When are economic phenomena persistent over time, and when are they not? If they are, do inequalities persist forever, or do they converge, and if so, at what speed? By analyzing the Indian state of Goa, this research makes use of a historical quasi-natural experiment to study the effect of Portuguese (catholic) colonialism. To achieve econometric identification, I apply a spatial regression discontinuity design alongside a border that was abandoned in the 18th century. This provides a rare opportunity to isolate and identify the effect of culture, holding constant geography, income, and institutions. I establish that historical disparities in female education can be overcome, albeit much slower than for males. In contrast, male-biased sex ratios stay virtually unchanged - highlighting the differential degree of persistence of deeply rooted preferences.