When investigating the nature of culinary works, it is easy to take for granted that they all share the same ontology. This paper argues that, on the contrary, the ontology of culinary works is really threefold. Some culinary works are edible concrete particulars, or dishes, as many of us may first assume. But others are types, or multiply realisable abstract entities. And, while some of these types are determined by one recipe, others are rather chased after by their indefinitely many (...) recipes. So, there are really three kinds of culinary works; only those belonging to one of the three are edible per se; and, each kind has a very different relationship to recipes. Indeed, it is very doubtful that culinary works consisting in edible concrete particulars are suitable to have one or more recipes: by exploring what are the requirements for being a recipe, the paper also examines under what necessary conditions there is a recipe for preparing a culinary work qua concrete particular. (shrink)