As Raymond Williams once said, the problem for analysis is not complex determination of x (identities or historical events, for example). Rather it's understanding the _order_ of determinations at a given moment. In precapitalist economies, the category of slave was determined differently than in 17th and 18th century colonial economies. Serfdom as an institution allowed many of the legal distinctions that supported slavery in the middle ages to fall away, while preserving its social forms.
But "race" as the byproduct of capitalist slavery is a pretty North American construct. In the transatlantic slave economies outside of North America (excepting perhaps Barbados), non-European racial otherness and slavery were not nearly as coextensive as in the North American new world colonies. (Olaudah Equiano's autobiography is instructive in this respect.) As Robert Brenner has argued, modern slave economies are in many ways non-capitalist, or only formally so, since they rely, like feudalism, on the threat of force to extract wealth, rather than on economic pressure or motives (ie. competition). In that sense, "race" as we understand it belongs to that twilight moment of development between feudal and capital social relations. It's not a product of capitalism, but of a residual social formation that supported it in the 17th-and 18th centuries.
Christian