The girl sat on a blanket, one scarecrow arm clutching her middle, the other held palm out to fend off the picture taker. She wore a one-piece bathing suit that showed skin so pale it was almost blue. She'd been hiding her face, but the camera caught her looking up, eyes enormous behind thick lenses. In the distance I could make out the horizontal slash of waves meeting shore.

As I stared at the wan little face, I ached inside. What could have prompted an attack on someone so fragile? Did a stranger force her at knifepoint, then strangle and leave her to the dogs? When did she realize she was going to die? Did she scream in terror, knowing no one would hear her cries? Had she died in her own home, to be hauled off and dumped? As her eyes closed for the last time did she feel terror or resignation or hatred or numbness, or merely bewilderment? Had she felt pain?

"Compare cranial features."

Kate was pul ing X-rays from a large brown envelope and popping them onto a wal il uminator

"This is a cranial series taken just four months before Savannah disappeared."

I got my X-rays from the athletic bag and clipped them next to the hospital films. Starting with the facial views, I compared the shape of the frontal sinuses. Varying from smal and simple to large and multichambered, these hol ow spaces above the orbits are as unique to an individual as his or her fingerprints.

Savannah's sinuses rose into her forehead like a crest on the head of a cockatoo, the configuration on her hospital X-rays matching exactly the one in the skul on my film. And the surgical burr hole was clearly visible in every view, the shape and position identical on the antemortem and postmortem films.

There was no doubt that the skul unearthed in St-Basile was Savannah Claire Osprey. But could we link the skul and femora to the partial skeleton found near Myrtle Beach?

Before leaving Montreal I'd removed a sliver of bone from the shaft of one of the femora and extracted a molar from the upper jaw in the skul , thinking that if relatives could be located, or antemortem samples of the victim's tissue or blood could be recovered, DNA sequencing might confirm the suspected identity. While the dental and radiographic evidence now rendered DNA testing unnecessary for purposes of identifying the bones from Montreal, I had another goal in mind.

Using a bone saw, I cut a one-inch chunk from al of the tibiae and fibulae that Kate had been saving al these years. She watched in silence as the circular blade buzzed through the dry bone, sending up a powdery white spray

"It's not likely the hospital wil come up with samples after al this time."

"No," I agreed. "But it happens."

It was true. Gal stones. Pap smears. Blood spots. Old DNA had been found in al sorts of strange places.