The problem posed by common names, in which the presence in a sample of different individuals with the same names can result in false linkages and thereby cause estimates of persistence rates to be biased upward, has of course been encountered by previous persistence studies. One solution, used by Stephan Thernstrom in his study of Boston, was to omit from his base-year samples any individual with a common name, with the latter defined as any name (first and last) that appeared two or more times in the Boston city directory for the base year from which the sample was drawn.

Lacking complete listings of the population of Barbados for the appropriate period, a different procedure was followed here. After obtaining the persistence estimates of Table C.1, which use all the names of purchasers that appear in the Royal African Company invoices, all the names of purchasers in the sample were compared to the listing of names obtained by the first federal census of South Carolina in 1790. Two groups of common names from the Barbados sample were then identified on the basis of comparison with the South Carolina listings. One group was the most common names, defined as those names (first and last) that appeared two or more times in the South Carolina census. Of the total of 2,886 names in the Barbados sample of purchasers, there were found to be 234 of these most common names.