THE MAN WHO GREW TWO BREASTS and Other True Tales of Medical Detection by Berton Roueche (Dutton: $21.95; 197 pp.) Doctors make mistakes. Hospitals too. (One is reminded of the man who submitted a specimen only to learn from the lab that his "horse has diabetes.") Mistakes make news. Successes have been lost in the shuffle, unless they came under the scrutiny of the late Berton Roueche. Roueche's "Annals of Medicine" rubric engrossed readers of the New Yorker for decades and helped to balance the medical scales with true tales of mysteries and marvels. Seven of his pieces are reprinted here, along with three unpublished accounts, all dealing with eldritch ailments and the persevering doctors who tracked them to the source; all are spiced with quirky asides and advice. (Van Gogh's madness may be traced in part to his penchant for eating leaded paint; "no pain, no gain" is nonsense: "Pain is not a badge of honor. It is a sign of tissue injury.") The title phenomenon, for example, is called gynecomastia; a 6-year-old boy similarly afflicted was relieved when his malady was traced not to the vitamins he took but to the machine that stamped the pills with a brand name. Members of a poker club in Nova Scotia were found to have come down with Query fever, first noticed in a slaughterhouse in Brisbane, Australia, in 1935. (Hint: Cherchez le chat ). The plague in rat-free Denver? Turns out that plague fleas indeed prefer rats, but when none is available. . . . Kudos, then, to Roueche's tenacious and open-minded physicians, one of whom points out: "When you hear hoofbeats, you don't necessarily think of a zebra."