Sir Arthur Conan Doyle made a significant contribution to the campaign to free Oscar Slater, wrongly convicted of murder in 1909, and imprisoned for eighteen and a half years. This paper examines the trial of Oscar Slater in the light of the argument made by the historian Carlo Ginzburg that the nineteenth century saw the development of a new evidential paradigm as exemplified by the method of Conan Doyle's creation Sherlock Holmes. This is discussed in the context of the development of the `reconstructive' trial in the late nineteenth-century, by looking at changes in the types of evidence admitted to trials, at the changes in the law of evidence and at the relation between the detective and legal counsel. It argues that, like detective fiction, the trial was structured around a `hermeneutics of suspicion' that was institutionalised in certain features of the reconstructive trial. This encouraged jurors and other observers to distrust appearances and to make judgments based on their interpretation of the evidence and the appearance of the accused. Ironically, this contributed to the miscarriage of justice in the case of Oscar Slater that Conan Doyle later sought to challenge by use of the same method.