Curtin developed 82 Curtin. the majority between I675 and I 725 . however.000 slaves were shipped to the Americas in the seventeenth and eighteenth centuries. The Portuguese Settlement on the Zambezi (London. I979). Other slaves were brought in by private merchants. I652-I795'. Census. I50. 54. While Armstrong does not hazard an estimate for total imports. the export trade probably totalled I50. between I652 and I795.110. I963). 85 J. I652-I820 (London. assuming losses at sea of io per cent. 'The slaves.000 slaves were imported from Africa before I 8 I O. although some of these may have come from the Bight of Biafra.000 slaves were imported into the Dutch settlement at Cape Town. La Traite des esclaves vers les Mascareignes au XVIIIe siecle (Paris.300 slaves imported on ships of the Dutch East India Company.84 By far the greatest number of slaves from Madagascar went to the Mascarene islands in the Indian Otean. As I have demonstrated.000.estimated at 25.400.494 PAUL E. in Richard Elphick and Hermann Giliomee. it is clear that the number must have been considerably larger than the volume of company imports in order to account for the size of the slave population by the I790s . most of them from Madagascar. M. where it is estimated that 7. I 973).85 Total volume for the nineteenth century may well have been on the order of ioo. Ivory and Slaves. other than for the engage trade. slaves from Madagascar are not included as a category. I 974). ships returning from the East Indies. 2I4.000 between I848 and the end of the century. 203-27.82 Finally. officials trading on their own account. THE VOLUME OF SLAVE IMPORTS INTO THE AMERICAS AND ATLANTIC BASIN The impression that Curtin's census has been steadily revised upwards is based on a serious misunderstanding of the debate which has arisen since I 969. eds. I70. xxii.

who in i 86I estimated the number of imports at I3. As Curtin demonstrated in his analysis of the historiography of earlier estimates of slave imports into the Americas. Although his figure of 9.on an American publicist for the Mexican government. 86-93. and Deerr's figure of I I. Census.VOLUME OF THE ATLANTIC SLAVE TRADE 495 have been revised. 88 Noel Deerr.ooo. This content downloaded from 73. The History of Sugar (London.based on shipping data for the non-Portuguese carriers . After all.usually considered a conservative estimate .877. but such increases in single components of the trade do not necessarily modify Curtin's original estimate of the total number of slaves imported into the Americas.87 although.500) does not include the 56. and I have estimated the volume of slave imports into the Americas for my export figure (Table 8). who in i 864 calculated the volume of imports at I5. I have adjusted their import totals . Except for Deerr.34. It is now clear that Curtin was right in suspecting the import data.IOO is often used for convenience. as is evident from the preceding section of this article.500. whose imports were equated with the Portuguese trade).000 (Table 8). My calculation for total imports (9. on the other hand.org/terms . Nonetheless. or Robert Dale Owen. although he was also low on the export data.and some guesses ranged as high as 50 million and more.to arrive at export figures.before my own. 217-220).was 8 per cent lower than his calculations for imports into the Americas (other than Brazil.392.jstor.566.970.000 are the only other estimates since I969 . Dunbar. The misunderstanding has arisen from a failure to distinguish between Curtin's various estimates for different portions of the export trade from Africa. Census. II. Rawley and others. most scholars relied . 269.knowingly or not . 3-13. 01 Jan 2017 18:51:06 UTC All use subject to http://about.88 The fashionable estimate was I5 million . upward revisions of the partial figures for slave exports do not necessarily affect Curtin's total estimate for slave imports.86 My synthesis of the partial totals supports Curtin's total estimate for imports into the Americas and the islands of the Atlantic basin. Noel Deerr is the only modern scholar to advance a serious figure before I969. and his overall projection for total imports into the Americas. Inikori's figure of I3.ooo libert 86 Curtin. with an allowance for error of several hundred thousand. and Curtin. For the period I76I-i8Io his calculations for exports . Curtin states that the total was 'about ten million'. Curtin did not attempt to reach a figure for total exports from Africa based on export data because of gaps in his material.soo. there is a logical connexion between the export trade and the level of imports . but he warned that a clear solution was not possible on the evidence available in I969. 1950). Curtin. it is now possible to calculate a figure which is based largely on export data. one Edward E.the difference being losses in transit.345.000 slaves was virtually ignored by other scholars.incorporating a percentage for losses at sea .110.778. Revisions of Curtin's census have concentrated on partial totals.75 on Sun. Curtin thought that this discrepancy was more likely to be a result of double counting in the import series than in an underestimation of the volume of exports from Africa.000 and Rawley's figure of I I. 8 Curtin considered the problem of estimating the volume of the trade by comparing data on slave exports with data on slave imports (Census. He claimed that 'it is extremely unlikely that the ultimate total will turn out to be less than 8. and the unwary scholar may well have been fooled by these revisions. In order to compare the various estimates of Deerr. Inikori.ooo or more than Io.ooo . indeed some of them upwards by substantial amounts.520. since revisions can cancel each other or fill in gaps in the export data. on the one hand. 284.

I450-I867 Exports (io % Exports (i5 % Exports (20% Source Imports loss at sea) loss at sea) loss at sea) Owen 15. 'History of the rise and decline of commercial slavery in America.313.000 Curtin 9. except for some parts of the trade in the first half of the eighteenth century and the second decade of the nineteenth century.970. Drescher. James A.I8I. with reference to the future of Mexico'. since there is an overlap in the data used. 1936).110.629. 'The origin of the Diaspora: The slave from Africa'. Noel Deerr.963. 1864).jstor.278. Kuczynski.254. 428. v.000 Rawley 11.particularly Postma's work on the Dutch trade . The Wrong of Slavery: The Right of Emancipation and the Future of the African Race in the United States (Philadelphia. The Transatlantic Slave Trade: A History (New York. In particular. 284. 8. and Fogel and Engerman increased Curtin's figure for North American imports by 20.778.244.957.000 1. Dunbar. since his estimate is based on shipping data and mine is based on population data in the Americas. I969).org/terms .400. except for the addition of Vila Vilar and Fogel and Engerman. LOVEJOY Table 8.431. I2. I(I86I).496 PAUL E.ooo lower than the global total of Table i. Tarikh. For the eighteenth and nineteenth centuries.642. Population Movements (Oxford. Other revisions .500. 269-70. but the differences This content downloaded from 73.000 I6.500 11. my synthesis of the partial estimates which relies on Curtin only to fill gaps in the different series . II. Census.000 17.345. Eltis and others largely replace the earlier calculations of Curtin.000 11. the partial figures of Anstey.000 19.75 on Sun. Robert R.000 13. iv (1978). The closeness of my results to Curtin's is not completely surprising. Consequently.082.000 12. Klein. however.348.34. The Atlantic Slave Trade: A Census (Madison.000 13.359.520. Deer and others.000 Deerr 11.000 Lovejoy 9. Edward E. although the figure for total exports does not include the transport of libertos after I870 and hence is 56.confirms Curtin's original projection of total imports.500 15. J.ooo 14.000 14.650. Philip D.000 I8. Rawley. E.000 Kuczynski 14. Stein.259. because the other estimates have also disregarded these slaves.000 I8.000 Sources: Robert Dale Owen. Inikori. Kuczynski.566. 3-13.300.600 slaves to Curtin's estimate for imports into Spanish America. 1950). my figures are largely the same as Curtin's for the period before I700.338.do not affect the estimate for this period. A comparison of Curtin's Census and the figures derived from the revisions suggests that Curtin's overall estimate was probably too high for the eighteenth century and too low for the nineteenth century.600 Inikori 13.000 Dunbar 13. Miller. The Mexican Papers.392.000 17.887. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 38.000 17.000 15.606.400. Estimates of the Atlantic slave trade. My own estimates are from Tables i and 9. Postma. I981). taken to Sao Thome after i876. For a discussion of the estimates by Dunbar.I00 10. even though there are important modifications.000 14.000 I6. As explained above. see Curtin. Vila Vilar added I35. Curtin. 268.235. The History of Sugar (London.

was 7-4 per cent (Johannes Postma. in Gemery and Hogendorn. Drescher.754 slaves) and 17 per cent for 1792 (31.000 slaves (Table 9). Buxton recorded losses in the British trade at 8-75 per cent for 1791 (I5. 'Mortality in the Dutch slave trade. Slaves imported into the Americas and Atlantic Basin Period Curtin's i969 census Revision Difference 1451-i600 274. In order to estimate imports between I70I and i8io.400 I70I-I8Io 6. Table 3 for I60I -1700. 15-6 per cent for ships from the Gold Coast (based on a sample of I56 ships) and i I-o per cent from ships from Angola (based on a sample of Ioi ships).500 I60o-1700 1. AFH 23 This content downloaded from 73. based on a sample was 205 per cent.05I. The difference between Curtin's original estimate and my revision would be 179. losses were considerably less. In the Nantes trade from 1748 to 1792.000 + 354.75 on Sun. Stein.600 -314. almost cancel each other. losses averaged I5-2 per cent.89 At the height of the British trade in the last three decades of the eighteenth century. while average losses between 1740 and 1795.400 + i8. (Herbert S.700 5. Curtin did not include the 56. I have taken my figure for total exports and allowed for losses at sea on the order of i5 per cent.253. but for purposes here these variations are not significant. If I3 per cent is used. The losses at sea for French vessels in the eighteenth century were I 2-2 per cent for ships from the upper Guinea coast (based on a sample of 175 ships). Klein and Stanley L. 255). this difference is not very significant. averaging I6-2 per cent from 1715 to 1775.500 +212. see text for i 8i I-67. then the number of imports would have been 5.VOLUME OF THE ATLANTIC SLAVE TRADE 497 Table 9. Table 4 for 170I-I800 and Table 6for i8oi-Io (allowance is made for losses at sea on the order of I5 per cent).566. 277-9: T. the difference between Curtin's I969 Census and later revisions cannot be compared in a simple.872. and given the quality of the data. 268. 01 Jan 2017 18:51:06 UTC All use subject to http://about.100 I8II-I867 I. F. 'A note on mortality in the French slave trade in the eighteenth century'.200 instead of 3 I 4. Anstey and Inikori appear to increase Curtin's estimate for I70I-I8I0 by several hundred thousand slaves.org/terms .737. Census. Uncommon Market. For the revisions.778.ioo 9.jstor.400 Sources.898. Other figures are cited in Curtin. The difference for the whole trade is 2I2. Hence I have subtracted these slaves in my calculation.34. Curtin.110.554 slaves).900 293. hence a figure of 12 or I3 per cent may be more accurate. Uncommon Market. was 13-1 per cent. The average losses for 44I ships.341. in Gemery and Hogendorn.600 Total 9. Because the gross figures for slave imports are based on a combination of estimated slave populations in the Americas and shipping data.494. straightforward way. see Table 2 for 145I-I600. and losses were slightly higher in the period before 1748. which would be closer to Curtin's original estimate for the volume of imports in this period.100 I. 100 slaves. Klein and Engerman demonstrate that average losses varied with the length of voyage. Engerman.400 2. This change would mean 89 Average losses at sea between i 68o and 1749.500 + I 5 3. I675-1795'. including 9 whose coastal origin is unknown (I 2-9 per cent losses).ooo libertos transported to Sao Thome between i876 and I900 in his estimate of the slave trade. Census. based on a sample of 130 ships.500 slaves. 269). but Curtin's import-based estimates appear to have been too high by almost that amount.

and from southeastern Africa. 'Transatlantic slave trade. primarily to Brazil (allowance is made for 25 per cent losses at sea).000 slaves came from Mozambique. secondly.000 slaves came from West-Central Africa. while 50.org/terms . for . and accounts for slaves who were liberated before reaching the Americas.. Nonetheless. African exports.700 slaves imported into Spanish America. however. I82I-43.a global estimate of 9. which allows for losses at sea of i o per cent. is uncannily close to that drawn by Curtin. 91 Eltis.000. Another way to estimate the total volume of imports is to adjust Curtin's original projection by incorporating the revisions of those scholars who have presented alternative figures for imports into the different receiving areas in the Americas. the French Caribbean and the United States. Spanish America and Brazil are adjusted according to the findings of Fogel and Engerman. and while Eltis has shown that Curtin's estimates need to be raised for the years I82I-43. I82I-I843.90 Finally. Inikori criticized Curtin's estimates for French 90 Average losses at sea between I82I and I843 varied with the coastal origin and destination. But then the present researcher is not alone in having toiled through the archives only to finish up with data which largely confirm the broad estimates contained in the Census. balance Curtin's overestimation for the eighteenth century. Other changes in the estimates for losses at sea would affect the global figure. I6-7 per cent (based on a sample of I 67 ships). Another 45. This content downloaded from 73.498 PAUL E. for example.ooo compared to estimates here of S I 1.300 slaves were imported between I 844 and I867. Curtin estimated I27. from the Bights of Benin and Biafra. In reaching a total for the period after i 8io. was io-i per cent (sample of 805 ships) Eltis. Rawley attempted to follow this procedure. the estimate for the last twenty-four years of the trade requires virtually no adjustment. . The period after I820 is particularly well documented. including i 8 from unspecified origins (i8 per cent losses).000 slaves were imported. LOVEJOY that my estimate for imports would be 347. In the I 82 I-43 period.000 slaves were imported. 292). 5I2. 6-7 per cent (based on a sample of 509 ships). Revisions for the nineteenth century. Vila Vilar and Eltis. I844-I867'. following the work of Eltis. a more accurate revision of Curtin's census can be obtained. As Eltis has noted for these years: the aggregate picture. the major determinant being distance of the voyage. Losses from Senegambia averaged 15-5 per cent (based on a sample of 28 ships). I844-67. for this reason all that I am claiming is that the revisions made since I969 do not significantly alter Curtin's original global estimate. 10'2 per cent (based on a sample of 83 ships). I8I I-20. 'Direction and fluctuation of the Transatlantic trade.000 slaves were imported into Brazil from the Bight of Benin on Portuguese ships. this revision must be considered tentative.300 slaves greater than Curtin's figure . from western-central Africa.110. The latter estimated total imports in this period at Sio.91 I have added another few hundred to account for Dutch exports from the Gold Coast.300. At present. but such a minor adjustment only serves to confirm Eltis's sense of the uncanny. The changes do require adjustments in the distribution of exports over time. Another 2 I 8. I have estimated imports for three distinct periods: firstly. and thirdly. too. This figure is derived as follows.34.750.75 on Sun. 01 Jan 2017 18:51:06 UTC All use subject to http://about.9I3. for I have not attempted to assess the volume of imports into the majority of receiving areas. if the figures for North America. however. I.jstor. The average for all ships. In the first period 44I. but he confused the import totals for different colonies with the carrying trade of the various European countries.

268. 30-I. 96 Curtin.96 As discussed above. 01 Jan 2017 18:51:06 UTC All use subject to http://about. 29I. despite Rout's claim to the contrary. Eltis has suggested that Curtin's import data for I821-43 is too low. Comercio de esclavos.92 On the other hand. hence the difference between the revisions of Eltis and Curtin's original calculation is about 340. Present revisions of the estimates for slave imports into different receiving areas suggest that Curtin's figure should be raised by 6oo.75 on Sun. Until additional research is done on the volume of imports into the different receiving areas. 'Transatlantic slave trade. I have estimated that 441.000 additional slaves. and it provides some confirmation that my global total is of the right order of magnitude. As a rough guide. The Bight of Benin became a second major source 92 Inikori. Evidence and Methods.898. which is very close to the estimate suggested in Rawley's calculation. still within the range of Curtin's original projection. Time on the Cross. Again it must be stressed. etc.95 Eltis has estimated 5 I 1.org/terms . for a total of over i-8 million for I82I-67. of course. CONCLUSION There is no question that the Atlantic slave trade had a tremendous impact on Africa. and it may well be that a thorough investigation of imports into the French Caribbean will result in an upward revision of Curtin's estimate.ooo higher than Curtin allowed.34.2I0.400 slaves imported from i8i i to i870. 182I-I843'.VOLUME OF THE ATLANTIC SLAVE TRADE 499 Saint-Domingue.94 Finally. since Curtin and Eltis made estimates for different periods. a review of Spanish America from i 640 to I 773 may require a downward revision of Curtin's estimate. this impact can be quantified in broad outline for virtually the whole of the trade. 94 Fogel and Engerman.jstor.) after i6oo. 202-4. Curtin estimated 1. 95 Eltis. and Inikori.ooo slaves. 617-22. that my synthesis is only the next stage in the continuing effort to calculate the volume of the Atlantic slave trade.600 slaves imported into Spanish America between I 595 and i 640. therefore. Despite the probable error in the regional breakdown of slave shipments. Britain and elsewhere in the eighteenth century and the trade to the offshore islands (Madeira. 93 Vila Vilar. Further revisions could well require the addition of another hundred thousand slaves or so in order to allow for the trade of the minor ports in France. Vila Vilar has accounted for an additional I 35.000 slaves were imported in the second decade of the nineteenth century. 'Rejoinder'. 18-2 This content downloaded from 73. Census. Western central Africa (the Angolan coast north to Cabinda) was drawn into the trade on a significant scale in the sixteenth century and remained a major exporting region until the end of the trade in the nineteenth century. where Eltis estimates that his total for I82I-43 is 30 per cent higher than Curtin's figure. once his tables are corrected for errors.750 slaves imported from I844-67 and just under I 3 million slaves from I82 I-43.000 slaves.110.93 Fogel and Engerman have established that North American imports were probably i 68. but my calculation suggests that Eltis is about I9 per cent higher. however. 'Assessment of Curtin and Anstey'. although it is difficult to establish a figure. therefore. 206-9. my global estimate for imports (9-8-9-9 million slaves) is preferred. an estimate for the slave trade derived from import-based data is approximately I0. This rough estimate is. Cape Verde.

34. 'Introduction'.97 Fage has emphasized the political impact while attempting to minimize the demographic consequences. place in total exports until the nineteenth century. While these broad patterns have been understood for some time.99 I have argued that the growth of the export trade was related to the consolidation of a mode of production based on slavery. and some scholars . Senegambia entered the trade early and retained a relatively fixed. particularly because of severe demographic losses. Various scholars have attempted to analyse the nature of this distortion: Walter Rodney talked about the process in terms of underdevelopment. 98 Fage. 99 Inikori. Southeast Africa. and instead relies on Curtin's study. For the time being. The Sierra Leone coast. by contrast. History of Africa. does not attempt a new estimate for the volume of the trade. now that Curtin's export figures have been revised and tested. tify the trans-Atlantic slave trade. but small.notably 91 Walter Rodney.100 These examples are but a few of the many studies which take the 'numbers game' seriously. which is a major reason why the study of slavery retains its significance. 101 Edward Reynolds. is is now possible to analyse internal African developments much more fully. I982). was drawn in relatively late. 01 Jan 2017 18:51:06 UTC All use subject to http://about. including the rivers to the north. while the supplies from the Bight of Biafra fell off in the I83os.jstor.500 PAUL E. his analysis of the impact of the slave trade on Africa is an important aspect of his study. there have been numerous revisions of different sectors of the trade. This content downloaded from 73.98 Inikori has argued that the trade retarded economic development. How Europe Underdeveloped Africa (London. it can be expected that detailed studies of political. 244-88.75 on Sun. Nonetheless. however. Scholars are likely to modify these estimates further. The Gold Coast ceased to supply large numbers of slaves by the first decade of the nineteenth century.101 Slaves were more than commodities and numbers. 13-6o. only after the movement to end the slave trade began to have an impact was it advantageous to seek slaves so far from the Americas. although the expansion of slave exports there grew relatively slowly until the middle of the eighteenth century. SUMMARY This article provides a synthesis of the various studies which attempt to quan-.110. Because slaves were a peculiar commodity.org/terms . The decline of slave exports occurred in parallel stages to the rise of the trade. Transformations in Slavery. Stand the Storm: A History of the Atlantic Slave Trade (New York. and from there the trade spread westward to the Gold Coast by I700 and eastward to the Bight of Biafra by the I740s. LOVEJOY of slaves in the second half of the seventeenth century. was another focal point of the export trade. the export trade distorted the historical process in an equally peculiar way. 100 Lovejoy. Curtin's pioneering estimates in i 969 (The Atlantic Slave Trade: A Census). when for two decades exports were very large. economic and social changes will attempt to correlate the volume of exports with local developments. Since the publication of Philip D. and southeastern Africa also continued to supply exports until the end. 1972). of course. The Bight of Benin and western-central Africa remained important sources of slaves until the I85os. representing labour power lost to local societies as well as a commodity whose value could be exchanged on the international market.

Inikori and Rawley have failed to distinguish clearly between imports by colony and exports by national carrier. while imports into the Americas and most other parts of the Atlantic basin are estimated to have been 9-8-9-9 million slaves . Further revisions are likely. economic and social repercussions on Africa. Many of the revisions are based on shipping data by national carrier. hence their global estimates have resulted in double counting.ooo slaves. both to test the regional breakdown of slave exports and to assess the demographic. the current state of research on the volume of the Atlantic slave trade is summarized in a series of tables which analyse the export trade by time period.110.VOLUME OF THE ATLANTIC SLAVE TRADE 50I J.well within range of Curtin's original Census. but until the completion of detailed research comparable to the studies of David Eltis. In the meantime. E. This content downloaded from 73.have argued that Curtin's global estimate for imports into the Americas is too low. and coastal origin. It is expected that the present synthesis will challenge historians to examine the impact of the slave trade on different parts of Africa. national carrier.hardly significant considering the quality of the data. it is apparent that Curtin's initial tabulation was remarkably accurate. The volume of exports from Africa across the Atlantic is here calculated at I I. political.34. 01 Jan 2017 18:51:06 UTC All use subject to http://about. rather than on series derived from estimated imports into different colonies in the Americas. The results of these substitutions shift the distribution of slave exports over time but do not affect estimates of the relative scale of the trade by more than 2-3 per cent .698. and a dozen other scholars it is not possible to estimate the extent of future modifications.jstor. Roger Anstey. Hence it is possible to substitute new data for much of the eighteenth and nineteenth centuries for many of the import-derived series used by Curtin.org/terms . Inikori and James Rawley . nonetheless. When the revisions are examined carefully.75 on Sun. however. Johannes Postma.