Hostname: page-component-7d684dbfc8-4nnqn Total loading time: 0 Render date: 2023-09-27T12:20:13.972Z Has data issue: false Feature Flags: { "corePageComponentGetUserInfoFromSharedSession": true, "coreDisableEcommerce": false, "coreDisableSocialShare": false, "coreDisableEcommerceForArticlePurchase": false, "coreDisableEcommerceForBookPurchase": false, "coreDisableEcommerceForElementPurchase": false, "coreUseNewShare": true, "useRatesEcommerce": true } hasContentIssue false

New Evidence on the Causes of Slave and Crew Mortality in the Atlantic Slave Trade

Published online by Cambridge University Press:  03 March 2009

Richard H. Steckel
Associate Professor of Economics, Ohio State University, Columbus, Ohio 43210
Richard A. Jensen
Assistant Professor of Economics, Ohio State University, Columbus, Ohio 43210.


The journals of slave ship surgeons of the 1790s are used to address questions on the relative importance of African conditions versus those on ships, crowding, the effectiveness of Dolben's Act, and the interaction between slave and crew health. In contrast with previous work we find that most slaves who died did so near the middle of the voyage. Crowding was important to health and mortality, but the restrictions of Dolben's Act did little to reduce losses. The crew was largely isolated from patterns of disease among slaves.

Copyright © The Economic History Association 1986

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)


1 See Lambert, Sheila, ed., House of Commons Sessional Papers of the Eighteenth Century (Wilmington, 1975), vols. 6773.Google Scholar

2 Curtin, Philip D., The Atlantic Slave Trade: A Census (Madison, 1969):Google ScholarKlein, Herbert S. and Engerman, Stanley L., “Slave Mortality on British Ships. 1791–1797,” in Anstey, Roger T. and Hair, P. E. H., eds., Liverpool, the African Slave Trade and Abolition (Liverpool 1976). pp. 113– 25;Google ScholarKlein, Herbert S., The Middle Passage (Princeton, 1978);CrossRefGoogle ScholarPostma, Johannes, “Mortality in the Dutch Slave Trade, 1675–1795,” in Gemery, Henry A. and Hogendorn, Jan S., eds., The Uncommon Market: Essays in the Economic History of the Atlantic Slave Trade (New York, 1979), pp. 239–72;Google ScholarEltis, David, “Mortality and Voyage Length in the Middle Passage: New Evidence from the Nineteenth Century,” this Journal, 44 (06 1984). pp. 301–8.Google ScholarPubMed

3 Miller, Joseph C., “Mortality in the Atlantic Slave Trade: Statistical Evidence on Causality,” Journal of Interdisciplinary History, II (Autumn 1981). pp. 317–29.Google Scholar

4 Cohn, Raymond L. and Jensen, Richard A., “The Determinants of Slave Mortality Rates in the Middle Passage,” Explorations in Economic History, 19 (Autumn 1982), pp. 269–82.CrossRefGoogle ScholarPubMed

5 A total of 7 slaves in the sample were returned to shore as sick. These unloadings were arbitrarily counted as deaths.Google Scholar

6 The estimated relationship is t-values are given in parenthesis. Simple nonlinear functional forms fit little or no better than the simple linear form. The available muster rolls show that relatively few crew members left the ship or were taken on board before the ship reached the West Indies. As an approximation it was assumed that crew size changed only through deaths. Thus the estimated number of crew at risk may be inaccurate for particular vessels.Google Scholar

7 The terms death rate and mortality rate are sometimes used loosely and interchangeably. According to usage by demographers death rates are based on the average number of person-years at risk during a time period and mortality rates are based on the number at risk at the beginning of a time period. If there is little change in the number of persons at risk during the interval, the difference between the measures is small. Given the loading and unloading of slaves it is appropriate to use death rates as opposed to mortality rates in this paper. For a discussion of terminologyGoogle Scholarsee Shryock, Henry S. and Siegel, Jacob S., The Methods and Materials of Demography (Washington, D.C., 1975), vol. 2, chap. 14.Google Scholar

8 The potential for contaminated water suggests that typhoid may have contributed to deaths from fevers. Typhus is an unlikely cause of fevers because it is usually associated with blankets and clothing characteristic of colder climates.Google Scholar

9 Symptoms of tropical diseases are discussed in Strickand, G. Thomas, Hunter's Tropical Medicine (Philadelphia, 1984);Google Scholarand Felsenfeld, Oscar, The Epidemiology of Tropical Diseases (Springfield, III., 1966).Google ScholarHoeppli, R., Parasitic Diseases in Africa and the Western Hemisphere: Early Documentation and Transmission by the Slave Trade (Basel, 1969), p. 62 notes that as late as the early nineteenth century dysentery was thought to be but one of the symptoms of malaria.Google ScholarPubMedSee also Lloyd, Christopher and Coulter, Jack L. S., Medicine and Navy, 1200–1900, vol. 3 1714–1815 (Edinburgh, 1961);Google Scholarand Scott, H. Harold, A History of Tropical Medicine, (London, 1939), 2 vols.Google Scholar

10 The absolute levels of crew death rates were in the low to middle range of magnitudes reported by others for European populations residing in western Africa during the eighteenth and nineteenth centuries. See Curtin, Philip D., “The White Man's Grave: Image and Reality,” Journal of British Studies, 1 (11 1961), pp. 94110;CrossRefGoogle ScholarThe Image of Africa: British Ideas and Action 1780–1850 (Madison, 1964);Google ScholarDavies, K. E., “The Living and the Dead: White Mortality in West Africa. 1684– 1742,” in Engerman, Stanley L. and Genovese, Eugene D., eds., Race and Slavery in the Western Hemisphere: Quantitative Studies (Princeton, 1975), pp. 8398;Google ScholarFeinberg, H. M., “New Data on European Mortality in West Africa: The Dutch on the Gold Coast, 1719–1760,” Journal of African History, 15 (1974), pp. 357–71.CrossRefGoogle ScholarPubMed

11 Data in Mitchell, B. R., European Historical Statistics, 1750–1970 (New York, 1975). table B6, show that crude death rates in European populations of the late 1700s and early 1800s commonly fell in the interval of 20 to 30 per 1000.CrossRefGoogle ScholarThe model West level 7 life tables of Coale, Ansley J. and Demeny, Paul, Regional Model Life Tables and Stable Populations (Princeton, 1966) has crude death rates in the neighborhood of 30 to 35 per 1000 (given reasonable assumptions of the range of European birth rates) and is a conservative (high mortality) benchmark for these populations. The average of the age-specific mortality rates for males in the level 7 table for age groups 10–15 to 30–34 is 10.4 per 1000. Soldiers not engaged in combat are a reference group against which the experiences of slaves and crew can be compared. The annual death rate among Dragoon guards serving in the United Kingdom from 1830 to 1836 was 14 per 1000 from disease and 1.3 per 1000 from suicide, murder, and accidents.Google ScholarSee Parliamentary Papers, Statistical Report on the Sickness. Mortalit, and Invaliding Among the Troops in the United Kingdom (London, 1839), p. 4.Google Scholar

12 The model West level I life table has crude death rates in the neighborhood of 47 to 51 per 1000 (given reasonable assumptions about the range of European birth rates). In this table the average of the age-specific mortality rates for males in the age groups 10–14 to 30–34 is 18 per 1000.Google Scholar

13 The “other” category includes well-defined but infrequent ailments such as yaws and dropsy, several poorly described diseases such as “sudden death,” “sulks,” “fits,” “delirium,” and “relaxed,” and many illegible entries.Google Scholar

14 See Lloyd and Coulter, Medicine and the Navy, pp. 293–328 for a discussion of the conquest of scurvy.Google Scholar

15 Ibid, pp. 348–52. Klein, The Middle Passage, p. 229, notes that inoculation pervaded the slave trade by the second half of the eighteenth century.Google Scholar

16 Dawson, Marc H., “Smallpox in Kenya, 1880–1920,” Social Science and Medicine, 13B (Dec. 1979), pp. 245–50;Google ScholarHerbert, Eugenia W., “Smallpox Inoculation in Africa,” Journal of African History, 16 (1975), pp. 539–59.CrossRefGoogle ScholarPubMed

17 The proximate determinants of an epidemic include the probability of contact between an infectious agent and individuals and the susceptibility of individuals. The amount of infectious material in the environment is a function of the number of infected individuals or hosts of the agent, the duration and extent to which infectious material is expelled by hosts, the rate of survival of the agent in the environment, the route of entry into the host, and the existence of alternative reservoirs or hosts of the agent. The availability of hosts varies with the mobility and interpersonal contact within the population group and the immunity or other means of resistance to the agent by individuals.Google ScholarSee Lilienfeld, Abraham M., Foundations of Epidemiology (New York, 1976). chap. 3;Google Scholarand MacMahen, Brian and Pugh, Thomas F., Epidemiology: Principles and Methods (Boston, 1970), chaps. 1–4.Google Scholar

18 Curtin, “White Man's Grave,” pp. 95–97.Google Scholar

19 The share of deaths occurring on days in which there were two or more deaths was so low that it justifies the use of a logit model as an approximation for calculating probabilities of death. Only 3.3 percent of crew deaths from fevers, for example, occurred on days in which there were two or more deaths. Among the eight data sets used, the largest share of deaths occuring on days in which there were two or more deaths (12.7 percent) involved slaves dying from gastrointestinal diseases during the voyage.Google Scholar

20 Strickland, , Hunter's Tropical Medicine; Russell, Paul F., West, Luther S., Manwell, Reginald D., and MacDonald, George, Practical Malariology (London, 1963).Google Scholar

21 It should be recognized that the large number of possible regional determinants of mortality and the lack of information about most of them will make it difficult to identify specific sources locational differences in mortality that may be found.Google Scholar

22 Strickland, , Hunger's Tropical Medicine, pp. 279–282; Felsenfeld, Epidemiology, pp. 130–34.Google Scholar

23 Bruce-Chwatt, Leonard Jan, Essential Malariology (London, 1980). pp. 117–20:Google ScholarRussel et al., Practical Malariology, pp. 237, 239–40.Google ScholarGreater incidence of fevers during the rainy season noted in Miller, Joseph C., “The Significance of Drought, Disease, and Famine in the Agriculturally Marginal Zones of West-Central Africa,” Journal of African History, 23 (1982), p. 23.CrossRefGoogle ScholarPubMed

24 The probabilities were calculated from the regression coefficients and the logistic function by evaluating other independent variables at their sample mean. Similar calculations giving relative risks of deaths for other variables can be done using the sample means given in the appendix Steckel, Richard H. and Jensen, Richard A., “Determinants of Slave and Crew Mortality in the Atlantic Slave Trade,” NBER Working Paper No. 1540 (Jan. 1985).Google Scholar

25 Seasonal patterns of rainfall are from Oliver, Roland and Crowder, Michael, The Cambridge Encyclopedia of Africa (Cambridge, 1981), p. 43.Google Scholar

26 Infective mosquitoes may have been sufficiently prevalent year-round that the chances of infection varied little by season of the year. It is also possible that infective mosquitoes were more prevalent during the rainy season and that patterns of activity, such as expeditions to the interior, were curtailed in ways that offset the greater risk of infection. Experiments with other variables include the price of sugar as a proxy for the price of slaves, a measure of association between slave and crew deaths, measures of duration that slaves had been on board, quadratic terms in number of slave or crew, the age and sex ratios of slaves, recent loading of slaves, year of activity, and month of the year. These variables were statistically insignificant with the exceptions that females and children had greater chances of deaths from gastrointestinal diseases, the incidence of slave deaths from gastrointestinal diseases declined over time, and crew deaths from fevers were lower during the dry season months of March and April (north of the equator).Google Scholar

27 See Lambert, , House of Commons, vol. 69, pp. 116, 121, and 142. The small sample size limited regional comparisons involving the crew to Biafra versus other regions. The negative coefficient for crew size in gastrointestinal diseases is unexpected. Errors in crew size estimates may have contributed to this result, but it does not register in this way in other regressions.Google ScholarThe coefficient is consistent with evidence in Dow, George Francis, Slave Ships and Sailing (Salem, Mass., 1927), pp. 8283 that some crew members on larger ships specialized in cleaning the vessel.Google Scholar

28 The last 13 days of observations for voyages going to Jamaica are not included in the regressions on the grounds that provisions acquired from a landfall on one of the outer islands may have lessened the comparability with non-Jamaica-bound voyages. Thus all voyages “ended” when a landfall was made.Google Scholar

29 Curtin, “White Man's Grave,” p. 95.Google Scholar

30 Strickland, Hunter's Tropical Medicine, p. 534.Google Scholar

32 Experiments conducted with other variables include the ratio of actual to expected voyage length, dummy variables representing interaction of slave and crew deaths from gastrointestinal causes within the previous week, measures of average duration that slaves had been on board, age and sex ratios, and quadratic terms in crew size and number of slaves. The latter four experiments were also conducted for loading and the results were similar with the exceptions that death of a slave from a gastrointestinal disease had a marginally significant (t = 1.85) positive influence on the chances of crew deaths and that the shares of females and children had positive but not statistically significant influences on gastrointestinal diseases. Possible interpretations of the interaction between slaves and crew are that crew health, and therefore resistance, deteriorated from loading to the voyage, and that the amount of contamination from slaves was greater during the voyage because slaves ill with dysentery could not be sent on sore. Females and children vulnerable to disease while loading may have acquired immunities and possibly more food by the time of the voyage. Mortality rates may have risen if rations were cut on unusually long voyages. However, the ratio of actual to expected length, interacted with day of the voyage or entered alone, had no systematic influence on crew or slave deaths from fevers or gastrointestinal diseases. This result is not surprising in view of the findings, shown in Figure 2, that deaths from the diseases in question were concentrated in the middle of the voyage. Expected length was calculated from the following regression: Senegambia, Sierra Leone, and windward Coast were omitted because there was no statistically significant difference in voyage length from these areas. The data source is the Return to an Order of the Right Honorable the House of Lords dated the 10th of July 1799, directing that the clerk of the Parliments do cause to be extracted from the several log books and Journals of Ships employed in the Slave Trade in each year 1791 to 1797 …, dated 28 Jult 1800. Only those observations from this source having all relevant information on tons, origin, destination, and dates were used.Google Scholar

33 Miller, “Mortality in the Atlantic Slave Trade.”Google Scholar

34 Cohn and Jensen, “The Determinants of Slave Mortality.”Google Scholar

35 Klein and Engerman, “Slave Mortality on British Ships”; Klein, The Middle Passage; Postma, “Mortality in the Dutch Slave Trade”; Eltis, “Mortality and voyage Length.”Google Scholar

36 Lambert, , House of Commons, vol. 68, p. 317; vol. 69, pp. 129, 142, 151, 161; vol. 71, p. 55.Google Scholar

37 Kiple, Kenneth F. and King, Virginia Himmelsteib, Another Dimension to the Black Diaspora: Diet, Disease, and Racism (Cambridge, 1981) discusses African diseases.CrossRefGoogle Scholar

38 One of the contributions of the study of heights has been to alert i esearchers to the nature and extent of interaction between nutrition and disease. See, for example, Steckel, Richard H., “Slave Height Profiles from Coastwise Manifests,” Explorations in Economic History, 16 (10 1979), pp. 363–80;CrossRefGoogle ScholarPubMedand Fogel, Robert W., Engerman, Stanley L., Floud, Roderick et al. ,, “Secular Changes in American and British Stature and Nutrition,” Journal of Inrerdisciplinary History, 14 (Autumn 1983), pp. 445–81.CrossRefGoogle ScholarPubMed

39 Lambert, , House of Commons, vol. 68, pp. 19, 33, 294; vol. 69, pp. 117, 119. 124, 125; vol. 71, p. 28; vol. 72, pp. 203, 205.Google ScholarSee also Sheridan, Richard B., Doctors and Slaves: A Medical and Demographic History of Slavery in the British West Indies, 1680–1834 (Cambridge, 1985), pp. 113–26.CrossRefGoogle ScholarScreening was important for control of bacillary dysentery because, for practical purposes, humans are the only hosts of the disease. Thus dysentery could not have been “created” on the ship from spoiled food or water. Dysentery may have been brought on board by slaves who had the disease but did not yet exhibit symptoms; the incubation period is about 1to 6 days. See Levine, Myron M., “Bacillary Dysentery: Mechanisms and Treatment,” Medical Clinics of North America, 66 (May 1982), pp. 623–38.CrossRefGoogle ScholarPubMed

40 Ibid., vol. 69, pp. 117, 119; vol. 71, p. 30; vol. 72. p. 302.Google Scholar

41 Ibid., vol. 69, pp. 120, 127; vol. 71, p. 47.Google Scholar

42 African-born slaves living in the Caribbean were 3 to 4 inches shorter as adults than slaves living in the United States, which suggests that the African diet was poor. Data and discussion of heights are available in Margo, Robert A. and Steckel, Richard H., “The Heights of American Slaves: New Evidence on Slave Nutrition and Health,” Social Science History, 6 (Fall 1982). pp. 516–38;CrossRefGoogle ScholarPubMedHigman, Barry W., Slave Populations of the British Caribbean, 1807–1834 (Baltimore, 1984), pp. 280–92.Google Scholar

43 Klein and Engerman, “Slave Mortality on British Ships,” p. 118.Google Scholar

44 Based on annual data given in Anstey, The Atlantic Slave Trade, p. 31.Google Scholar

45 Data on slave prices are available in Bean, Richard Nelson, The British Trans-Atlantic Slave Trade, 1650–1775 (New York, 1975), P. 76;Google ScholarDickson, William, Mitigation of Slavery (London, 1814), pp. 259–60.Google Scholar

46 Dow, Slave Ships, p. 86 discusses the lack of space for food on slave ships.Google Scholar

47 The point on voyage length has been made by Cohn, Raymond L., “Deaths of Slaves in the Middle Passage,” this Journal, 45 (09. 1985), pp. 685–92. Information on time spent in loading in an earlier era of the trade is available in Postma, “Mortality in the Dutch Slave Trade”.Google ScholarPubMed

48 Smith, Robert S., “Compensating Wage Differentials and Public Policy: A Review,” Industrial and Labor Relations Review, 32 (April. 1979), pp. 339–52.CrossRefGoogle Scholar

49 Lambert, , House of Commons, vol. 69, pp. 160, 171, 173; vol. 71, pp. 39, 55. Data collection for study of these issues is in progress.Google Scholar

50 Some work has been done in this area. See, for example, Riley, James C., “Mortality on Long-Distance Voyages in the Eighteenth Century,” this Journal, 41 (Aug. 1981), pp. 651–56;Google ScholarPubMedCohn, Raymond L., “Mortality on Immigrant Voyages to New York, 1836–1853,” this Journal, 44 (06 1984), pp. 289300;Google ScholarPubMedGrubb, Farley, “Mobility and Mortality on the North Atlantic Passage: Evidence from Eighteenth-Century German Migration” (unpublished manuscript, University of Delaware, 1984).Google Scholar

51 The text of the act is given in Donnan, Elizabeth, Documents Illustrative of the History of the Slave Trade to America (Washington, 1931), vol. 2, pp. 582–89.Google ScholarSanderson, F. E., “The Liverpool Delegates and Sir William Dolben's Bill,” Transactions of the Historic Society of Lanschire and Cheshire, 124 (1974), pp. 5784 discusses the debate.Google Scholar

52 If more than 40 percent of the slaves were children (defined as no more than 4 feet 4 inches) then every five children over the 40 percent counted as four slaves.Google Scholar

53 Judy Collingwood conducted the search. Logs that failed to give dates of slave deaths were ignored. Information sometimes missing from the available papers includes dates of loading (2 logs), crew deaths (27 logs), destination (I log), causes of crew deaths (5 logs out of the 65 containing crew deaths), and causes of slave deaths (9 logs).Google Scholar

54 There is one known exception. The surgeon on a ship for which the day of arrival is known continued to record deaths for nearly two weeks after arrival.Google Scholar

55 Incentives are discussed in Anstey, The Atlantic Slave Trade, p. 34.Google Scholar

56 Information on the time trend of mortality is available from Klein, The Middle Passage; Eltis, “Mortality and Voyage Length”; Curtin, Atlantic Slave Trade, chap. 10; Cohn, “The Trend.”Google Scholar