I read one of the later editions which has a bunch of endnotes where he points out a bunch of things in the text which were either ambiguous, out of date, or in error. The errors I remember are that one of the prisoner's dilemma strategies which he said was stable turned out to actually be meta-stable, and that someone discovered a viable model for the sexual-selection handicap principle, which he had wagered would be shown to be unstable.
I'm onto reading "Blind Watchmaker" now. One minor thing that bugs me more than it should is that in it Dawkins makes two pedantic statements and one otherwise uncontroversial statement, which when taken together make a contradiction. The first bit of pedantry a statement to the effect that birds and mammals should properly be classed as reptiles because they are descended from reptiles and that more generally things are supposed to be classified by descent (basically, the entirety of chapter 10), the second bit of pedantry is that whales aren;t fish because they are mammals, and the third otherwise innocuous statement is the phrase "modern fish, other than sharks" in chapter 4. This last part demonstrates that he does not define "fish" to specifically mean the ray finned fish (and nor does he define them specifically as the sharks). Given this we have a definition of fish that includes the last common ancestor of both the bony and carilaginous fish, and therefore all the bony fish, including the lobe-fins, and therefore, by the monophyletic clades argument, also the reptiles, birds, and mammals.
EDIT:
Unless he means to define "fish" exactly as class Chondrichthyes