SOURCES AND METHODOLOGY

The Middelburgsche Commercie Compagnie (hereafter MCC) was a Dutch company that dispatched some 118 slaving voyages to Africa between the 1730s and 1790s. Its records are among the most detailed of all slave-trading groups. They have long attracted the attention of students of the slave trade, but to our knowledge they have not previously been used to reconstruct the day by day slave purchase and slave mortality patterns that the high survival rate of logs, journals, and other papers for MCC ships makes possible. Records of this sort exist for 70 (or 60 percent) of known MCC slaving voyages; of these 39 are for voyages destined for West Africa or “Guinea” and 31 for West-Central Africa, principally the Loango Coast. We shall focus here on the Guinea voyages. “Guinea” in this context is defined as the coastal area between the River Senegal in the north and Cape Lopez in the south and comprises Senegambia, Sierra Leone, the Windward Coast, the Gold Coast, the Bight of Benin (or Slave Coast), and the Bight of Biafra.⁴

The Appendix summarizes the data on the names of the ships, masters, and surgeons as well as on ships' tonnages and voyage schedules for the 39 voyages included in this study. Thirty-six of the 39 voyages were dispatched before 1780, the remaining three in 1788–1794. All but one ship returned home to Middelburg after journeying to Africa and thence to America with slaves. Even the one ship that failed to complete its triangular voyage—the Zang Godin, which sailed in 1773—delivered slaves to America before being abandoned as unseaworthy. Fourteen different ships made these 39 voyages, with just over half the voyages (20 out of 39) being made by just four ships—the Willem V, Philadelphia, Johanna Cores, and Nieuwe Hoop. Concentration of management of voyages was much lower, however, for no fewer than 20 different captains were involved in the voyages. Only two made more than two voyages in this capacity. Of the 29 surgeons involved, only one made more than two voyages. Such patterns of management do not appear unusual in the eighteenth-century slave trade.⁵

By contrast, the mean time taken to complete the MCC voyages in our sample group—572 days (range 357–1,370)—was seemingly longer than that of contemporary British and French traders.⁶ One important factor determining this was time spent at the African coast, a point to which we return later.

In common with other ships logs, those of the MCC provide daily reports on ship location, wind strength and direction, and rainfall from the start until the conclusion of voyages. What distinguishes the MCC records from those of other traders is that logs are often accompanied by trade books that provide daily accounts of slave purchases, deaths, and sales. In almost every case the age and sex of the slaves concerned is noted. Interspersed with these data are accounts of coastal purchases of water, food, wood, and other consumables as well as periodic inventories of slaves on board, purchased and dead. The last provide useful cross-references for our own calculations of the same. For the most part, the causes of the death of slaves are not reported, but there are anecdotal reports of accidental deaths, of suicides, of uprisings, and of epidemics, notably of smallpox, scurvy, and diarrhea. Reports relating to the crew paralleled those relating to slaves.⁷

Combined into one dataset, the MCC records permit a day-to-day reconstruction of life and death on 39 slaving voyages to Guinea in the second half of the eighteenth century. The total number of enslaved Africans involved in these voyages was 9,880. To facilitate analysis of their experience, we divided voyages into five phases: (1) the journey from Middelburg to the first place of purchase of slaves; (2) the purchase or trading period in Africa (hereafter the “loading phase”); (3) the sailing time from last place of purchase in Africa to the principal place of slave sale in the Americas (conventionally known as “the middle passage,” but hereafter the “crossing phase”); (4) the time spent in the Americas selling slaves and obtaining a return cargo (or “selling phase”); and (5) the homeward voyage. In this article we concentrate on phases (2) to (4) when slaves were on board ship.

Following Curtin's pioneering census of the slave trade published in 1969, historians have typically divided trading locations in Atlantic Africa into seven regions—Senegambia, Sierra Leone, Windward Coast, Gold Coast, Bight of Benin, Bight of Biafra, and West Central Africa.⁸

We have followed the same procedure in analyzing the data of our 39 MCC voyages, one of which, though initially destined for Guinea, completed its loading in West Central Africa. Where the specific location of any slave purchase was not given, we assumed this was the same as those for immediately adjacent days of slave purchase. The key variables used to track daily events through time were location (or port); number of slaves traded and number of slaves died, each by age and sex; number of crew entering or leaving the ship; and number of crew deaths. Data describing location were cross-matched between the logbooks and trade journals. Where there was conflict between sources, preference was given to data from trade journals. Cross-matching slave purchase data with muster rolls and ship specifications allowed us to calculate packing levels of slaves through time and to assess the experience of particular shipmasters and surgeons in handling slaves.

For the reconstruction of patterns of slave mortality on board ship one ideally requires data allowing one to construct “life tables” describing the number of days on board ship from initial purchase to sale or death of each individual slave in the sample. Such information is available for only one voyage in our sample.⁹

The next best option is to calculate daily hazards and exposure-based death rates of slaves. Daily hazard is defined as the number of slave deaths per day relative to the total number on board on that particular day. It can be calculated for both slaves and crew and in the case of slaves disaggregated for both age and sex. Exposure is the sum of the days spent on board ship by a group of people expressed in person-years (or py). This provides the denominator for a calculation of the true death rate on board ship, reported as the number of deaths per thousand person-years exposure (or d/py). The use of true death rates in captivity allows for a standardized comparison of slave and crew mortality across different groups and periods. It is also to be distinguished from the percentage loss in transit, which relates to the total number of slave deaths as a proportion of slave purchases irrespective of the length of the voyage or the time spent on board ship and which has been the primary measure of slave mortality used in most studies of shipboard mortality since the late 1960s.

SLAVE PURCHASING STRATEGY

We begin our analysis by looking at the time spent on board ship by slaves and at the slave purchasing strategies of MCC shipmasters that underlay it. In investigating these issues we use pooled data for 9,880 slaves purchased across the 39 voyages (mean purchase 253 slaves, range 71–390). As a group, these 9,880 slaves were actually on board ship for 61.3 percent (or 13,646) of the 22,238 days taken to complete the voyages. Put another way, the typical MCC slaving voyage to Guinea in 1751–1797 lasted some 570 days (or 19 months) (95 percent confidence interval (hereafter CI) 513–627 days), during which the ship would have some slaves on board for 350 days (95 percent CI 316–383 days), or just under a year. The last figure includes 223 days (95 percent CI 199–247 days) in the loading phase, 71 days (95 percent CI 65–77 days) in the crossing phase, and 56 days (95 percent CI 42–70 days) in the selling phase. The mean length of time spent by MCC ships in the crossing does not appear exceptional in the period 1750–1775, but the company's ships do seem to have spent much longer than those of other carriers acquiring slaves in Africa.¹⁰

To our knowledge, however, there are no data currently available on the ratio of days with slaves on board to total voyage days. This ratio—roughly six out of ten days in these 39 MCC voyages and, when allowance is made for longer loading times by MCC ships, perhaps five out of ten days on most other voyages—underlines the need for historians to reconsider the concept of the “middle passage” to include the time spent by slaves on board ship before departing the shores of their homeland. More specifically, the fact that MCC ships typically took up to six or even eight months in 1751–1797 to acquire a shipload of slaves requires us to take account of the pace at which slaves were acquired in Africa when seeking to assess life and death on board slave ships.¹¹

It is possible to ascertain the African coastal place of purchase of over 99 percent of the slaves acquired by the 39 MCC voyages in our sample. All but one of the ships began their purchase of slaves at Upper Guinea which comprised the subregions of Senegambia, Sierra Leone, and the Windward Coast west of the Gold Coast; the Windward Coast was by far the most common initial trading destination. Upper Guinea was, however, not the only source of slaves for these MCC ships, for while typically starting trade there, most went on to complete their purchase of slaves further south, with the Gold Coast being the preferred destination for the great majority. This pattern is underlined by the data shown in Table 1, which show that of the 9,880 slaves bought by MCC ships, 6,718 (or 68.0 percent) were taken at the Windward Coast and 2,224 (or 22.5 percent) at the Gold Coast. The remaining 938 slaves (or 9.5 percent) were taken at various places north of the Windward Coast or east and south of the Gold Coast. This pattern of trade differentiated that of the MCC ships from that of British, British colonial, and French traders. They sometimes purchased whole complements of slaves at the Windward Coast. In other cases they collected a few slaves, “troque au vol” (or in chance or casual barter), as they sailed along the Sierra Leone and Windward Coast to places further south.¹²

Rarely, however, do they seem to have acquired two-thirds of their slaves troque au vol and the remainder at markets further south. The unusual sequential regional pattern of slave purchases by MCC ships probably helps to explain their slow turnaround times in West Africa in 1751–1797 noted earlier. It also needs to be taken into account when exploring not only the levels and patterns of shipboard slave mortality on MCC ships but also how far they were representative of the trade more generally.

At the Windward Coast, MCC ships spent about 153 days (95 percent CI 132–174 days) acquiring slaves whereas at the Gold Coast purchases took about 49 days (95 percent CI 38–59 days). Slaves were acquired at the Windward Coast at various places, with no single location predominating. Some spreading of purchases across trading venues was also evident at the Gold Coast, but 44 percent of slaves taken there were bought at Elmina and a similar percentage at Anomabu, Apam, and Axim combined. Widening the perspective to the regional level again, the implication of the regionally sequential patterns of purchase was that two-thirds of the slaves in our sample—mostly those purchased at the Windward Coast—spent up to six months longer (or 48–200 days) on board ship than the rest.

As Elmina was a major trading venue for Dutch slave ships from the late seventeenth century onwards, it is unsurprising that MCC ships purchased over 40 percent of their Gold Coast slaves there.¹³

More surprising is the fact that up to two-thirds of the slaves acquired by MCC ships trading to Guinea came from the Windward Coast. Dutch traders certainly visited this region from the late seventeenth century, but the second West India Company, which dominated Dutch slaving before its demise in 1730, seems largely to have neglected it as a source of slaves.¹⁴ The trading pattern of the MCC probably represented, therefore, an innovation in Dutch slaving activity, which, when combined with simultaneous moves by British traders and British colonial traders, contributed to a remarkable upsurge in slave exports from the Windward Coast in the third quarter of the eighteenth century.¹⁵ This, in turn, seems to have been a response to more general pressures on slave supply in other regions of West Africa in the face of growing international slave exports after 1750.¹⁶ In the case of MCC ships, the concentration of slave purchases at the Windward Coast may also have reflected a decline of Dutch competitiveness relative to the British and French in the larger Gold Coast, Slave Coast, and Bight of Biafra slave markets after the demise of the second West India Company. In the MCC period, the French dominated the Slave Coast and the British the Bight of Biafra and to a lesser extent the Gold Coast.

During their stay in Africa, MCC ships trading at Guinea (defined as rows 1–6 of Table 1) acquired some 1.19 slaves per day on average. This was lower than loading rates achieved by British ships trading to Sierra Leone and the Gold Coast in 1791–1797, but broadly similar to daily loading rates of Liverpool ships trading at the Windward Coast and Gold Coast in 1750–1775 computed by other means.¹⁷

Unlike earlier estimations, however, rates derived from MCC accounts allow us to quantify and visualize short-term fluctuations in the rate and sex and age distribution of slave purchases across the whole loading phase, thereby exposing previously unknown features of the process. Purchasing strategies typically involved buying a few slaves every few days, with ships actually taking on board small clusters of slaves less than twice per week during their stay in Africa. During most of the purchase period, this pattern of acquisitions was fairly homogeneous or constant, as is shown in Figure 1, with slaves being acquired at the rate of up to five a week up to 20 weeks before departure from Africa, and rising to about eight a week (95 percent CI 7.3–8.8 slaves) between 20 and six weeks before departure. But during the last four weeks of loading acquisition patterns changed for all ships, but particularly for those finishing off their purchase at the Gold Coast. The records for the 32 ships in the last category show that most placed an order for a large number of slaves about a month before departure. An easing of loading rates to about three slaves a week then ensued, to be followed by a surge of acquisitions to close to 40 slaves a week just before departure. Indeed, of the 2,023 slaves acquired at Elmina, Anomabu, Apam, and Axim on the Gold Coast, some 1,350 (or two-thirds) embarked during the last week of purchase. This pattern of loadings is less evident for the remaining five ships in our sample.¹⁸ Taking all ships in our sample, it appears that, while the loading phase typically lasted eight months, over 78 percent of the slaves were on board a month before the ship left the coast and that 21.8 percent (or 2,151) of the 9,880 slaves embarked during the last month and 14.7 percent (or 1,390) during the final week before sailing.

MCC accounts also provide day-to-day information on the age and sex of the slaves purchased. We do not know the age at which MCC shippers distinguished children from adults, but it was likely to have been around 14 years. Of the 9,880 slaves purchased in our sample, some 81 percent were described as adult. Sex ratios were 55 percent male, 45 percent female. These proportions varied little by age or between voyages. Such age and sex distributions of slave purchases were broadly similar to those achieved by other carriers of slaves from the same regions and at the same time.¹⁹

In this respect at least, MCC voyages may be seen as representative of all voyages to the Windward Coast and Gold Coast after 1750. Further investigation of MCC accounts shows, however, that age and sex ratios of slave acquisitions commonly varied during the course of individual ship purchases, as illustrated for one typical voyage in Figure 2. From the beginning, masters bought mixtures of slaves and overall bought more adults, including men, on the Windward Coast than other parts of West Africa. Typically, however, they also tended to buy proportionately more girls and boys early in their purchase than throughout the whole purchase. Thus, half the final purchase of girls and boys was normally completed in less than 45 percent of the total loading phase of ships. This contrasts with the purchase pattern for women and men, where half the purchase of slaves in these two categories was achieved only after 60 percent and 69 percent, respectively, of the loading phase. The implication of this is that the last minute surge in purchases, described earlier, centered disproportionately on adults, especially men. This is underlined by the fact that whereas the typical voyage embarked 6 percent and 9 percent of the total purchase of girls and boys during the final week on the coast, the same figures were 17 percent and 12 percent for men and women respectively. In short, the loading of adult slaves and especially men by MCC ships proceeded at a more uneven rate than that of children. The combination of sequential sex-specific purchasing preferences and selective last-minute purchases had other important implications. It meant that, as most ships started trading at the Windward Coast, this region contributed proportionately more children to the final slave purchase than regions visited later. The reverse was true of the Gold Coast, which contributed proportionately more adults, especially men. Overall, seven times more children were taken from the Windward Coast than from the Gold Coast on our 39 MCC voyages (or 1,441 compared to 216). For adults this ratio was just 2.5 (or 5,777 compared to 2,008). These ratios contrast with an overall ratio of slaves embarked in the two regions of three to one. Similar calculations show that the Gold Coast contributed a disproportionate share of the adult slaves taken by MCC ships from the two regions.

These age and sex patterns of slave purchase meant that, ceteris paribus, children would have been expected to spend more time on board ship than adults before MCC ships left the coast. Indeed, the MCC accounts show that on the day before they left Africa girls and boys had typically spent 132 days (95 percent CI 115–150 days) and 124 days (109–139 days), respectively, on board ship whereas women had spent 101 days (86–116) and men just 91 days (70–113). In other words, children were typically exposed to on-board conditions for 41 days longer on average than adults before the crossing started and often much longer. Widening the perspective to include the crossing and selling phases of voyages, it appears that, while children comprised only 19 percent of the slaves purchased by our MCC ships, they accounted for 23 percent of the total exposure time lived by the 9,880 deportees on board ship. In other words, the slave-purchasing strategies of MCC shippers put children at risk for longer periods of time than adults, particularly men.²⁰

This has important implications for interpreting the age- and sex-related shipboard slave mortality rates described later.

Explaining age- and sex-related slave purchasing strategies by MCC shippers is more difficult than documenting them. We have found no evidence in instructions to shipmasters by the MCC bearing on such matters, but evidence from other accounts reveals that masters trading to the Windward Coast were sometimes instructed to be selective in terms of the quality of, and price paid for, slaves early in their transactions.²¹

The precise strategy adopted by MCC shipmasters was likely to have been a response to several factors. The most obvious was that adult slaves, particularly males, were more expensive to buy than children. Premiums on prices of adult males varied widely through time and place, but rates for adult males of 20 percent above the mean price for slaves were not uncommon, and the mark-up on adult males was even higher relative to children.²² Price data found in MCC records confirm these impressions. As time spent on board ship was widely considered a health hazard for slaves, a deferral in purchasing high-value adult males was perhaps to be expected. In other words, it reflected a policy adjustment by MCC shippers to coastal price differentials among categories of slaves largely dictated, it seems, by planter preferences for adult males in the Americas.²³

It would be misleading, however, to attribute slave purchasing strategies by MCC shipmasters purely to coastal relative prices for slaves or to demand-side factors. Such strategies were conceivably shaped, too, by African-centered factors as well as perceptions of the shipboard survival capacity of different categories of slaves. The intensity of MCC activity at the Windward Coast after 1750 almost certainly reflected pressures on slave supply capacity as well as the strength of competition for slaves in other regions of Atlantic Africa at that time.²⁴

It brought MCC shippers, too, into contact with emergent slave supply systems that were perhaps less efficient than the well-established systems of the Gold Coast in satisfying shippers' preferences for slaves. This might explain the higher ratios of children and females taken by MCC ships from the Windward Coast relative to the Gold Coast. Differences in regional patterns of slave exports may also have been linked to a higher propensity of slaves shipped from the Windward Coast to rebel on board ship. This rebelliousness has been increasingly well documented by historians. It was attributed by some contemporaries not just to Windward Coast slaves but to adult males generally and the Gold Coast in particular.²⁵ The relative lateness in finalizing the loading of adult males may therefore have been just as much an adjustment to fears of slave rebellion as to the price premium on adult male slaves. Whatever weight one attaches to particular factors, however, the slave purchasing strategies followed by MCC shippers were likely the rational outcome of a complex set of demand- and supply-side factors. We turn now to explore their relationship to shipboard mortality of slaves.

SHIPBOARD MORTALITY

It is widely recognized that there was a large spread and skewed distribution of slave mortality on board ship from voyage to voyage and that this presents problems for establishing a meaningful measure of mortality levels.²⁶

The slave mortality experience on MCC ships was no exception.²⁷ The pooling of data relating to 9,980 slaves, however, enables us to discern some fairly clear and robust macroscopic trends in shipboard mortality on 39 MCC voyages to Guinea in 1751–1797.²⁸ Our focus on such trends limits our capacity to interpret the short-term fluctuations in slave mortality that other historians have heroically attempted to describe but it does reveal previously unnoticed differences in slave mortality experience as well as some persistent time trends in mortality over voyages.²⁹ It also sheds new light on the impact of shipboard and African conditions on such mortality.

We begin by providing in Table 2 some basic mean and median values of voyage duration as well as slave purchases, exposure, loss in transit, and death rates for the 39 voyages in our sample. To give a very general impression: the typical MCC voyage to Guinea in 1751–1797 purchased 253 slaves (range 71–390), lost 33 slaves (range 2–174) during the course of its voyage, and sold 220 slaves (range 45–349) in the Americas. The loss in transit, defined in this case to include the loading, crossing, and selling phases, averaged 13 percent (range 1–65 percent). Though problems exist in making comparisons, this seems not to have been untypical at this time.³⁰

When account is taken of actual time spent by slaves on board ship, the data reveal a mean slave death rate for the 39 voyages of 259 deaths per 1,000 person-years (range 29–922 d/py). Given the age structure of the populations involved, this was an extraordinary rate by any standard.³¹ It is witness to the enormous human costs of the Atlantic slave trade.³²

Our methodology allows us to go beyond these crude measures of mortality, and to investigate the impact on mortality of risk factors such as length of time on board ship, geographical location of trade, and the age and sex of slaves. Table 2 provides information relating to the first of these. It shows that the loading phase accounted for almost 66 (or 54.5 per cent) of the 121 person-years spent on average by slaves on board ship. It also accounted for 41 percent of slave deaths. This amounts to a much higher proportion of total slave deaths on board ship at the coast than has sometimes been estimated.³³

Nevertheless, though ships spent more days loading slaves than sailing to the Americas, death rates were lowest during loading (199 d/py). Thereafter, they averaged 379 d/py during the crossing, and 342 d/py during the selling phase for all 39 voyages in our sample.³⁴ Four of the 39 voyages, however, experienced slave losses in transit of over 30 percent. Discounting these produces the result in Figure 3.³⁵ This reveals that, even in the case of the 35 voyages where losses in transit were under 30 percent, slave mortality rates doubled in more or less linear fashion from 146 d/py four months before ships sailed from Africa to 290 d/py six weeks after sailing from the continent. These findings confirm the suggestion of Richard Steckel and Richard Jensen, based on British data, that slave mortality rates were normally lower during the stay of ships in Africa than during the Atlantic crossing.³⁶ But they also indicate that on MCC ships mortality rates rose systematically the longer slaves were on board ship, whether ships were at the coast or at sea. Only landfall and sale of slaves in the Americas seems to have tempered such trends, but as Table 2 reminds us, even in the selling phase mean mortality rates remained higher than at the African coast. The imperative for MCC traders to sell slaves as quickly as possible after reaching the Americas was evident.

Our data allow us to investigate risk factors for slaves other than time. Table 3 refers to the coasting period (or phase 2) and provides data on the coastal location of ships at the time of slave deaths. These show that while trading on the Windward Coast ships experienced slave mortality rates of 167 d/py. Rates rose to 189 d/py during trading on the Gold Coast. It is likely that this rise in mortality rates between the Windward Coast and Gold Coast was in part a result of time factors, with sick or exhausted slaves bought on the Windward Coast dying while ships completed their slave purchase further south. But one cannot rule out the possibility that it may also have been caused by unhealthy Gold Coast slaves dying soon after purchase.

The data offer a rather clearer picture when it comes to looking at age and sex as risk factors for slaves. Table 4 refers to the entire period when slaves were on board (phases 2–4) and offers disaggregated data on slave mortality by age and sex. Some caution in interpreting the data on mortality by age is needed when relying on only two categories—adults and children—and when the boundaries between the two are blurred and the precise age distributions within them are unknown. The data suggest, however, that mortality rates for adults were systematically higher than for children. Moreover, among adults, they were much higher for males than for females. Thus, at 296 d/py, the mortality rate for adults was more than double that of children, while at 402 d/py, the mortality rate of men was some 88 percent higher than the 214 d/py for women. Such findings inform Figure 4, which reveals survival rates for groups of slaves by age and sex during the crossing. It shows a steeper decrease in surviving adult males compared to other groups. On MCC ships at least, it appears that being an adult male slave was a risk in itself.³⁷

Multivariate analysis allows us to assess the relative contribution of these and other risk factors to shipboard slave mortality. Our analysis uses all daily records for all 39 voyages during the loading, crossing, and selling phases in a nonlinear multivariate model, with separate regressions for loading and crossing phases. The results are expressed as relative risks and refer to odds of dying on any given day.³⁸

The constant for each regression is the adult male on the first day, located on the Windward Coast, for the loading phase, and departing from the Windward Coast for the crossing phase. Table 5 presents the results for the loading phase, Table 6 for the crossing.

During the loading phase (Table 5), the odds of dying for an adult male on the first day of loading was equivalent to a death rate of 54.76 d/py. For adult women, the odds were 0.59 of those for men (32.30 d/py); for girls 0.54 (29.57 d/py); and for boys 0.40 (21.90 d/py). These baseline odds worsened with every month the ship took to load (× 1.21), during the rainy season from June to August (× 1.47) or during the following trimester to November (× 1.40).³⁹

They were maybe better on the Gold Coast (× 0.77) and certainly better in the Bight of Biafra (× 0.17) but worse in West-Central Africa (× 2.64) and much worse still in the Bight of Benin (× 5.61).⁴⁰ No other significant risk factors were found during loading.⁴¹ During the Atlantic crossing (Table 6), the odds of dying started at 124.54 d/py for men. They were lower for women (× 0.60), for boys (× 0.37) and for girls (× 0.36). They increased every month during the crossing (× 1.47) and during the rainy season (× 1.64). Strangely, mortality seems to have increased with increased experience of the ship captain (× 1.16). It is important to note, however, that not all variations in slave mortality can be explained by the factors referred to above (random factors of × 1.76 and × 1.58 for loading and crossing). Nevertheless, these results suggest that, together with time-related variables and seasonality, age and sex were critical factors in shaping the survival chances of enslaved Africans deported on MCC ships from West Africa to the Americas in 1751–1797. On the basis of this MCC evidence, therefore, Africans plainly did not enter the trade with an equal probability of surviving.

IMPLICATIONS

The principal finding of this longitudinal study is that slave exposure and mortality patterns on board MCC ships trading to Guinea after 1750 were not homogeneous. Depending on the region of embarkation, large proportions of slaves spent much longer on board ship than others. Adult males tended to be purchased later than children. There was a slow but continuous rise in mortality throughout the loading and crossing phases of voyages, with a dramatic rise in mortality at the end of very long crossings. Adult males were at higher risk than other slaves, especially children. The overall slave mortality rates found on MCC ships in 1751–1797 were, at 199.0 d/py and 379.0 d/py, respectively, for the loading and crossing phases, much higher than those found by Steckel and Jensen for British ships in 1792–1796 in the only other study using standardized shipboard slave mortality rates known to us.⁴²

This is true whether we use mean or median values for mortality on MCC ships to compare with the British data. We cannot explain this discrepancy, but it is worth noting that Steckel and Jensen acknowledge that their estimates of mortality rates were low. Our study, nevertheless, endorses their finding that mortality rates during the crossing were much higher than during the loading phase of the voyage.

In some respects MCC voyages to West Africa were atypical of the slave trade in the second half of the eighteenth century. They had an unusual balance of trade between the Windward Coast and the Gold Coast. They typically took over 60 percent longer to purchase slaves at the coast than their contemporary rivals, yet experienced mortality rates among slaves in the crossing from West Africa to the Americas that were relatively low. The juxtaposition of the last two may in part have reflected the concentration of MCC West African trade at the Windward Coast and Gold Coast. Ships leaving these two regions typically experienced lower mortality in the crossing than those leaving Senegambia, the Bight of Benin, and Bight of Biafra and which are included in more general calculations of mortality. The apparently exceptional nature of some features of MCC voyages to West Africa suggests the need for similar micro-studies of slave mortality for other regions of trade in Atlantic Africa.

Even with these caveats, however, our investigation of shipboard experience of slaves on MCC vessels trading to West Africa encourages reflection on conceptualizations of the middle passage; on the impact of age and gender on slave purchase and mortality patterns; and on the relative importance of African and shipboard conditions in determining mortality trends. These are inter-related issues. By way of conclusion we briefly examine each of them.

The image of the Atlantic slave trade as a triangular trade linking three continents—Europe, Africa, and the Americas—has been a powerful factor in shaping the historiography on the middle passage. Conventionally defined as the ocean voyage between Africa and the Americas—and thus the second of the three legs of the triangular trade—the middle passage has become a metaphor for the horrors of transatlantic slavery. Viewed as an ocean crossing, it has also become the principal focus of efforts to measure changes through time and across space of shipboard slave mortality, with the proportions of slaves leaving Africa lost in transit being the key measure of such mortality. Reconstruction of voyages by MCC ships reveals, however, that the Atlantic crossing occupied little more than one-fifth of the time that ships had “human cargo” on board. This may have been a lower proportion than in most instances, simply because MCC ships typically spent longer acquiring slaves at the coast than other ships. Nevertheless, it remains the case that the Atlantic crossing typically took only a fraction of the time that most ships spent trading in Africa. This was especially true of the period 1751–1775 in which the MCC voyages included in this study were concentrated.⁴³

To better understand life and death on board slave ships, therefore, we need to adopt a broader definition of the middle passage. This should embrace all the time when slaves were on board ship and needs to take account of differences in time spent on board ship by individual slaves or groups of slaves as a result of shippers' slave purchasing strategies. This study is a first step in redefining the middle passage in this more comprehensive fashion and in clarifying the differences in time spent on board by different groups of slaves.

Redefining the middle passage places the slave purchasing strategies of shippers at the heart of discussions of shipboard slave mortality. The reconstruction of MCC voyages reveals a bias towards purchasing children and females at the start of loading and towards men in the final weeks. In the case of MCC ships, this had regional implications as ships tended to begin trade at the Windward Coast and conclude it on the Gold Coast. In terms of mortality, an important finding is that, against a background of rising mortality throughout the loading and crossing phases for all slaves, death rates of children and women remained lower than those of men. This fact may have been appreciated by some MCC shipmasters, at least half of who made multiple slaving voyages. It may thus have reinforced other factors such as the higher price of adult males and fear of shipboard rebellion, to which we alluded earlier, in shaping the slave purchasing strategy of such masters in Africa.

The age- and sex-related differentials in mortality on MCC ships contrast sharply with the findings of other studies of mortality using loss in transit data, which, if anything, show higher losses of children than adults.⁴⁴

The MCC mortality findings need to be confirmed by other studies, but should they prove representative, they have potentially wider implications. Confirmation of our findings would underline the role of gender as well as economic factors in shaping shippers' preferences for adult male slaves. Such preferences typically resulted in substantial premiums on prices of adult males at the coast. It might also help to explain the well-documented increases in proportions of children entering the Atlantic slave between the late seventeenth and early nineteenth centuries, increases that to date have largely escaped convincing explanation.⁴⁵ Understanding why slaves on board MCC ships had differential rates of mortality by age and sex and the typicality of these experiences has, therefore, a significance that extends well beyond the voyages covered by this study.

Finally, patterns of mortality on MCC ships help to inform the on-going debate over the relative importance of shipboard and African conditions in influencing mortality rates of slaves in transit. Differential rates of loss of slaves by African region of embarkation—notably between the Bight of Biafra and other regions—and allegedly high incidences of mortality in the early stages of the Atlantic crossing have tended to be seen as indicative of African influences on slave mortality. The MCC records used here offer very little new direct evidence on either of these arguments. Only a tiny share of the slaves included in this study came from the Bight of Biafra. Extraordinary variations in mortality over short intervals also make it unwise to do more than track macroscopic time-related trends in voyage mortality. Nevertheless, differentials in mortality rates of slaves by age and sex raise questions about the effectiveness of developments in shipboard sanitation and hygiene in lowering shipboard mortality rates equally for all slaves, and point, instead, to the condition of slaves at the time of embarkation as critically important in determining their survival rates.⁴⁶

This was particularly the case with adult males, relatively high proportions of whom embarked late in the loading phase, yet succumbed at higher rates than other slaves in the crossing. Precisely why this happened is uncertain. It could have been associated with pressures on captains to take adult males to satisfy owner's and buyer's expectations, thereby encouraging them to be less rigorous in slave selection, especially in the final stages of transactions at the coast when risks to the slaves on board ship were rising. Equally, it may have been associated with differences in enslavement and slave marketing processes between the Windward Coast and Gold Coast that caused adult males to reach the coast in less healthy condition than others. Having said that, it is worth remembering that men were typically seen as the primary instigators of shipboard rebellions and suffered the highest casualties during such incidents. Rebellions were not unknown on MCC ships. Moreover, men were commonly incarcerated on board in rooms separate from women and children and were often held in irons, particularly while ships were close to Africa. To our knowledge, the implications of differences in slave management strategies on shipboard mortality rates have until now escaped attention, principally because age and sex differentials in slave mortality have largely gone unnoticed. But the housing of children with women, some of whom may have been related, conceivably helped to nurture material and spiritual support systems among slaves lodged in the “women's rooms” that were less easy to inculcate and sustain among isolated and shackled men. Though our evidence may point towards the primacy of African influences on age- and sex-based differentials in slave mortality, it would clearly be premature to disregard slave resistance, shipboard management practices, and even mutual support systems among the enslaved as factors affecting differences in their survival rates on MCC ships. Patterns of shipboard slave mortality on MCC voyages to Guinea in 1751–1797 open up, therefore, new avenues of enquiry into the complex inter-relationship between African conditions, trading and management strategies, and life and death on board slave ships.