Sorry this is a bit of a back-up from Chapter 4's recent posts, to section 8. where TW is talking about what exactly is a shared language, and how exactly they (or it) is possible.

1. If there is something that binds together the common practice of using a word with a given meaning, then there is a shared understanding of that word.
   
2. If there is a shared understanding of that word, then there is either a mutual similarity of the constituents or it is the complex interrelations of the constituents.
   
3. So, if there is something that binds together the common practice of using a word with a given meaning, then the something is either a mutual similarity of the constituents or the something is the complex interrelations of the constituents.
   
4. It is not the case that the something is a mutual similarity of the constituents.
   
5. Thus, if there is something that binds together the common practice of using a word with a given meaning, then the something is the complex interrelations of the constituents.
   
6. There is something that binds together the common practice of using a word with a given meaning.
   
7. Therefore, the something is the complex interrelations of the constituents.
   

TW proposes that the something which is the complex interrelations of the constituents, is above all the constituent's causal interrelations. And that “the idea that a shared understanding of a word requires a shared stock of platitudes depends on the assumption that uses of a word by different agents or at different times can be bound together into a common practice of using that word with a given meaning only by an invariant core of beliefs.” So, the something is an invariant core of beliefs (foreshadow moment, I think... not JBs, not JTBs, no esatblishing the entailment links between UA, UK, UJ or UT. Just plain old beliefs / knowledge. Like I said not sure).

The premises that stick out here are 1., 2., 4., and 6.

How about 1. and 6.? Though perhaps wishful and hopeful, the truth of premises 1. and 6. are hard to resist. If there isn't some sort of something that binds the my use and meaning of words (or terms) with those of my fellow man. Well frankly I'm sunk, and so is everyone else. To deny either of these two would invite skepticism about a binding theory of meaning and usage. Bad.

How about 4.? The conditions placed on having a mutual similarity of the constituents, is that they have “some invariant feature, shared by all the constituents and somehow prior to the complex as a whole.” To be fair TW qualifies these conditions by stating that this feature almost never occurs. So, even if a strict cataloging were to take place and this feature was found to occur amongst constituents some of the time, this would still not yield the widespread commonalities within usage and meaning that seem so apparent.

How about 2.? The antecedent seems to be fine for the reasons given above for 1. and 6. But the consequent is a little odd, if not false. A disjunction does not seem à propos in this instance (in all fairness perhaps it is my reconstruction of what is going on, but this does seem to be the way TW is setting things up). Rather, having a mutual similarity of constituents is a feature of the complex interrelations of those constituents.