Over the past few years, there has been an ongoing debate between John Collis and I about the question of how to reconstruct or interpret European Iron Age societies. While we do agree on many points, there are some major areas of disagreement as well. The latter mainly relate to the question of the definition of the term ‘Celts’, and to the question of how to use analogies in the process of reconstruction or interpretation, and what sources to use to draw potential analogies from. While Collis has argued for the use of non-European, e.g. African ethnographic sources as our main source of analogies, I have argued for using ‘Celtic’ sources as our first point of call in a strategic approach to finding useful analogies. In some of his recent work, Collis has criticised my approach for allegedly being rooted in old, outmoded culture-historical approaches to archaeological explanation and for being based on an equally outdated linguistic Stammbaum-model of cultural evolution; and has tried to explain our differences by reference to his perception of our respective academic background. This paper is a response to Collis’ most recent criticisms, as well as a counter-critique of his assumptions about the ‘true’ definition of the term ‘Celts’ and arguments for a particular choice of sources for analogies. In it, I argue that Collis’ definition of the term ‘Celts’ is rooted in an essentialist tradition, while I argue that the definition of the term should be seen as a purely nominalist one. I then discuss how analogies should be used, and why certain sources for analogies are preferable as a first point of call in a strategic research process, despite the fact that ultimately, we should consider all possible sources for analogies equally; and that all explanatory models created by means of using analogies must always be tested against the evidence, and only those retained which pass this test.