This is a purely editorial issue. The wording currently reads:

Thus, a call x(arg1,...) is interpreted as x.operator()(arg1, ...) for a class object x of type T if T​::​operator()(T1, T2, T3) exists and if the operator is selected as the best match function by the overload resolution mechanism ([over.match.best]).

What does it mean for T::operator()(T1, T2, T3) to exist? What are T1, T2, and T3? The idea here, I think, is to clarify that operator() is called if it exists, as opposed to some function pointer conversion that could happen and be a better match... I can't propose a better wording since I'm not entirely sure what the wording is currently intended to convey.