„But this position cannot be maintained either: against it speaks, first, "the possibility of crosswise definition of the logical 'primitive signs' of Frege and Russell" (TLP 5.42), for "it is self-evident that the '⊃' [→] defined by means of '∼' and '∨' is identical with the one that figures with '∼' in the definition of '∨'; and that the second '∨' is identical with the first one; and so on." (ibid.)
Secondly, logical primitive signs would have to mean logical objects, which, however, leads to an absurdity: "...if there were an object called '¬', it would follow that '¬¬p' said something different from what 'p' said, just because the one proposition would then be about ¬ and the other would not." (TLP 5.44) In reality, however, both propositions say the same; the logical constant *disappears*."