The weaker and the stronger Anscombe claims

 I think everyone working on the Anscombe essay (that includes one Victor Reppert) has been making a mistake in not distinguishing between Anscombe criticizing Lewis for, for example, not making the distinction between GC and CE on the one hand, and saying, and making this claim at the end of her piece:

I do not think that there is sufficiently good reason for maintaining the “naturalist” hypothesis about human behaviour and thought. But someone who does maintain it cannot be refuted as you try to refute him, by saying that it is inconsistent to maintain it and to believe that human reasoning is valid and that human reasoning sometimes produces human opinion.

That seems to imply a refutation, and a claim that once Anscombe's distinctions are drawn, the naturalist is off the hook. Lewis comes back and says "OK, I see your distinctions, but you're actually making it worse instead of better for the naturalist." At the end she goes from saying "You didn't refute the naturalist for these reasons," to "You can't refute the naturalist this way." It's more than just a correction. McGrath, who sees the whole thing as a win-win on both sides (just a friendly correction), overlooks this passage. But she doesn't repeat the more ambitious claim. 

Lewis on Supernaturalism

What Lewis means by "supernatural"
To call the act of knowing--the act, not of remembering that something was so in the past, but of 'seeing' that it must be so always and in any possible world--to call this act 'supernatural', is some violence to our ordinary linguistic usage. But of course we do not mean by this that it is spooky, or sensational, or even (in any religious sense) 'spiritual'. We mean only that it 'won't fit in'; that such an act, to be what it claims to be--and if it is not, all our thinking is discredited--cannot be merely the exhibition at a particular place and time of that total, and largely mindless, system of events called 'Nature'. It must break sufficiently free from that universal chain in order to be determined by what it knows. From Miracles, Chapter 3.