The author offers ``a defence of free logic, specifically a free logic admitting both empty names and empty definite descriptions as genuine singular terms''. He presents an interpreted first-order language containing both its own denotation predicate and the definite description operator, and argues that (1) such languages provide a positive support for a free logic, (2) a free logic admitting definite descriptions as genuine terms needs something like the constructions involved. Guided by \textit{S. Kripke's} ``Outline of a theory of truth'' [J. Philos. 72, 690-716 (1975)], he describes a least fixed-point construction for a class of partial free structures appropriate to the language mentioned above; the resulting structure possesses several desired properties. The construction leans upon the technique derived by \textit{S. Feferman} [J. Symb. Logic 49, 75-111 (1984; Zbl 0574.03043)] and is, for the most part, routine. The author distinguishes carefully between the terms having the empty value in a structure and those undefined in the sense that (the facts described by) the structure do not determine whether they are empty or non-empty. The presentation is preceded by a short general discussion on arguments for free logic against classical logic. The approach taken up in the paper has already been sketched by the author [Theoria 50, 178-211 (1984; Zbl 0581.03002)].