From Introduction, pp. 163-5. 'My most fundamental thought is this: logical constants are not proxies for anything. The logic of the facts cannot have anything going proxy for it. (4.0342) Here he is contrasting logical constants with names, which 'go proxy' for their objects: 'The possibility of sentences,' he has just said, 'rests upon the principle of signs as going proxy for objects'--and what this principle in turn amounts to is the possibility of logical picturing through one fact's having the same logical form as another--for only in the context of the proposition will a sign go proxy for an object. Sentences thus cannot represent, and nothing in them can stand for, 'the logic of the facts'; they can only reproduce it. An attempt to say what it is that they so o repreoduce leads to stammering.... [I]f we try to explain the essence of a relational expression to ourselves, we reproduce the relational form in our explanation. For as we have seen, we must make the distinction between 'aRb' and 'bRa' and if we do this by e.g. saying that in one the relation goes from a to b, and in the other from b to a, we produce a sentence which employs the essential relational form; for it reproduces the distinction produced by exchanging the places of the terms. All the logical devices--the detailed twiddles and manipulations of our language--combine, W tells us at 5.511, into an infinitely fine network, forming 'the great mirror'--that is to say, the mirror of language, whose logical character makes it reflect the world and makes its individual sentences say that such-and-such is the case. The simplest and most characteristic mark of this is that we do not have to learn the meanings of all the sentences of our language; given the understanding of the words, we understand andconstruct sentences, and know what they mean without having it explained to us.... It was at one time natural to think that the field of logic was the field of what was a priori true, i.e. true independently of all existence. On this W says at 5.552: 'The "experience" that we need to understand logic is not that something is thus or thus, but that something is: but that is not an experience. Logic precedes any experience--that something is thus. It comes before the How, not before the What.' According to the Tractatusthe 'what' is conveyed by the simple names, which cannot be taken to pieces by definitions 93.261) and which name the 'substance of the world' (2.0211). Thus even when a simple name is replace by a definite description, the description is merely 'about' the object, it could not 'express' it (3.221). Walto