This paper presents a meta-analysis of the “cluster literature” contained in scientific journals from 1969 to 2007. Thanks to an original database we study the evolution of a stream of literature which focuses on a research object which is both a theoretical puzzle and an empirical widespread evidence. We identify different growth stages, from take-off to development and maturity. We test the existence of a life-cycle within the authorships and we discover the existence of a substitutability relation between different collaborative behaviours.  We study the relationships between a “spatial” and an “industrial” approach within the textual corpus of cluster literature and we show the existence of a “predatory” interaction. We detect the relevance of clustering behaviours in the location of authors working on clusters and in measuring the influence of geographical distance in co-authorship. We measure the extent of a convergence process of the vocabulary of scientists working on clusters.