This article takes one form of extreme event - namely forest fires - and asks how the ways in which they come to be understood might contribute to core questions about evidence. In particular, I shall suggest that, by considering the moment-ousness of extreme events, we are offered tools for considering the tension between 'rupture' and 'continuity' which emerges throughout the articles of this Special Section, not only at the methodological level but also in terms of understanding these processes themselves. The controversial and often political relationship between evidence, modelling and prediction - that is, taking fragmentary evidence of past events to speculate on future process - is an issue to which we return in the conclusion when the question 'what are scientists for anyway?' comes to the fore. 'Uncertainty' - as a variable, or as a sign of failure - emerges as key in what the article calls a 'clash of modernities'.