A point made in the third of these essays, on the value of history, was the widespread human enjoyment of a good story. It was suggested that history played a part in satisfying this need. The consistent success of fiction based on a simple form of counterfactual history — Robert Harris's "Fatherland" is a good recent example — seems to indicate that this type of history is equally appealing. Sometimes known as "what if", or "alternative" history, or, in the title of a recent collection of serious counterfactual essays edited by Niall Ferguson, "virtual history", it takes as its starting point some historical event, assumes that it turned out differently — Harold wins at Hastings or Napoleon at Waterloo — and develops a possible course of events from then on.

All of this has great potential for some intriguing speculation, particularly so in the case of events within living memory, as shown by Harris's bestseller. However is that all that can be said for counterfactual history? Is it no more than a supply of good storylines for novelists, shading perhaps into something not unlike science fiction? It certainly does do these things, no doubt thereby adding to the gaiety of nations, but I believe that there are some more serious points to be made in its favour, and that it is a wider concept than what has just been described. We shall however have to attempt to deal with some highly unfavourable opinions of counterfactualism held by many professional historians. Consideration of these views may in fact help us to a better understanding of the true meaning of counterfactualism.

I think there is certainly one dimension of alternative history which has as much philosophical content as merely fictional; it seems to provide philosophers of a particular metaphysical bent with some welcome grist to their mill. I refer to those for whom the notion of parallel worlds holds fascination, since every venture into alternative history involves by definition the creation of a parallel world. As I shall try to show, such parallel worlds, if they are to contain any meaningful truth, will need to correspond as closely as possible to the real world, but it is clear that an element of conjecture, or fantasy if a harsher term is preferred, is a necessary part of any such exercise. While this necessary limitation may lie behind the reluctance to accord counterfactualism academic respectability, it in no way invalidates the contention that there does seem to be a real human urge, and therefore legitimate grounds for philosophical investigation, to ponder over the questions raised by such fictional works as the "Back to the Future" films, or novels like Wyndham's "Random Quest"; the latter is of course a book, and subsequently a film, whose plot is firmly embedded in a world where twentieth century history has turned out very differently.

Good counterfactual history calls for the consideration not only of the main counterfactual — Napoleon's victory at Waterloo — but also for an intelligent appraisal of what in a different context E H Carr called the "significant" facts among a potentially infinite number of knock-on changes, the "Garden of Forking Paths" of the eponymous short story by Jorge Luis Borges. Such a need for careful analysis of cause is inherent in any serious writing of counterfactual history, and seems to me to speak strongly in its favour as a discipline for any historian.

There is one form of counterfactualism which we are all constantly using throughout our lives. Every time a businessman runs a sensitivity test on the economics of a new project he is undertaking a comparison of a number of possible worlds, of which at best one, more probably none, will represent what actually takes place, and obviously is itself not counterfactual. The businessman is working before, not after the event, but he is basing his assessment partly on his knowledge of the past. A different time perspective is found in certain legal situations,