In the first section of this paper I present the measurement-theoretic fallacy of 'over-assignment of structure': the unwarranted assumption that every numeric relation holding among two (or more) numbers represents some empirical, physical relation among the objects to which these numbers are assigned as measures (e.g., of temperature). In the second section I argue that a generalized form of this fallacy arises in various philosophical contexts, in the form of a misguided, over-extended application of one conceptual domain to another. Three (...) examples are given: (i) the reduction of arithmetic into set theory, (ii) the ascription of full-blown intentional states to (at least some) non-overtly intentional creatures, such as Wittgenstein's builders, and (iii) the analysis of some modal notions as involving quantification over possible worlds (or their substitutes). The discussion of the third example gives rise to a novel account of possible-worlds talk. (shrink)