A context and introduction are provided for nine contributed articles on Forecasting Environmental Change. Since the first quantitative studies of the ‘predicament of mankind’ at the end of the 1960s, large‐scale forecasting models have attracted substantial criticism on the grounds of failing both to inform the policy process and to embrace the notion of surprise‐rich futures. Almost all attempts at developing models of environmental systems, at whatever scale, have been dogged by the sceptic's question: Is the model capable of predicting conditions substantially different from those observed in the past? The question, strictly speaking, is not answerable. The paper defines therefore three approaches to the development of modelsthe mechanical, metric, and linguistic paradigmsand constructs thence a reorientation of the customary problem of prediction in order to explore the concept of reachable, and radically different, futures. Such reorientation depends crucially on a juxtaposition of the linguistic and mechanical descriptions of a system's behaviour. The nine papers of this special issue are then summarized in the context provided by this reconsideration of how we formalize our thinking about the behaviour of systems, and of what might be meant by the systematic analysis of reachable futures. Copyright © 1991 John Wiley & Sons, Ltd.
