The doctors, as the saying goes, were stumped. Doctors are always stumped at some point in a Berton Roueche' article, just as detectives are always stumped at some point in an Agatha Christie story. The facts were all assembled -- nearly 100 persons between February and July had been struck by the same mysterious ailment. All of them were residents of a three-block area of garden apartments in Queens. All of them had what looked like virulent cases of chicken pox, with raging fevers, swollen glands and pimplish rashes, but most of the victims had already had chicken pox. By midsummer some residents of the apartments were moving away rather than risking sickness.

The facts were all there. But the facts didn't seem to be enough. Federal epidemiologists (specialists in the mystery-cracking branch of medicine) and local physicians were all at a loss to explain just what was striking everybody and how the illness was being passed along. It took one man with a hunch -- in this case, an exterminator with a special expertise in ticks and mites and the incredible patience required to find mite-bearing mice in the apartments' basements -- to solve the mystery. In most of Berton Roueche''s stories, all it takes is one man or woman with a hunch, or a fresh view of the facts or a talent for making links between apparently unrelated cases. And the reader is along for the delightful ride as Roueche' weaves his stories, fact by implacable fact.

Roueche' has perfected a style that has been imitated by a generation of science writers. This is his 12th nonfiction book; like most of its predecessors, it is a collection of his "Annals of Medicine" column in The New Yorker. We read about the sweet-tempered Philadelphia couple in their sixties who suddenly begin to act deranged (diagnosis: carbon monoxide poisoning from a blocked furnace flue); the violent seesaw of fever and chills in an 18-year-old newlywed in Georgia (diagnosis: malaria, contracted when an Army sergeant, recently returned from South Korea, drove through the neighborhood in his convertible with the top down); the raging fevers and vomiting of a 6-year-old girl in Denver (diagnosis: the plague, contracted by touching a dead squirrel in the city park).

The Roueche' style includes some deft touches. His best tales begin almost clinically, in unemotional prose that seems to

The reviewer is the author of the forthcoming "How a Woman Ages." say that the facts of the story are themselves so incredible that just stating them will capture your rapt attention. In most cases this is absolutely true.

What reader can stop after reading: "Around ten o'clock one September evening in 1934, a native of Puerto Rico, whose name shall here be Roberto Ramirez, was sitting at ease in the kitchen of his three-room flat on West 114th Street, a cigarette in his hand and a bottle of beer at his elbow, when he became aware of an odd and pungent odor," or, "Among the more obviously ailing suppliants who appeared at Mount Sinai Hospital, on upper Fifth Avenue, on the afternoon of Tuesday, February 15, 1949, was a man I'll call Arnold Schneider. Schneider was only thirty-seven years old, but that day he could have passed for sixty-five or seventy."

There are some problems with the pieces, the oldest of which was first published in 1947. Although they are arranged in roughly chronological order, their year of publication is stated only at the end. This can make for some confusion. In one story, Roueche' charts the case of a man with leprosy that was first diagnosed in 1934. It is only at the end of the story, when the man has been followed until his death in 1951, that we realize that the "now" in the narrative refers to a time nearly 20 years after the story begins -- and a time more than 30 years before the story appears in this collection.

There is probably nothing wrong with presenting the stories exactly as they first appeared, without distracting inserts and parenthetical updates. But some updating would have helped ease confusion. When salmonellosis (a form of food poisoning) is first described, in the story of a 1954 outbreak spread by contaminated watermelon, we are told that there are "now" more than 400 species of salmonella. Some 150 pages later, in another story, the author says that 1,500 species are "now" known. How many other times, one wonders, is a statement that is given as fact no longer true?

These complaints are minor, though, next to the great pleasure of reading another Roueche' collection. Where else can you find such true-life dramas as the five people in Harlem who died of tetanus transmitted through impure heroin, or the 33-year-old Topeka housewife wracked by muscle spasms so severe that she broke and rebroke both hips while bedridden? Where else can you read in detail the eight-day nightmare of a man who forgot his identity, or the two-week ordeal of a woman covered head to toe with poison ivy rash? This book leaves one believing, along with one of the experts quoted here, that life is "not that dependable. In fact, it's the most delicate thing in the world. It's a miracle. And one false step and it's over."