The Feynman anecdote is irrelevant. I've discussed this before, on HN even: it was a single unspecified test as a kid for middle school. For all anyone knows (and I have looked into all discussions of this, finding no sources beside the brief mention in Gleick), 125 IQ was the ceiling of that test - school tests, starting from the original Binet test, are often not intended to measure gifted children but in the normal or low range.