This paper examines the relative roles of human and physical capital for the creation of scienti
c knowledge. To address the endogeneity of inputs, I analyze a shock to human capital, the dismissal of scientists in Nazi Germany, and a shock to physical capital, WWII bombings of universities. In the short-run, a 10% decline in human capital reduced output
by .2 sd whereas a 10% decline in physical capital reduced output by .05sd. The human capital shock persisted in the long-run, while the physical capital shock did not. To explore
mechanisms for the persistence of the human capital shock I show that the dismissal of star scientistswas particularly detrimental as they are key for attracting other successful
researchers and for the training of PhD students.