Anthrax is a disease that is nearly always deadly to cows and other animals. The nineteenth-century French veterinarian Louvrier devised a treatment for anthrax that was later shown to be totally without merit. His alleged “cure” was tried on two cows, selected at random from four cows that had received a powerful dose of anthrax microbes. Of the two he treated, one died and one recovered; of the two he left untreated, one died and one recovered. The reasons for recovery were unknown. Had Louvrier tested his “cure” on the two cows that happened to live, his treatment would have received impressive but spurious confirmation. What was the probability of Louvrier choosing, for his test, just those two cows that chanced to live?