A CORPUS-BASED MODEL OF THE ECONOMY OF SUSPICION IN AGATHA CHRISTIE'S NOVELS, 1920–1929

Agatha Christie is celebrated for hiding the culprit in plain sight, yet recent scholarship on her work has been predominantly thematic, and the mechanics of concealment remain unmeasured. This article models how Christie distributes suspicion among her characters in all eight novels she published in 1920–1929 (235 chapters; c. 557,000 words). For 64 characters, 10,706 name mentions were identified, and a sentence-level Suspicion Index (SI) was computed against a broad and a narrow suspicion lexicon and assessed with exact, permutation and Mann–Whitney tests. Three layers emerge. The chief red herring is among the three most frequently named characters in every novel that has one (p = .0012). Peak suspicion density falls, in all eight novels, on a secondary “lightning rod” who is neither the culprit nor the red herring. The culprit’s SI is indistinguishable from that of innocent background characters (p = .34 and .76), and no culprit is the least-suspected character. The “least likely person” view is therefore not supported: Christie hides the murderer within the average of suspicion. Three concealment strategies (in plain sight, at the margin and in the narrating voice) and a double-bluff exception are identified, with implications for translation, teaching and computational story generation.