Anscombe attempts to make clear the famously confusing 5.541 et seq. by relating it to the also famously mysterious 3.1432. Here are the two props in question: 5.541 At first glance it seems as though there is another way in which a proposition can occur in another. Especially in certain propositional forms of psychology, like "A believes that p is the case" or "A thinks p" etc. Here it seems superficially as though the proposition p stands in a kind of relation to an object A. (And in the modern theory of knowledge (Russell, Moore, etc.) those propositions have been understood in just this way.) 5.542 It is, though, clear that "A believes that p" "A thinks p" "A says p" are of the form "'p' says p": And here it is not a question of a coordination of a fact and an object, but rather of the coordination of facts by way of the coordination of their objects. 3.1432 "The complex sign `aRb' says that a stands in the relation R to b." No, not that, but rather "That `a'stands in a certain relation to `b' says that aRb." According to Anscombe, since, in order for "p" to picture p, the prop and fact must have the same form/multiplicity, that is the key point W was trying to make here. "[W]hat was cler to him was that for anything to be capable of representating [sic]the fact that p, it must be as complex as the fact that p; but a thought that p or a belief or statement that p, must be potentially a representation of the fact that p (and of course actually a representation of it, if it s a fact that p)?.'A believes p' or `conceives p' or `says p' must mean `There occurs in A or is produced by A something which is (capable of being) a picture of p' She then relates the two props back to 3.1432 which she claims to be "really not particularly obscure" if we only note that writing `a' and `b' in different colors won't do the trick that relating the two `signs' to each other does. But it seems to me not only that Anscombe has forgotten the whole point of 5.541 and 5.542 here (which is to deny that one prop can occur inside another without there being a truth-functional relationship between the two), but to have gotten the explanations reversed. (I speculate here that maybe she got a couple of her conversations with W mixed up.) First, noting that for "p" to picture p the object and sign must share a multiplicity/form does not help explain the non-truth-functional relationship between "p" and "A believes that p." What I think would do that is pointing out that "'p' says that p" is not an empirical but some sort of conceptual truth. (And interestingly, she spent a good chunk of time and trouble discussing and defending Reach's critique of Carnap on this very matter. I note that it's also something that Martin and I have discussed in the past at Analytic Borders, where Martin has cleverly suggested that a term might be both mentioned and used at the same time.) Using a different name for p than "p" simply produces different results than using "p" does. [See Reach's "The Name Relation and the Logical Antinomies" Journal of Symbolic Logic (1938)]. I don't claim that one can get "A judged that p" from "'p' says that p" based on this, but I do think both that it's clear that utilizing a non-empirical prop was the point of 5.542-3 and that Anscombe ought to have realized this, based on her attack on Carnap. Secondly, it seems to me that the real point of 3.1432 IS to require the isomorphism that Anscombe instead discusses with respect to the other two above props. That is "a stands in the relation R to b" would seem to have precisely three names, while it is unclear and controversial how many the sign "aRb" has. But that a stands in a certain relation to b, leaves open Griffin's interpretation of nothing but a and b configured, and so does what it is said to analyze, viz., that aRb. And, in fact, even if Griffin is incorrect, the mirroring relation will be possible, since there is no particular interpretation of the "R" as a name or non-name in either the analysans or the analysandum of W's preferred analysis. However we take the form of the one will be consistent with the form of the other, since both are rather plastic. Walto