The conclusions presented in this work rest upon an analysis of extensive statistical data drawn from the federal manuscript censuses for 1850, 1860, 1870, and 1880. Although these censuses are indispensable tools for the economic and social historian, any scholar who has used them knows that they are potentially deceptive as well as insightful. To maximize their value, historians must use the quantitative evidence that they contain judiciously and with a constant awareness of their limitations – in the same manner, in other words, as they would approach any other form of historical artifact. This means applying rigorous standards both in the selection of evidence (sampling) and in its interpretation (statistical inference).

The data sets developed for this study fall into one of two basic categories. The statistics on wealth distribution presented in Chapters 2 and 3 are based on estimates of real and personal wealth recorded in the manuscript population censuses and include the entire population of white farm households for all eight sample counties. A farm household, as I have defined it, included any household in which one or more members reported a farm-related occupation to the census enumerator, whether planter, overseer, farmer, tenant, farm laborer, or farm hand.