In general: When such propositions are analysed, while the words ‘thing’, ‘fact’ etc. will disappear, there will appear instead of them a new symbol,
of the same form as the one of which we are speaking
which is a thing, or fact, respectively
; & hence it will be at once obvious that we cannot get the one ˇkind of prop. from the other by substitution.
  In our language names are not things: we don't know what they are: all we know is that they are of a different type from relations etc. etc. … The type of a symbol of
a relation is partly fixed by type of symbol of thing, since a symbol of latter type must occur in it.