I once read that by the 1920s, every major mathematics department in the country allowed women to be PhD students except Princeton, which didn't grant PhDs to women until the 1970s. (Side note: female grad students at Harvard got Radcliffe degrees until the 1960s.)
One spring day in the early 1980s, in Princeton's math department Common Room, someone asked the Graduate Chair about the next entering class of graduate students. After he told us about one of the star students, we asked if Princeton had accepted any women. He told us of several women they thought about admitting but decided against.
A typical case was a student at one of the all-female Seven Sisters schools (he had trouble remembering which one). She had top grades and great letters of recommendation. Princeton rejected her because her college didn't have a history of sending students to Princeton for math grad school (not surprising since Princeton had been a men's school until recently). The Graduate Chair didn't know what these good letters and grades really meant, unlike those from all-male colleges where he knew the faculty and Princeton had an established network.
I learned from this that Princeton's history as an all-male university continued to skew its graduate admissions decisions for years after the university decided to admit women. In much the same way, the Ivy League's practice of recruiting heavily from historically all-male prep schools like Exeter and Andover continued to skew its undergraduate admissions for years after the universities went coed.