Hi Walter: I've fallen a little bit behind, but let me offer a few comments here. I'm reading Griffin's book, by the way, and think it's pretty informative. ..if a bit quirky. I don't think it's right to say Russell "eliminated" definite descriptions (DDs); it was one of his most important and lasting contributions to philosophical logic (cf. Neale, "Descriptions, " Cambridge, MA: MIT Press, 1990). And W, I think, basically adheres to the analysis of DDs given by Russell, namely, that they are disguised existential and uniqueness statements. Agreed, of course, that no DD is atomic. The atomic sentences in predicate logic are just the strings of symbols of the form R(a, b, c, ...), where R is an n-ary relation, and each a,b,c, etc. is either a constant or a term containing no variables. So, if W. has this logic in mind for the way the world of all that is the case operates, then he has to at some point concede that there are atomic facts and corresponding atomic propositions. I don't see any way around this, unless the TLP is vacuous. At the same time, I think I can see how "The leaf is green" might be analyzed into atomic propositions. We need names for simples and we need either names for spatial locations or predicates that define spatial locations (otherwise, there's no language for the arrangment of objects). So "The leaf is green" becomes "Leaf is here" & "Green is here". In this way, every atomic fact is just a thing at a place. But then we need properties or effects, like green, warm, liquid, noisy, etc., to be things. Why can't there be names for complexes? Why can't I have a name for a house, "Tara", for example? Propositions about Tara resolve into more elementary propositions about Tara's component objects: "Tara is burning" = "The roof is burning" & "The porch is burning" & .... It would also appear that species and genera would acquire names and logical structures through disjunctions of some sort: water = this_water_1 | this_water_2 | .... Individual things, even if they are composite, like Tara, would be given by logical conjunctions Tara = this_stud & that_stud & this_rafter & this_brick & that_joist & .... It's practically impossible to carry the analysis of complexes as far as it would need to go, and there is always the possibility that empirical investigation leads to the discovery that what we had named a simple is in fact analyzable into components. Thus, there is no dire need to provide examples, and this, I would suggest, is why W. did not provide concrete examples. Although, as you point out, he did indicate their form: just what first-order predicate logic would demand. Thanks! --Ron --- On Tue, 8/3/10, walto <calhorn@rcn. com> wrote: