Education is recognized to be a key factor of economic development, not only giving access to technological progress as emphasized by the Schumpeterian growth theory, but also entailing numerous social externalities such as the demographic transition (Murtin, 2009) or democratization (Murtin and Wacziarg, 2010). If the evolution of world distributions of income and longevity over the last two centuries have been described by Bourguignon and Morrisson (2002), changes in the world distribution of education have remained unexplored until now, despite their major importance. How has global education inequality evolved over the twentieth century? How should it be measured? Up to now, existing studies on education inequality have had limited spatial and time coverage. For example, Castello and Domenech (2002) and Thomas et al. (2001) provide a descriptive analysis of years of schooling inequality for a broad panel of countries, but their study starts only in 1960. Also, they remain at the country level and do not consider the world distribution of years of schooling, which takes into account educational differences both within and between countries. In contrast, this paper depicts the world distribution of education over 140 years, improving and extending the database recently released by Morrisson and Murtin (2009), which focuses on average years of schooling. The authors provide both average years of schooling and the distribution of education as summarised up by four quantiles in each country. Importantly, this new database is cross-validated by historical data on illiteracy rates. Then, they describe average stocks of primary, secondary and tertiary schooling by region since 1870, and estimate world inequality in years of schooling, which has been dramatically reduced since 1870. Focusing on the measurement of education inequality, this paper raises an important methodological issue. The authors show that a substantial share of inequality in years of schooling can be mechanically explained by a single component of the distribution of education, namely the population that has not attended school, subsequently called the illiterate population. Actually, they find that the observed decrease in inequality in years of schooling over the XXth century is almost entirely explained by the decline in illiteracy. They believe that this result, derived both theoretically and empirically, could help to reconsider an empirical fact discussed in the literature on education inequality (see Berthelemy (2006)), namely the cross-country negative correlation between the average of and the inequality in years of schooling. This correlation mainly reflects the negative and mechanical correlation between average schooling and the illiteracy rate. In line with a recent macroeconomic literature (see for instance Hall and Jones (1999)), the authors then turn to human capital as defined by Mincer (1974), in order to confer a monetary dimension to education. They propose estimates of the world inequality in human capital, examining several definitions for human capital. They focus on one functional form in particular, which accounts for the existence of diminishing returns to schooling. It is the only one that can account for the cross-country negative correlation between Mincer returns to schooling and average years of schooling, as described by Psacharopoulos and Patrinos (2004). At the national level, they find that that human capital inequality within countries has increased then stabilized or even decreased in most regions of the world. When plotted against average years of schooling, human capital inequality within countries has clearly followed an inverted U-shape curve, namely a "Kuznets curve of education". At the global level, they also find that human capital inequality has increased from 1870 to approximatively 1970, then has decreased. They interpret these findings as a consequence of mass education and the existence of diminishing returns to schooling. (Contains 6 tables, 6 figures and 14 footnotes.)