While human capital has been used with some success to analyze recent changes in racial income differences, scholars have repeatedly pointed to a major empirical problem that appears to severely limit the historical relevanlce and scope of skill-based theories as applied to racial questions. The challenge they raise is legitimate. Put simply, if measured skill disparities between the races narrowed throughout the twentieth century, why did income ratios first begin to converge in the 1960's? In this paper, I address this question relying on some unexploited census data by race on education, literacy, occupations, and income. Using these data that begin with the 1890 Census, I present new estimates of agespecific relative income positions of black men for all postslavery birth cohorts. In addition to reconciling the apparently inconsistent skill and income series, these income ratios offer a very different historical record than many economists believe to have been the case. To cite a prominent example, Gunnar Myrdal's classic work (1944) saw the economic position of his contemporary black America not only as dismal, but made even more so by its sense of hopelessness, given the absence of any hint of progress or change. While Myrdal's pessimism is understandable, it appears that even in his day seeds had long been sown that were already permanently altering and improving the relative economic status of black men.