Adopting the viewpoint of abstractive morphology, and in line with Blevins (2016), the article shows that the classification of sets of words along the pattern of inflectional paradigms can be inferred from clues provided by the words themselves (internal evidence). This situation is not observed with derivational series because the coherence of a given series is based on information that comes from other series or context (external evidence). As for derivational paradigms, which are conceived of as conceptually rooted networks, derivational series play a crucial role in the way the various cells they involve can be discriminated.