In Wittgenstein's Logical Atomism, James Griffin says that one common view of what W meant by "analysis" in the Tractatus is mistaken. The confused commentators have said that as W admired Russell's elimination of definite descriptions by the use of bound variables, when he talked about analysis of propositions down to ultimate elements, he must have been thinking of the Russellian model for elimination of definite descriptions. But Griffin points out that no such expression as "(Ex) Fx and (y)if Fy then y=x" can be elementary because it contains logical terms, and, in any case, if there is a problem of ambiguity with respect to "the biggest guy in the room" it won't be eliminated through by Russellian analysis. In Griffin's view, W's propositional analysis is strictly analogous to chemical analysis, and an analysis of "the broom is brown" will start with such sentences as "the bristles are brown, the broomstick is brown, and the bristles are connected to the broomstick." It will in this way analyze "the broom" into smaller and smaller referents, just as a physical analysis would break down the broom itself. As Griffin understands the Tractatus, what's being claimed is that the multiplicity of language mirrors that of the world partly because the most elementary words (names) can designate only the most elementary objects (simples). And, like Leibnizian monads, no atomic object can be altered or destroyed, only moved around and/or combined with others. Similarly, on Griffin's view of W, no name can be of any complex, but only of a simple object. If that's true, it's unsurprising that W couldn't provide any examples of atomic propositions. With that intro in mind, here are a couple of interesting excerpts from Griffin's book: Every element in a proposition will be either a name or defined by names. But this means that descriptive words like `broom', `brush' and `stick' will be defined by names. But if names are of particulars, how can they define general words? `Broom', after all, can be used to describe many things, and how can I possibly give the meaning of this general word in terms which refer to particular objects? It would almost seem on the basis of this that names, other evidence to the contrary, cannot be restricted to particulars. Now, however, we should see a way out of this difficulty. I said earlier that analysis explains that what I mean by `the broom' is `the brush in a certain relation to the stick'. What it explains, in other words, is what I mean on this occasion; I mean this brush in a certain relation to this stick'. And analysis is definition in this sense; by moving from statements about complexes to statements about particulars, I eventually define what I now mean by the signs in the unanalysed sentence?. [S]ince particulars configured in such and such a way constitute a broom, names configured in such and such a way will say that these objects constitute a broom. The role of general words in propositions is, in other words, taken over in the elementary proposition by the configurations of its signs. What Griffin attempts to deduce from this is that no prop of the form Fa can be an elementary proposition. In a recent post, I reproduced this: 4.123 A property is internal if it is unthinkable that its object should not possess it. (This shade of blue and that one stand, eo ipso, in the internal relation of lighter to darker. It is unthinkable that these two objects should not stand in this relation.) and I mentioned some difficulties it seems to engender. Griffin handles them as follows: If a shade of blue can have an internal property, then it also has a structure; and if it has a structure, then it cannot be an object in the strict sense. It is called an object because it and a darker blue are spoken of as standing in a relation to one another, and speaking loosely we can call terms of a relation objects. So, at least when the "F" in "Fa" is a colour, "F" cannot refer to an object and "Fa" cannot be elementary?.The "a" in "a is blue" must therefore be complex. A blue object is an object whose elements have a certain structure. Now, this way of talking, along with W's earlier talk of physicists' points as examples of simples, makes his account of blue very close to that of physics: a blue object is blue because its surface is structured in a certain way, and it is blue rather than, say, red, because to be red it would have to be structured differently?. [B]oth colours and shapes, i.e. what we see, and sounds, i.e. what we hear, turn out to be analyzable?.These are?good grounds for entertaining seriously the idea that W thinks all `F''s in "Fa" are to be analysed away. All facts, it seems, are quite literally objects in some configuration?. In analyzing `the broom is in the corner' we pass through several stages in which we talk of the brush and the stick and then, presumably, of sub-descriptions of these. The final stage comes when, leaving descriptions altogether, we mention only particulars. Thus, names appear only in the final stage. This means that a name will appear in a proposition only when all the rest of the signs in it are names too. This in turn, would seem to mean that since the propositional sign "Fa" has the sign "F" in it, which is not a name, "a" cannot be a name?.Consider 3.221. `Objects can only be named'; in other words, I cannot describe them; I cannot say of an object that it is an F. 3.221 does not say just this, but I think we can surmise it. It does say that I can only state how a thing is and not what it is. That I can only say how a thing is means, I think, that I can only say how an object stands in realtion to other objects; I can only give its configuration with other objects. I think it's worth mentioning here that the method by which Griffin?and, he says, Anscombe too?attempts to make this interpretation of objects and simple props consistent with 4.24 (in which W explains his symbolism) doesn't seem entirely convincing. On the Griffin interpretation of 4.22, it seems to conflict with the simplest reading of 4.24, which certainly suggests that there are atomic props of the form `Fa'. Walto