This paper focuses on three collections of languages, in the form of collections of parts of texts: the Traité des chiffres by Blaise de Vigenère (1586), the Thresor de l’histoire des langues de cest univers by Claude Duret (1613) and the Traittez des langues estrangeres, de leurs alphabets et des chiffres by François Colletet (1660). To make an inventory of the world during the Renaissance involved, along with the inventory of its territories, its plants and its animals, the inventory of its languages. It is the task that Vigenère, Duret and Colletet assigned themselves, and is the reason their works function as cabinets of linguistic curiosities. They are assemblages of two kinds of ‘parts’: first, samples of languages (alphabets and prayers); secondly, fragments of texts, written by humanists or missionaries especially, about languages. These three pieces of work, though presenting some differences and being separated by almost a century, are deeply connected. The idea is to consider the practice of collecting and its relations with the elaboration of books concerning languages. How does the literary habitus (i.e. incorporated by those who work on or with books) weigh on how they write? What did it mean in the sixteenth and seventeenth centuries to elaborate knowledge by collecting, when to read was concomitantly to write? Vigenère’s, Duret’s and Colletet’s books are indeed crossroads: built up from parts of texts, but also ‘reservoirs’, useful for later works. They are in some respects textual hubs, to which texts arrived and from which they departed. Composite texts are susceptible to ‘re-adaptation’ or reconfiguration of the knowledge they contain.