A robust body of social science research has investigated the spatial mismatch hypothesis (SMH), considering the consequences of geographic disparities between black residential locations and potential opportunities for employment. Focusing on U.S. urban areas between the 1970s and the present, studies have produced equivocal evidence on the implications of spatial mismatch for black employment. In this paper, we argue that the mixed evidence may result from a misspecification in both the historical time period and mechanisms whereby spatial mismatch affects black employment opportunities. We show that national declines in black employment and labor force participation, particularly among black women, were especially pronounced in the Jim Crow era (1880s-mid 1960s), rather than the post-industrial era (1970s to present) in which the SMH has generally been tested. We then investigate the extent to which the SMH should be formulated as a commuting problem, involving the difficulties that blacks face in reaching non-residential sites of employment, or a problem of residential ecology, in which blacks who do not live near entrepreneurs or white neighbors are less likely to obtain jobs. Analysis of census micro-data between 1910 and 1970 suggests that residential segregation provides the most consistent account of black-white employment gaps, insofar as employment under Jim Crow suffered when black housing was separated from the homes of business owners and work opportunities in residential locales.