The Virginia evidence comes from Walter Minchinton, Celia King, and Peter Waite, eds., Virginia Slave-Trade Statistics, 1698-1775 (Richmond, 1984). David Richardson, ed., Bristol, Africa, and the Eighteenth-Century Slave Trade to America, 3 vols. (Bristol, 1986), supplied information on additional voyages and on African ports of embarkation for a number of the ships appearing in the naval office shipping returns, as did a preliminary version of David Eltis, Stephen D. Behrendt, David Richardson, and Herbert S. Klein, eds., The Trans-Atlantic Slave Trade: A Database on CD-ROM (Cambridge, 1999), kindly shared by the authors (hereafter cited as Trans-Atlantic Slave Trade Database. Page references to Trans-Atlantic Slave Trade Database are to the manual accompanying the CD-ROM).An earlier analysis of the Virginia material is Douglas Brent Chambers, "He Gwine Sing He Country": Africans, Afro-Virginians, and the Development of Slave Culture in Virginia, 1690-1810” (Ph.D. diss., University of Virginia, 1996), chaps. 4, 5.

The Maryland data are compiled from extant Maryland naval office shipping records; "List of Ships Trading to Maryland from 30 Apr[il] 1689 to 1693"; accounts of the collection of the provincial duty on imported slaves; Customs 16/1, Public Record Office; advertisements in the Maryland Gazette; Maryland Vice Admiralty Count Records, 1754-1777; Chancery Records; and Notary Public Record Book, 1744-1778, all available in manuscript or on microfilm at the Maryland State Archives, Annapolis; James Carroll Daybook, 1714-1721, Jesuit Provincial Archives, Georgetown University Library; ships trading to Maryland cited in Elizabeth Donnan,Documents Illustrative of the History of the Slave Trade to the Americas, 4 vols. (Washington, D. C., 1932-1935), esp. vols. 2, 3; and information on additional ships trading to Maryland found in Trans-Atlantic Slave Trade Database. For Maryland naval districts and records, John J. McCusker, "The Naval Officer Shipping Lists: An Introduction to the Records of Early British American Overseas Trade, 1657-1775" (manuscript in progress), was helpful. No naval office records survive for Maryland for the years 1702 to 1741, and returns are incomplete for several of the naval districts in subsequent years. Accounts of all direct shipments from Africa between 1699 and the end of 1708 are available in Donnan, Documents of the Slave Trade, 4:17-18. For an earlier analysis of the port of entry books see Darold D. Wax, "Black Immigrants: The Slave Trade in Colonial Maryland," Maryland Historical Magazine, 73 (1978), 30-45.

The Trans-Atlantic Slave Trade Database and the Chesapeake database were designed to answer somewhat different research questions, so although there is considerable overlap between the two, the Chesapeake database is not just a subset of the Trans-Atlantic Slave Trade Database. The Trans-Atlantic Slave Trade Database is designed to track and measure the transatlantic slave trade and hence only includes information on voyages from Africa to one or more American destinations. Ships carrying slaves from one colony to another but that were not involved in transatlantic carriage were deliberately excluded. The naval office records note all ships carrying slaves into Virginia and Maryland, as do records of the provincial duties collected on all imported slaves (in Maryland from 1695 and Virginia from 1699); the latter were used, where available, to compensate for missing naval office returns. Because the issue of the proportion of seasoned or locally born slaves brought from the West Indies looms large in Chesapeake historiography, the Chesapeake database is designed to answer the question: What was the overall composition of the slave trade to this region? Therefore it includes entries for both transatlantic and intercolonial voyages.

Several decisions that were made in the creation of the Chesapeake database should be noted. The Trans-Atlantic Slave Trade Database includes all ships whose intended destination was stated as Virginia or Maryland but whose arrival in one of these colonies either cannot be verified due to gaps in the naval office records or is not recorded for years for which the naval office returns are ostensibly complete. I excluded from the Chesapeake database some intended voyages included in the Trans-Atlantic Slave Trade Database that would have arrived in the Chesapeake in years for which the naval office records are complete or that were reported as captured or lost before arriving in the Chesapeake. I included in the Chesapeake database intended voyages in which slaves were successfully delivered to some North American destination that would have arrived in years when naval office records are missing or incomplete. The assumption was made that most of these ships did arrive at their intended destination, since Eltis et al. found that 82 percent of vessels for which both intended and actual ports of arrival in the Americas are known did go to the ports where they were bound at the outset of the voyages (Trans-Atlantic Slave Trade Database,18). For Maryland I also included direct shipments from Africa to Maryland between 1768 and 1772 listed in Customs 16/1. As these ships are not identified by name, captain, or port of origin, they are not included the Trans-Atlantic Slave Trade Database.I excluded from the Chesapeake database any ship listed in the naval office returns as both importing and exporting the same number of dutiable Negroes (usually one or two); it seems likely that these individuals were part of the ship's company rather than slaves imported for sale.

In tabulating the numbers of slaves arriving from Africa or the West Indies, I counted all slaves carried on ships included in the Trans-Atlantic Slave Trade Database as Africans, whether they came on vessels sailing directly from Africa to the Chesapeake or on ships coming from Africa that first entered at a West Indian port. I also classified as Africans those coming from the West Indies on other vessels not included in the Trans-Atlantic Slave Trade Database when those slaves were identified as "new" (as opposed to "seasoned") "Negroes" in Virginia naval office records between 1770 and 1775. All other slaves who arrived on ships from West Indian ports were classified as coming from the West Indies, although some of these were doubtless newly arrived Africans.

Returns from the Accomack naval district (a total of 6 ships) were tabulated with the Lower James (or from October 1736, Hampton) district. Four ships arriving in 1772 for which port of entry is given only as James River were assigned to the Upper James, as this seemed the most likely destination for large cargoes of captives. Those cases where ships disembarked slaves in two or more Chesapeake naval districts were counted as separate voyages, with the number of slaves disembarked in each district assigned to the two respective districts, rather than allocating the whole to the first port of entry. This produces a slight overcount of the total number of slave-carrying ships entering the two colonies but allows an accurate count, by region of origin, of the number of slaves brought into in each district. Since only four ships are known to have traded in two different African regions, the region of origin assigned is that given in Trans-Atlantic Slave Trade Database as the major buying region. (For alternative estimating procedures see ibid., 22-23.) In cases where information came from several sources, there are minor discrepancies in the reported numbers of slaves delivered. The numbers used here are those used in the Trans-Atlantic Slave Trade Database, which occasionally differ slightly from the numbers published in Minchinton et al.

Where direct information is lacking for the number of slaves disembarked at the end of a voyage, Trans-Atlantic Slave Trade Database offers estimated (imputed) numbers. This procedure compensates for missing information but does result in a mix of reported and estimated numbers with which some readers may be uncomfortable. Where the number of slaves embarked in Africa is known,Trans-Atlantic Slave Trade Databaseestimates the number of arrivals by discounting for mortality in the Middle Passage. The probability that estimations derived from numbers embarked are relatively reliable seems high (ibid. , 19-20). Where no information is available on the numbers either embarked or disembarked, the number of arrivals is imputed from the mean number of slaves carried on comparable vessels for which this information exists (ibid., 20-22). The likelihood that estimations of the second kind are less reliable than the first is clearly greater. Following the recommendation of Eltis, the imputed numbers used in the Chesapeake database for voyages with no information on how many slaves were carried differ from those used in the full Trans-Atlantic Slave Trade Database, since the mean number of slaves per ship brought to the Chesapeake was lower than for the transatlantic trade overall. These imputed numbers are 143 for ships coming from Upper Guinea, 200 for ships coming from the Gold Coast south to Angola, and 189 for ships from unspecified African regions. All tables in the text include both reported and imputed numbers. For those who prefer to rely only on reported numbers, these are given in Tables A1 and A2. The number of imputations made for Virginia are comparatively few; more are employed for the less complete Maryland returns. I did not attempt to estimate numbers of slaves carried between colonies on ships not involved in direct trade with Africa; the number of ships lacking this information is small, and the number of slaves they potentially carried was likely small. All tables were initially tabulated using both reported and reported plus imputed numbers and the results compared, whether the inclusion of imputed numbers produced different conclusions from those derived from reported numbers alone. Any differences were minor.

Two difficulties posed by missing evidence, which could affect the robustness of the patterns identified from voyages for which there is full information, should be noted. One is that the port of entry is not known for 60 of the 1,153 ships carrying slaves into Virginia. For four of the six time periods used in the tabulations, the number of slaves such ships carried is too small to have any significant effect on the overall findings. From 1704 to 1709, however, port of entry is unknown for 23 of a total of 32 arriving ships, with the former carrying an estimated 4,331 captives. Where might they have disembarked? From 1704 to 1718, 83 percent of the slaves for whom port of disembarkation is known entered the York naval district, and another 11 percent entered the Rappahannock. In addition, 20 of the 24 ships for which port of entry is unknown were registered in London, and the remaining 4 were from Bristol. Of the 29 London-owned ships for which port of entry is known, 21 went to the York; Bristol traders were active in both the York and the Rappahannock. Given both the dominance of the York district in the trade of this period and evidence that London traders favored that river, it is probable that most of those ships for which port of entry is unknown also entered there, with perhaps two or three going to the Rappahannock. The region of embarkation is known for 40 percent of the slaves carried on these ships; 70 percent were taken from the Bight of Biafra and 24 percent from Senegambia. Thus it is likely that some Senegambian captives disembarked on the York in these years (none appear among the ships for which port of debarkation is known) and a somewhat higher proportion from the Bight of Biafra were taken to the Rappahannock.There is no reason to suppose that the trading regions of the ships for which both region of embarkation and region of disembarkation are missing differed from those for which region of embarkation is known. Assigning all slaves who were taken to an unknown port of entry to the two most likely districts and their regions of embarkation assumed to be the same as those of the knowns would not substantially alter the percentages reported in the text for these districts.

For the years 1748 to 1759, eleven transatlantic voyages carrying about 2,100 slaves are documented in the Trans-Atlantic Slave Trade Database for which port of entry in Virginia is also unknown. Between 1746 and 1760, half of the ships for which port of entry is known entered the Upper James and another third stopped in the York. It is likely that most of the 11 ships whose port of entry is unknown went to one of these rivers. The region of embarkation is known for 80 percent of these ships; their origins were nearly evenly divided among the Bight of Biafra, West Central Africa, and Upper Guinea, proportions similar to those found on ships known to have entered both the York and the Upper James. Thus it appears unlikely that missing information on port of entry has much potential for weakening the patterns identified.

A more serious problem is that information is unavailable on African region of embarkation for a third of the slaves estimated to have been carried into Virginia and for more than half of those taken to Maryland. The problem is most serious for the early eighteenth century and for slaves transported on ships registered in London. The areas where Liverpool ships traded are documented for 84 percent of transatlantic voyages originating from that port, so the regional origins of Africans carried on Liverpool ships are firmly established. The African trading region is available for only 59 percent of Bristol ships coming to the Chesapeake, with the most problematic period being that of 1731 to 1745 for the York and Rappahannock districts and for Maryland. In a study of the Bristol slave trade overall, Richardson concluded that up to 1745 merchants from that port concentrated their trade in the Bight of Biafra and Angola. Since Bristol figured most prominently in the Chesapeake trade between 1719 and 1745, a possible effect of missing data for Bristol voyages is an undercounting of the proportions of slaves taken from these two more southerly African regions into the York, Rappahannock, and Maryland. The concentration of Africans from areas south of the Gold Coast in southern Virginia may thus well be understated, and the proportion from Upper Guinea in Maryland may be overstated, if one assumes that less well-documented Bristol shippers did not discriminate between lower and upper Chesapeake markets. For voyages originating in Liverpool or Bristol, one can at least suppose that the estimated numbers of slaves delivered approaches total imports, because the data are compiled from nearly full records of ships departing from and/or returning to these two ports.

London's trade in the Chesapeake is the least well documented of the three major British ports and the least studied overall, in part owing to the absence of surviving port records after 1696. Records of the London-based Royal African Company supply information on company-sponsored voyages, but much less evidence is available for independent shippers, some of whom may have traded in places in Africa other than those favored by the chartered company. And whereas Virginia naval office returns afford a nearly complete record of ships from London arriving in that colony, information is almost entirely lacking for an unknown number of London-based ships delivering slaves to Maryland between 1709 and the mid-1740s. Among London ships arriving in the Chesapeake, trading area in Africa is known for only 51 percent. The York from 1698 to 1718 and Maryland in all periods are the destinations most affected by missing data for London vessels. Among London voyages made before 1746 for which the African trading area is known, approximately half the captives were taken from Upper Guinea, with the remainder evenly split between the Bight of Biafra and West Central Africa. After 1745, two-thirds were from Senegambia and the remaining third from West Cental Africa. Most of these later slaves were taken to Maryland. It seems unlikely that the missing information would substantially alter the patterns observed for the York district. There, a likely undercount of slaves from Upper Guinea transported on London ships is offset by a likely undercount of slaves from regions south of the Gold Coast brought by Bristol traders. Moreover, many of the London-based ships arriving in the York were owned by independent traders, some of whom were trading at the turn of the century in the Bight of Biafra rather than in Upper Guinea. The regional origins of Africans brought to Maryland on London ships from 1698 to 1745 are the least certain.

The numbers reported in the William and Mary Quarterly article reflect additions and corrections made to the database since publication of preliminary estimates in Lorena S. Walsh, "New Findings about the Virginia slave Trade," Colonial Williamsburg Interpreter, 20:2 (1999), 11-21.