Evidence drawn from records of auctions held by the Royal African Company in Barbados between 1673 and 1723 is used to obtain annual estimates of slave prices by demographic category. These price series are then used to investigate the implications of an economic analysis of the demographic composition of the slave trade. The results provide quantitative support for the prediction that rising slave prices in the West Indies caused an increase in the share of children among the population of slaves in the transatlantic trade. This economic effect may have been a significant link between American slave markets and the demographic history of black populations in both Africa and America.
1 For accounts of the introduction and spread of sugar cultivation in Barbados, see Dunn, Richard S., Sugar and Slaves: The Rise of the Planter Class in the English West Indies, 1624–1713 (Chapel Hill, 1972), Chap. 2;Sheridan, Richard B., Sugar and Slavery: An Economic History of theBritish West Indies, 1623–1775 (Barbados, 1974), Chap. 7.
2 McCusker, John J. Jr “The Rum Trade and the Balance of Payments of the Thirteen Continental Colonies, 1650–1775,” (unpublished Ph.D. dissertation, University of Pittsburgh, 1970), p. 699.
3 On the wealth of the planters, see Dunn, Sugar and Slaves, Chap. 3.
4 Galenson, David W., White Servitude in Colonial America: An Economic Analysis (Cambridge, 1981), Appendix H.
5 Davies, K. G., The Royal African Company (New York, 1970), p. 229;Davies, K. G., The North Atlantic World in the Seventeenth Century (Minneapolis, 1974), p. 118.On the Company's difficulties in enforcing its monopoly, see Davies, Royal African Company, Chap. 3, andThornton, A. P., “The Organization of the Slave Trade in the English West Indies, 1660–1685”, William and Mary Quarterly, Third Series, 12 (07 1955), pp. 399–409.
6 On total slave imports to Barbados in the period, see Curtin, Philip D., The Atlantic Slave Trade: A Census (Madison, 1969), p. 55.
7 Menard, Russell, “From Servants to Slaves: The Transformation of the Chesapeake Labor System” Southern Studies, 16 (winter 1977), pp. 355–90;Main, Gloria L., “Maryland and the Chesapeake Economy, 1670–1720”, in Land, Aubrey C., Carr, Lois Green, and Papenfuse, Edward C., eds., Law, Society, and Politics in Early Maryland (Baltimore, 1977), pp. 134.
8 The homeward bound invoice account books are held at the Public Record Office, London, Treasury 70/936–59. The present paper extends the analysis and empirical findings of my earlier paper based on these records; see Galenson, David, ‘The Slave Trade to the English West Indies, 1673–1724’, Economic History Review, Second Series, 32 (05 1979), pp. 241–49.
This investigation is based on the records of all auctions held in Barbados that are included in these volumes. The analysis of prices in this paper is based on the records of all slaves recorded in the surviving invoice accounts that were delivered alive and sold in Barbados. (The one exception to this statement involves a number of slaves sold during the years 1673–1686 who were sold for sugar rather than currency; lacking a farm price series for sugar for this period, these slaves were excluded.) The principal categories of slaves accounted for in the records that are not included in the analysis of prices are therefore those described as having died aboard ship or on shore prior to sale, and those described in the invoices as given away for any reason by the Company, principally to pay either freight charges or the commissions of ships' captains or agents. The analysis of the demographic composition of the trade in this paper is based insofar as possible on all slaves carried by the Company from Africa to Barbados, and therefore when the necessary information is available in the invoices includes slaves listed as having died and those listed without positive prices.
9 Curtin, Atlantic Slave Trade, p. 55.
10 Davies, Royal African Company, Chap. 7.
It might be noted that the figures given in Table I should not be taken to represent total deliveries of slaves by the Royal African Company to Barbados in this period; what they represent is the total numbers delivered and sold at auctions for which records were placed in the Company's account books (see fn. 8, above). More complete figures on the Company's total deliveries to Barbados and the other British West Indian colonies, obtained by supplementing the auction invoices with evidence drawn from a number of other sources, are given by K. G. Davies in ibid., Appendix Ill, pp. 361–63.
11 In general, descriptions of the condition or quality of the slaves sold do not appear in the records. An exception to this is occasional comments describing slaves in particularly poor condition. These typically appear in transactions that occur late in auctions. Thus for example in an auction held in Barbados in April 1674, in an auction of 70 total transactions, the heading “Refuse & sickly negroes” appears to describe the slaves sold in transactions 62 through 67, followed by the comment “A lame negro” for the man sold in transaction 68, and “Almost Dead” to describe the total of 9 slaves sold in transactions 69 and 70.
K. G. Davies concluded that virtually all sales of slaves by the Royal African Company involved credit. Interest charges often appear to have been included in the recorded prices of slaves in the invoices, with the time of payment agreed upon at the time of purchase. Information on interest rates charged and the length of credit is normally not recorded, perhaps because these were standard; as a result no attempt has been made here to separate the spot prices of slaves from interest charges. In the absence of significant changes in conditions of credit, the series presented here should closely follow trends in spot prices, and in view of the apparent prevalence of credit in slave sales in the West Indies, their levels should be comparable to those of slaves sold by other traders in the period. On credit in slave sales, see Davies, Royal African Company, pp. 316–25.
12 The estimating equation for the prices reported in Table 2 was:
where i = index of transaction number, P = amount paid per transaction, M = number of men, W = number of women, B = number of boys, G = number of girls. The regression line is constrained to pass through the origin.
Throughout the paper, price estimates have been treated as missing information in reporting results when estimated coefficients failed to be significant at the 0.10 level.
In all tables reporting prices, the figures reported for “n” represent the total number of individual slaves on which the reported price estimate is based.
In some years the distributions of slaves sold in the four categories were quite dissimilar over the course even of the first halves of auctions. In order to check the extent to which this might have affected the estimated prices, a second set of regressions was estimated on the same samples, in which account was taken of the order of the transactions. The specification allowed for separate linear trends in the price of slaves in each category. Prices were estimated for only three of the categories; the fourth, girls, had to be dropped because of insufficient observations.
The estimating equation was:
where t = transaction number within each auction, expressed as a percent of total transactions (that is, the estimates were obtained for 0 < t ≦ 50).
Transactions in which girls were recorded were excluded from the sample. For definitions of other variables, see above.
The resulting price series are available from the author.
The general similarity of the results obtained in this manner to those presented in Table 2 lends support to the validity of the estimated price series. The simple correlation coefficient between the two price series for men is.94, while the women's series yield a correlation of.63, and the boys' series a correlation of.89; all three correlations are significant at the.01 level.
13 The specification of the estimating equation was the same as that of the second equation given above in fn. 12, except for the omission of the two variables involving B. Transactions involving boys and girls were excluded from the sample. The sample was defined as 50 < t ≦ 100.
14 The correlation coefficient between the two price series for adult men of Tables 2 and 3 is.70, while that between the corresponding series for adult women is.54; both correlations are significant at the.01 level.
A separate set of regressions was estimated on the sample underlying the estimates of Table 3 in order to check their robustness in view of one other characteristic of the data. Frequently in the latter stages of an auction transactions appeared in which very large numbers of slaves were sold at extremely low prices. In order to asses the impact of these apparent distress” sales on the estimated prices, variables were added to the equation used to estimate the prices shown in Table 3 that controlled for the effect on the prices of both men and women of individual transactions involving a total of ten or more slaves.
The estimating equation was:
where d = 1 if (Mi + Wi) ≧ 10, and d = 0 otherwise.
For definitions of other variables, see fn. 12.
The resulting price series are available from the author.
There is considerable similarity between the movements of the price series obtained in this manner and those of Table 3.Thus the correlation coefficient between the two men's series is.89, while that for the women's series is.77, with both correlations significant at the.01 level.
15 One source of such information is a census of property holders in Barbados taken in December 1679, which includes information on land, servants, and slaves owned by each planter; Public Record Office, London, Colonial Office 1/44/142–279. Results obtained for four auctions held in Barbados during 1680 and 1681, for each of which 25 or more purchasers could be matched to the census, do suggest that wealthier planters tended to buy earlier in the auctions. Using land holdings as a proxy for wealth (excluding residents of Bridgetown), the following table shows the mean number of acres on the island of Barbados owned by slave purchasers, displayed by the quarter of each auction (defined by transaction rank) in which their purchases occurred, expressed as a proportion of the mean acres owned by the purchasers in the first quarter of the respective auctions:
Although the results shown here are erratic, in 11 of 12 cases the mean number of acres owned by purchasers in quarters 2–4 was below that of those buying in the first quarters of the respective auctions. This suggests that in fact the wealthiest planters may have tended to open the sales, paying the highest prices for the highest quality slaves. Although more information is necessary concerning the format of the auctions before firm conclusions can be drawn, if all planters interested in buying at a given auction had assembled at its beginning, it would not seem surprising that transactions might have been ordered by the purchasers' wealth, since wealthier planters, who placed a high value on their time, would demand, and receive, first service from the sellers. A number of hypotheses could explain a positive correlation between planter wealth and slave quality; the phenomenon might be analogous to that of larger (wealthier) firms paying higher wages to their employees.
16 For a summary description and full references to these, see Davies, Royal African Company, Appendix VII.
17 Heavner, Robert, “Indentured Servitude: The Philadelphia Market, 1771–1773”, this JOURNAL, 38 (09 1978), pp. 710–11.
18 It has been proved that under conditions in which bidders know the value of a good to them, the expected vaiue of the winning bid for the good in an auction is the same under both English and Dutch rules—that is, whether the sale price is approached from below or above; see Butters, Gerard R., “Equilibrium Price Distributions and the Economics of Information”, (unpublished Ph.D. dissertation, University of Chicago, 1975), pp. 76–88. An implication of this result is that the most efficient way to sell homogeneous goods under conditions of good information is to set a single price; that slaves were not sold in this way implies the existence of considerable heterogeneity in their quality.
19 A comment is in order concerning the difference between the current estimates and two previous uses of the Royal African Company auction records to produce price series for slaves. Davies's, K. G. Barbados price series, presented in Royal African Company, p. 364, represents the mean price of all slaves sold in the colony in each year, without regard to age and sex; his series was obtained by dividing total receipts in a given year by total slave sales. The series therefore does not permit the separation of price changes within a demographic category from changes in the shares of categories overtime. (I am grateful to Davies, K. G. for a description of his procedure, in a letter of October 18, 1976.)
My own previously published series for adult men and women in Barbados, given in Galenson, “The Slave Trade to the English West Indies, 1673–1724,” p. 242, were constructed using those transactions from the auctions in which either only a single category of slaves appears or in which, although more than one category appears, the price of each is marked. Those series were therefore based only on a relatively small subset of the auctions' information, unlike the current estimates.
A third slave price series that includes the period considered here is that of Bean, Richard Nelson, The British Trans-Atlantic Slave Trade, 1650–1775 (New York, 1975), pp. 132–36, reprinted with some changes in U.S. Bureau of Census,Historical Statistics of the United States, Colonial Times to 1970 (Washington, D.C., 1975), Part 2, Series Z-166, p. 1174. This series is based on a variety of sources, most of which did not provide information about the demographic composition of the slaves purchased, and as a result cannot distinguish between price changes within demographic categories and changes in the shares of categories in the trade.
20 Fogel, Robert William and Engerman, Stanley L., Time on the Cross: The Economics of American Negro Slavery (Boston, 1974), p. 75.
21 The ages of the adults and children are not stated in the invoices. While some variation in the definitions may have occurred over time, in 1720 the Company listed its usual definitions of the invoice categories, with men from 16 to 40, women from 15 to 35, boys 10 to 15, and girls 10 to 14; Donnan, Elizabeth, Documents Illustrative of the History of the Slave Trade to America (New York, 1969), Vol. II, p. 244; also see Davies, Royal African Company, p. 300. Determination of the slaves' ages must have been subject to considerable error, as it was based largely on such physical evidence as height and weight, and condition of teeth and hair. For some descriptions of the process,see Dow, George Francis, Slave Ships and Slaving (Salem, Massachusetts, 1927), p. 61;Grant, Douglas, The Fortunate Slave: An Illustration of African Slavery in the Early Eighteenth Century (London, 1968), pp. 54–55;Rodney, Walter, A History of the Upper Guinea Coast, 1545–1800 (New York, 1980), pp. 190–91.
Fogel and Engerman's prices for the Old South, 1790–1860, show the ratio of the average price of a boy of 14 to that of a man of 27 to have been 0.71, and that of a girl of 14 to a man of 27 to have been 0.65; Time on the Cross, p.76.
22 One striking feature of the weights might be noted, as they indicate that adult women tended disproportionately to be sold late in auctions. The reason for this is not known, although it might be recalled that a comparison of Tables 2 and 3 indicated that the price of women relative to men appears to have risen as the auctions progressed.
23 For example, see Sheridan, Sugar and Slavery, pp. 252–53;Carl, and Bridenbaugh, Roberta, No Peace Beyond the Line: The English in the Caribbean, 1624–1690 (New York, 1972), p. 252.
24 Davies, North Atlantic World, pp. 130–31, 188–89.On sugar prices, see Sheridan, Sugar and Slavery, pp. 392, 397, 496–97.
25 For discussion, see Galenson, White Servitude in Colonial America, pp. 151–54.
26 On trends in the demographic composition of the slave trade, see for example, Sheridan, Sugar and Slavery, pp. 242–43;Klein, Herbert S., The Middle Passage: Comparative Studies in the Atlantic Slave Trade (Princeton, 1978), pp. 242–47.
27 It might be noted that this formulation of total cost assumes that transport charges were paid only on slaves surviving the voyage to the West Indies. This is based on Davies's discussion of the method of payment (Royal African Company, p. 198), and the rates used in the empirical analysis here are those given by Davies, Cohn Palmer, however, has cited examples of contracts in which partial payment of a quoted passage charge was made for slaves who died during the voyage;Human Cargoes: The British Slave Trade to Spanish America, 1700–1739 (Urbana, 1981), p. 12. Since slaves carried for any part of the trip would impose some costs on the shipper, it would seem likely that in the case of contracts of the kind described by Davies the charge per slave delivered alive would have been set at a level sufficiently high to cover the total transport costs for all slaves carried in view of expected losses in passage; the expression for total shipping costs in the equation above would then yield the true mean value for actual total cost if a contract was of the type described by Davies as long as the appropriate values of t were used.
28 Thus the amount the Royal African Company paid for slave transportation from Africa to particular West Indian destinations did not vary with the age or sex of the slaves, but was a flat sum per slave levied on live deliveries;Davies, Royal African Company, p. 198.
29 Klein, The Middle Passage, p. 242. If mortality did vary significantly by age, it can be seen from equation (2) that this would have tended to lower the implied African relative price of the group with the higher mortality rates. Higher mortality rates for a particular demographic group would therefore have tended to lower that group's share in the transatlantic trade.
30 See the age-price profiles given by Fogel and Engerman, Time on the Cross, pp. 72–76, as well as those shown inHigman, B. W., Slave Population and Economy in Jamaica, 1807–1834 (Cambridge, 1976) pp. 190–205. More generally, although the peak age of the profile varies, the age wealth profiles of free workers also rise early in life and fall later, where wealth is defined as the discounted sum of future earnings; Becker, Gary S., Human Capital, Second ed. (New York, 1975), pp. 223–30.
31 Illustrative calculations of this type have been carried out, using the Barbados slave prices by sex presented above, supplemented by information on the relative prices of slaves by age estimated by Fogel and Engerman for the nineteenthcentury South, and evidence on transportation costs and passage mortality. Weighting the derived annual ranges in age by sex obtained in this manner by the estimated shares of the included age groups within the African population yields predicted annual shares of children in the slave trade which are strongly correlated with the actual shares observed in the Royal African Company invoices, shown in Table 5. Details of these calculations are available from the author.
32 Direct evidence on the demographic structure of African populations in this period is not available, but a census which provides evidence on the age structure of the population of Angola in 1777 is reported in Thornton, John, “The Slave Trade in Eighteenth Century Angola: Effects on Demographic Structures”, Canadian Journal of African Studies, 14 (1980), Table 1, p. 421.
33 This argument assumes that the islands of the British West Indies were price takers in the African slave market, and therefore that changes in freight rates to the islands would have tended to change slave prices there without affecting African prices, that is, that changes in specific to the region could have changed PB by the same amount, leaving PA unchanged. Further, it is assumed that changes in West Indies slave prices in this period that occurred when transport costs were stable were due to market forces that affected wider areas—whether changes in African supply conditions or changes in the general demand for labor in the Americas due to changes in the world market for sugar—which would have affected African slave prices.
34 Davies, Royal African Company, pp. 198–99.
35 See fn. 21, above.
36 The price variable of the regression is based on the annual mean of the four price series indicated in the text, converted to sterling. Missing annual values for women's prices were interpolated as 0.85 of the appropriate men's price. Estimation of the regression using the sterling price index of Table 4 yields results virtually identical to those obtained with this price variable.
37 The discussion in the text concentrates on the relationship between slave prices and the relative, rather than absolute, quantities of the different demographic categories in the trade because evidence on the total annual volume of slaves delivered to Barbados by all suppliers is not available. The Royal African Company deliveries are deficient as a measure of this for, although as noted above the Company's share of the total slave trade to Barbados clearly changed over time, the precise share of the total trade made up by the Company's deliveries is not known, and no other evidence is available that would allow accurate estimation of total annual deliveries to Barbados in this period. The predicted sign of the relationship between West Indian slave prices and the total volume of slaves carried in the trade would of course be positive for price increases due to rising American demand for labor, and negative for price increases due to falling African slave supply.
The analysis of age composition presented here properly applies to the slave trade as a whole; the application here to shipments of slaves to a single destination is done under the assumption that, lacking evidence for the whole trade, these can be taken as a representative sample with respect to the overall trade's age composition.
38 Palmer, Human Cargoes, pp. 52–54.
39 An important example of this may be afforded by females. Philip Curtin has found that within Africa, female slaves were valued more highly than males; see his Economic Change in Precolonial Africa: Senegambia in the Era of the Slave Trade (Madison, 1975), p. 176. As Curtin notes, this could account in part for the disproportionate numbers of males carried to America relative to their representation in African populations. Analysis of the evidence of the Royal African Company invoices shows that the relationship between the share of girls among females in the trade to Barbados and slave prices there is considerably weaker than the corresponding relation between the share of boys among males and slave prices. A possible explanation of this difference involving relative demands for slaves by age and sex within Africa that differed systematically from those that existed in America will be presented in a forthcoming paper on the demographic composition of the Royal African Company slave deliveries to all the West Indian colonies during 1673–1725.
Email your librarian or administrator to recommend adding this journal to your organisation's collection.
* Views captured on Cambridge Core between <date>. This data will be updated every 24 hours.
Usage data cannot currently be displayed