埃尔提斯教授对大西洋奴隶贸易总量计算差额的答疑

[笔者按]在利用戴维·埃尔提斯教授等人所编写的《跨大西洋奴隶贸易光盘数据库》时，笔者对计算大西洋奴隶贸易总量时出现的差额感到困惑不解，于是在2005年6月11日向埃尔提斯教授本人请教，当天就得到了埃尔提斯教授热情的回答。鉴于笔者认为这是利用该光盘时一个需面对的问题，下面就公布埃尔提斯教授的答疑，希望能对学界有关同仁有所助益。

笔者询问：
You say the numbers of trans-Atlantic slaves are 11.8 million departures and 10.3 million arrivals (The Trans-Atlantic Slave Trade: A Database on CD-Rom, Cambridge University Press, 1999, p.5), but I get the different numbers by searching the database, which are 6246598 departures and 5338548 arrivals. The method is that:
Full time period and Region of disembarkation broadly defined = United States: 317748(embarked); 270976(disembarked).
Full time period and Region of disembarkation broadly defined = Caribbean: 4084565; 3446600.
Full time period and Region of disembarkation broadly defined = Brazil: 1308479; 1165366.
Full time period and Region of disembarkation broadly defined = other Americas: 535806; 455606.
Addition: 6246598(embarked); 5338548(disembarked).
So there is a margin between your conclusion and my calculation. When I search "Full time period and Region of departure broadly defined = Africa", the result is even more confusing: 15839(embarked) and 13643(disembarked). Is the result tells us that the whole numbers of trans-Atlantic slaves are below 20000? It is impossible. Surely there must be a problem when I search the database, but I cannot understand where there is.
At the same time, I get other numbers of trans-Atlantic slaves between 1519 and 1867 by my calculating the table 2 and table 3 you provide on The William and Mary Quarterly, which are 11062000(departures) and 9599000(arrivals). (See David Eltis，“The Volume and Structure of the Transatlantic Slave Trade: A Reassessment”，The William and Mary Quarterly, Vol.58, No.1, January 2001, table II & III.)
Thus the numbers of trans-Atlantic slaves are very different. I think the main problem is my unfamiliarity with the database. Could you answer my question? I really need your help. Thank you very much.

埃尔提斯教授答疑：
As the Introduction to the CD explains, the database does not pretend to include every voyage that left Africa with slaves. The higher numbers to which you refer are the result of our projections which make allowance for the many voyages missing from the CD. These cannot be computed directly from the data in the database.
As you note, there are two of these projections. The first was calculated in 1998 and produced the estimates in the CD ROM. The second was the result of new calculations carried out the following year, after the Introduction went to press. The second projection appeared in the WMQ article, where the procedures for the projection are explained. The 1999 projection includes some new data which were not available when the CD-ROM appeared as well as slightly different assumptions about the nature of the missing material. The second set replaces the first.
Since then we have added many new data (7,000 new voyages) and I am developing a third set of estimates (as well as an expanded database). This new projection (along with explanatory spread-sheets) will of course replace the second.
The odd results you got for Africa "broadly defined" is the result of a minor glitch in definitions. "Broadly defined" in this context means only those regions where Africa alone is specified as a point of departure. In the vast majority of records a more specific region or port in Africa is indicated and such records are not included in the calculation. Thus broadly defined for the Americas is not the same as broadly defined for Africa. This will be corrected in the next edition (at least two years away).
I hope this helps.
原载《中国美国史研究会通讯》2006年第1期