The purpose of this article is to show that the unit of translation (UT), which the author calls transleme, can only be bi-textual; for its existence depends on the model of translation equivalence underlying the "textual pair", that is, source text (ST) and target text (TT). From this view, a brief outling of the several approaches to the topic of the UT is attempted, and some methodological pitfalls and their implications for the specification of such a unit are dealt with.