Difficulties in a kind of averaging procedure for constructing gauge-invariant operators out of gauge-variant ones

We prove that a kind of averaging procedure for constructing gauge-invariant operators (or functionals) out of gauge-variant ones is erroneous and inapplicable for a large class of operators (or functionals).