It was Wednesday morning, Sept. 23, when Dr. Courand Rothe picked up the phone in his office at the San Antonio Metropolitan Health District and called Atlanta. Rothe was worried. The city was in the middle of an outbreak of typhoid fever, the worst in the United States since 1973. At least 15 cases had been diagnosed during the month and there appeared to be no link.

Rothe wanted help.

In Atlanta, Dr. Wayne Shandera was in his office at the Centers for Disease Control, where he is one of 115 members of the CDC's Epidemic Intelligence Service. They are a kind of medical SWAT team, on call 24 hours a day to respond to outbreaks of disease.

On this Wednesday, Shandera's number was up, and he believed he might be on his way to Chicago to look into an outbreak of campylobacter, a new strain of diarrhea. Then one of his superiors, Dr. Mitch Cohen, came into his office. "You may be going to Texas this evening," Shandera remembers Cohen saying.

For Shandera, the assignment had special meaning. He had grown up in San Antonio and graduated from high school in one of the older neighborhoods on the city's north side. Now he was being called back by his hometown to help solve the typhoid riddle.

Shandera, 29, is an epidemiologist -- a medical detective. The tools of his trade are not a black bag or a microscope, but intuition, shoe leather and a knowledge of statistics. There are about 500 cases of typhoid nationally every year, and for about two-thirds of them, the source is never found. With the count rising rapidly in San Antonio, Shandera was being asked to beat the odds.

When Shandera arrived the next day at the public health office a block south of city hall, he was impressed with what local officials already knew. Four public health nurses had been tracking each new case of typhoid, and had put together "surveillance forms" on each.

In their tiny office, a city map on the wall pinpointed the residence of each case. The pins showed that the outbreak was largely concentrated in the Mexican-American neighborhoods on the west side of town. But despite interviews conducted by the public health nurses, there were no solid leads.

In any typhoid outbreak, there are two suspects: food and water. But the disease is usually transmitted through human hands, by someone who is carrying typhoid bacteria. The process of transmitting it is simple.

The bacteria are excreted in human feces, and if the carrier doesn't wash his hands after a bowel movement and then touches food, the food is contaminated. City officials had quickly ruled out the water supply as the source of the fever, which was once deadly but is now treated relatively simply with antibiotics.

That left food as the suspect. But where? Any number of things came to mind: restaurants, fast-food outlets, church festivals, large picnics, school gatherings. The list was almost infinite, and as the team of experts gathered in Rothe's office last Thursday, the number of typhoid cases was about to grow to 20.

The team that met with Rothe that morning included Shandera; Dr. Tom Betz, 36, a CDC official assigned to the Texas Department of Health in Austin; Jeff Taylor, 28, a epidemiological investigator who had come with Betz; Dr. R. F. Bell, head of Rothe's communicable disease division.

After reassuring themselves that water was not the problem, the team members set out on the more difficult task of trying to identify the food outlet that was the source. "Tom and I decided that the only way to find out was to talk to all the people individually," Shandera said.

The three epidemiologists spent much of Thursday afternoon refining a questionnaire developed by the nurses. It was a long form that asked whether the patients had eaten various foods during August or September and where they had purchased the foods. The list looked like a menu from a Mexican restaurant: tamales, tortillas, tacos, enchiladas, raspas, and barbacoa, a speciality that is the meat from a cow's head that has been steamed under pressure for eight to 10 hours, picked off the bone and served with tortillas.

That evening, the group fanned out around the city to conduct their interviews. What they were looking for was a common thread, the name of an eating establishment that somehow linked all the cases. At 10:30 p.m. they quit in discouragement. "We came up with absolutely nothing," Shandera said. "It was totally fruitless."

Shandera, Betz and Taylor were working together on the interviews, and had saved three for Friday morning at Baptist Hospital. None of these three patients lived on the west side, and the team hoped they might provide the clue they were looking for.

They got it from a 13-year-old boy. The boy's mother remembered going to a tortilla mill and Mexican take-out restaurant called La Frontera on Old Highway 90. Excited, the team members put the same question to the other two typhoid patients at the hospital. Yes, they had purchased barbacoa or tortillas at La Frontera.

Shandera put in a quick call to the nurses and told them to ask other patients if they had eaten at La Frontera. When the responses came back positive, the medical detectives knew they had a lead.

But it was far from conclusive. All but one of the patients had purchased barbacoa or tortillas at La Frontera, but at least three-fourths of them also had purchased food at one of the fast-food chicken outlets on the west side, and many of them had eaten sno-cones purchased from vendors along the streets. A new round of interviews was needed, and this time the team used what is called the "case control method."

Under case control, the team needed a control group of interviewees, persons of the same sex, roughly the same age and from the same neighborhood as the typhoid patients. Betz and Taylor drew up a 22-question form that asked specific questions about food purchased at La Frontera and several other suspected restaurants.

Shandera outlined the method for finding control interviews. On Friday night, they were ready to go again. "I was confident it was La Frontera," Betz said. "It was the only thing that fit. But if it didn't work, we were back to square one, which is a tight position to be in."

But by Saturday noon, frustration had set in again. Shandera, Betz and Taylor met for lunch on the west side and shared their results. Not only the typhoid victims had gone to La Frontera. So had most of the control group. "We weren't finding anything different," Betz said.

But in the outlying areas, another pattern was emerging from the control group interviews. Virtually no one had been to La Frontera, and when the seven interviewers reassembled at the public health office that afternoon to combine their data from the 75 interviews, they knew they had found the source.

It took several hours to distill the information, and then with a $19 pocket calculator, Shandera and Betz determined the mathematical probability that it was strictly a coincidence that all the victims had been to La Frontera. The odds were less than 1 in 100,000.

Shandera called Rothe at 9:30 p.m. with the final figures, which had been run on the CDC computer in Atlanta.

At 6 a.m. on Sunday, the team converged on La Frontera and ordered it closed. Shandera, exhuasted, slept most of the day.

But the case was still not completely solved. The team needed to find bacterial evidence of typhoid on the scene to back up the epidemiological evidence. That work is done in the laboratory, and the group that closed La Frontera quickly gathered stool and blood samples from the employees and shipped them to the enteric bacteriology lab in Austin.

At 9:40 a.m. yesterday CDT , Tom Betz was in his Austin office when the phone rang. It was Gloria Pierce, head of the state bacteriology lab. Betz listened for a moment and then turned to Taylor. "We've got the carrier," he said.

Pierce's analysis of the stool samples showed that a 26-year-old woman who worked at La Frontera had the typhoid germ. Betz called Rothe in San Antonio with the good news. "You've got everything nailed down," he said.

But Betz turned out to be overly optimistic. The woman did not begin working at La Frontera until late August, some weeks after the disease had first been transmitted to the outlet's customers. Later today, tests showed traces of the same germ in another employe of La Frontera.

Shandera, though greatly encouraged by the findings, called Betz and told him not to call the woman a carrier. Shandera later said the actual carrier may never be found.

And with the incubation period for the disease of one to three weeks, more cases undoubtedly will be diagnosed. As of yesterday, the count was up to 51.

But for San Antonio, the worst is over. "The outbreak is hopefully under control," Shandera said yesterday.