A critically discerning discussion of path integral bosonization is given. Successively evaluating the conventional path integral bosonization of QCD it is shown without any approximations that gluons must be composed of two quarks. This contradicts the fundamentals of QCD, where quarks and gluons are independent fields. Furthermore, bosonizing the Fierz reordered effective four quark interaction term yields gluons, too. Colorless ``mesons'' are shown to be Fierz equivalent to a submanifold of gluons. The results obtained are not specific to QCD, but apply to other models as well.