Angela Cannings's successful appeal against her convictions for murder has revived an old controversy about the competence of juries to evaluate expert evidence. In response to criticisms of the jury system in the wake of a series of controversial poisoning trials, the Victorian jurist J.F. Stephen argued that juries were well equipped to decide on behalf of the community which experts should be treated as authorities, whose opinions the lay public could accept for practical purposes as ‘beyond reasonable doubt’. Such practical decisions did not, Stephen argued, require that juries fully understand the experts’ reasons for their conclusions. This article draws on recent work in social epistemology to argue that Stephen's view of the jury remains tenable, and that his authoritarian arguments can be recast in more democratic terms. It also concurs in Stephen's blunt recognition that the courts’ need to make decisions despite the uncertainties of science renders some convictions of the innocent inevitable.