Morton went further. With an impaired clotting system, he argued, Lucas' body could not handle the hours of in-and-out pressure from being hooked up to a ventilator. That was why his lungs began to bleed on day two in the hospital, Morton reasoned.

Hospital records showed that Lucas got a first dose of vitamin K at 9 a.m. that day - more than 17 hours after he was admitted, and about six hours after his lungs began to hemorrhage and his heart rate crashed.

The bleeding lessened after the vitamin dose, but by then, Morton wrote, the baby had suffered irreversible brain damage. If the vitamin had been given soon after admission, he contended, Lucas would be alive today.

The impact of the vitamin dose was enough to tell Morton that a bleeding disorder was at work. But there was also the PIVKA-II test, which came back six days after Lucas died. It's an unusual test, performed by only a few labs in the country, and it takes time.

But it is considered the most specific way to pinpoint a vitamin K problem. Clearly, Morton wrote, Lucas had a problem.

"The opportunity was missed," Morton wrote. "The infant died of a treatable bleeding disorder."

Lisa had been saying that all along. But she worried that the legal system wouldn't agree.

The showdown

The attorney general's board met in November 2005 in Harrisburg.

Morton and Rorke-Adams presented their case.

Then came Paul J. Bellino, a Geisinger pediatrician and specialist in diagnosing abuse, along with Samuel Land, an independent pathologist who had ruled Lucas' death a homicide.

Committee proceedings are not public. But in Land's written report, he concluded that Lucas had died of blunt-force trauma to the head. He wrote that while there was no sign of external trauma, the hemorrhaging and the rib fracture meant Lucas had been killed.

Bellino's report stated that it wasn't clear if the baby's head had struck a hard surface, as Land concluded, but that the internal injuries had been as bad as those that would occur in a high-speed, head-on car crash. Bellino cited the rib fracture, assorted bruises, and the brain and retinal hemorrhages, which he said cannot occur spontaneously in a child.

"Given these findings, there can be no explanation other than this child was the victim of shaken-baby syndrome," Bellino wrote.

Neither Land nor Bellino was made available to comment on the case. But Geisinger's chief medical officer, Bruce H. Hamory, said in a July 9 interview that he supported Bellino and his hospital colleagues in their findings.

"I'm certain there is a slight tendency on the part of experienced pediatricians everywhere, if there is an error to be made, to err on the side of a report" to the authorities, Hamory said. "And that's because of a concern for the safety of children."

Asked if his Geisinger colleagues might have been wrong, Hamory said the cases of both Sara and Lucas were well within the bounds of what should be reported as suspected abuse, though he said the hospital accepted the district attorney's decision to drop charges in the Amish case.

"I think in any instance, there is always a chance that somebody could be wrong, whether that's Dr. Bellino or Dr. Morton or whoever. That's why, at the end of the day, there is a legal proceeding."

Prosecutors are not doctors, but in cases with little else but medical evidence, they must take their cue from them. The state must make black-or-white decisions when the medical experts may span the spectrum of gray.