This same distinction between what can be shewn by the language but not said, explains the difficulty that is

felt about types – e.g. ˇas to difference between between things, facts, properties, relations. That

M
so & so

is a thing can't be said: it is nonsense: but something is shewn by the symbol M. In same way that a prop. is a subject-predicate prop. can't be said: but it is shewn by the symbol.
:. a theory of types is impossible. It tries to say something about the types, when you can only talk about the symbols. But what you say about the symbols is not that this symbol has that type, which would be nonsense for same reason: but you say simply This is symb the symbol, to prevent a misunderstanding. E.g. In „aARbB’, R is not a symbol, but that R is between one name & another symbolises. Here we have not said this symbol is not of this type but of that, but only: This is the symbol & symbolises & not that. This seems again to make same mistake, because ‘symbolises’ is ‘typically ambiguous’. The true analysis is: R is no proper name, &, that R stands between a & b ,(expresses a relation). Here are 2 props. of different type, connected by ‘and’
It is obvious that, e.g. with a subject-predicate prop., if it has any sen sense at all, you see the form, as soon as you understand the prop., in spite of not knowing whether it is true or false. Even if there were props. of form ‘M is a thing’

they would be superfluous (tautologous) bec. what this tries to say is something wh. is already seen when you see M.
  In the above expression ‘aRb’, we were talking only of this particular R, whereas what we want to do to is to talk of all similar symbols. We have to say: inc any symbol of this form R ˇwhat corresponds to R is not a proper name, & fact that … expresses a relation. This is what is sought to be expressed by ˇthe nonsensical assertion: [s|S]ymbols like this are of a certain type. This you can't say, bec. in order to say it you must first know what the symbol is: & in knowing this you see the types, & therefore also types of symbolised. I.e. in knowing what symbolises, you know all that is to be known; you can't say anything about the symbol.
  For instance: Consider the 2 props. (1) “What symbolises here is a thing”, (2) “What symbolises here is a ˇrelational fact (or relation = relation)”. These are nonsensical for 2 reasons: (a) bec. they mention ‘thing’ & ‘relation’ (b) bec. they mention them in props. of same form. But The 2 props. must be expressed in entirely different forms, if properly analysed; & neither the word ‘thing’ nor ‘relation’ must occur.
  Now we shall see how properly to analyse props. in wh. ‘thing’ ‘relation’, etc. occur.
(1) analysed = &(∃x). x symbolises.

N.B. here ‘thing’ doesn't occur

(2015–) Wittgenstein Source Bergen Nachlass Edition (WS-BNE). Edited by the Wittgenstein Archives at the University of Bergen under the direction of Alois Pichler. In: Wittgenstein Source, curated by Alois Pichler (2009–) and Joseph Wang-Kathrein (2020–). (N) Bergen: WAB.