This paper reads Arthur Conan Doyle’s Sherlock Holmes stories beside James Joyce’s “Ithaca” chapter of Ulysses. I argue that both authors critique the social sciences as practiced by demographers like Charles Booth. For Doyle and Joyce, knowledge about people is unreliable, the knower is always implicated in the production of the known, and the “facts” are prone to cultural distortions. They engage the “culture versus science” debates of Matthew Arnold and T. H. Huxley and suggest that these two modes of knowledge production, the scientific and the cultural, are interrelated or even inseparable. The high modernist literary experimentalism of the 1920s has, in this respect, an antecedent in popular detective fiction. Both genres parody social science’s claims to encyclopedic knowledge of people and imply that human beings defy stable categorization.