The case for public funding of research without a foreseeable application is usually made with a list of accidents: the laser, the web, magnetic resonance imaging, each traced back to work undertaken for other reasons. The list is accurate and it is a weak argument, because it is assembled retrospectively from the successes and says nothing about the base rate. A defence of basic research that rests on the discoveries it happened to produce concedes the premise that its value lies in downstream application, and having conceded it, is obliged to compete on evidence it does not have. The stronger case is uncomfortable to make in a funding submission: that a society capable of investigating what it does not understand is worth having, and that the applications, when they come, are a by-product rather than the justification.