This paper makes three main claims. First, it points out that scholars involved in metonymy research do not make it sufficiently clear whether they espouse an essentialist position, whereby metonymy is taken to be a Platonic concept; this engenders conceptual and analytical inconsistences. Second, it argues that scholars do not distinguish clearly between general cognitive abilities (Langacker’s reference point ability) and their correlates in language; this obfuscates the importance of form in linguistic analysis. Third, adopting a radical stance akin to Croft’s (2001), this paper claims that each “metonymic” example should be analysed in its own right by relying on a variety of parameters that do not necessarily cohere into clear-cut categories, not even prototype-based ones.