Princeton/Stanford Working Papers in Classics, May (2007)
Abstract
Any reconstruction of the Roman slave supply depends on two variables: the total number of slaves, and the relative contribution of particular sources of slaves to overall supply. Due to the nature of the record, these issues are at best only dimly perceptible. A simple comparison with the history of US slavery highlights the severity of this predicament: decadal census counts not only record the number and distribution of slaves but also permit us to calculate rates of natural reproduction and even to assess the patterns of the domestic slave trade. This body of data gives us a good idea of the scale and development of the underlying slave system. In the study of the world history of slavery, by contrast, an evidentiary basis of this kind is the exception while uncertainty and guesswork are the norm. Roman slavery firmly belongs in the latter category: hardly any genuine statistics are available, and historians face two similarly unpalatable options. Thus, we may decide to eschew speculative quantification altogether and focus on what our sources readily provide – that is, qualitative impressions of the prevalence of slave-ownership and the provenance of slaves.
This humanistic approach allows us to draw a rich canvas of slaveholdings large and small and of a variety of sources of supply from capture in war all the way to voluntary self-enslavement. What it cannot do is to give us even a remotely reliable notion of the representative value of scattered references. Conversely, we may choose to advance broad probabilistic estimates of the demand for slaves and the likely weight of different sources of supply. This approach is likewise fraught with serious problems: it depends on inherently aprioristic notions of plausibility; if these notions are backed up by comparative evidence, they run the risk of circularity; yet in the absence of comparative contextualization, they invite arbitrary implausibility; it may be hard if not impossible to link broad models to qualitative source references; and models may at best produce a range of competing probabilities instead of a single authoritative reconstruction. What they can do, however, is to enhance our understanding of overall structure and scale in ways that would not otherwise be possible.
Walter Scheidel (University at Stanford)
Princeton/Stanford Working Papers in Classics, May (2007)
Abstract
Any reconstruction of the Roman slave supply depends on two variables: the total number of slaves, and the relative contribution of particular sources of slaves to overall supply. Due to the nature of the record, these issues are at best only dimly perceptible. A simple comparison with the history of US slavery highlights the severity of this predicament: decadal census counts not only record the number and distribution of slaves but also permit us to calculate rates of natural reproduction and even to assess the patterns of the domestic slave trade. This body of data gives us a good idea of the scale and development of the underlying slave system. In the study of the world history of slavery, by contrast, an evidentiary basis of this kind is the exception while uncertainty and guesswork are the norm. Roman slavery firmly belongs in the latter category: hardly any genuine statistics are available, and historians face two similarly unpalatable options. Thus, we may decide to eschew speculative quantification altogether and focus on what our sources readily provide – that is, qualitative impressions of the prevalence of slave-ownership and the provenance of slaves.
This humanistic approach allows us to draw a rich canvas of slaveholdings large and small and of a variety of sources of supply from capture in war all the way to voluntary self-enslavement. What it cannot do is to give us even a remotely reliable notion of the representative value of scattered references. Conversely, we may choose to advance broad probabilistic estimates of the demand for slaves and the likely weight of different sources of supply. This approach is likewise fraught with serious problems: it depends on inherently aprioristic notions of plausibility; if these notions are backed up by comparative evidence, they run the risk of circularity; yet in the absence of comparative contextualization, they invite arbitrary implausibility; it may be hard if not impossible to link broad models to qualitative source references; and models may at best produce a range of competing probabilities instead of a single authoritative reconstruction. What they can do, however, is to enhance our understanding of overall structure and scale in ways that would not otherwise be possible.
Click here to read this article from the Princeton/Stanford Working Papers in Classics
Sponsored Content