Earlier quoted context omitted.
Actually, that 0.7 intergenerational correlation only tracks surnames—i.e., the male line. It completely ignores the fact that ~50% of the population changes status by marriage, which is invisible in surname analysis. Think about it: when a blacksmith’s daughter marries a baron, her social mobility doesn’t show up anywhere in the data. She just becomes part of the baron’s lineage going forward. So Clark has discovere…
I think you're overestimating how far families married outwith their class. Given the scandal of Mrs Simpson or Ms Markle, how often do you think Barons married commoners? It's the stuff of fairy-tales.
Edit: or to put it differently, which of the two scenarios has been more likely in the past 500 years: the daughter of a blacksmith marrying a baron, or the son of a blacksmith becoming a baron through merit?