We build a comprehensive database of father-son pairs in academia, measure their publications,
 and develop a general method to disentangle two determinants of occupational persistence:
 nepotism vs. inherited human capital. This requires jointly addressing measurement error
 in human capital and selection from nepotism. Exploiting multi-generation correlations and
 parent-child distributional differences, we identify the structural parameters of a Markov process
 of intergenerational transmission with nepotism. The human capital elasticity is lower than
 in standard multi-generation estimates ignoring nepotism. Nepotism was lower in science vs. law,
 in Protestant institutions, and declined during the Scientific Revolution and the Enlightenment,
 testifying to the rise of meritocracy.