Metonymy has been studied for at least a thousand years by rhetoricians, for two hundred years by historical semanticists, and four about ten years by cognitive linguists. However, they all have neglected one peculiar aspect of metonymy: its serial nature. Both historical semanticists and cognitive semanticists still use a substitution view of metonymy and focus mostly on one-to-one relations, as for example between CONTAINER and CONTENTS. They also focus mostly on discourse-dependent metonyms, such as "The appendicitis wants a cup of tea". In this article we want to study in depth an aspect of metonymy which distinguishes it quite profoundly from metaphor and the radial network of senses that surrounds a metaphoric source-word, namely the way metonyms form chains. These chains are either synchronic lexicalised chains, as in the case of paper (material, product, contents) or diachronic chains, as in the case of lavatory (container, room, special room, container). The article has two parts: a theoretical part in which we discuss the differences between metaphor and metonymy, between metonymy and ellipsis, and between the referential-pragmatic and the cognitive-conceptual view of metonymy; an empirical part, in which we analyse a number of diachronic metonymic chains. The article concludes with a discussion of the cognitive significance of what we call 'serial metonymy'. If one of the cognitive prerequisites of language is the ability to infer the referential intentions of others, serial metonymy can be seen as the result of this ability in (linguistic) action. (Brigitte Nerlich)