“There have been studies of the individual kinds of reservoir conditions, but nobody’s really looked into the syndrome as a whole.” Kelly sank, if anything, farther back into the couch. “Why does it happen? Why does it happen in specific parts of the body? How is it that the virus is contained? Everything we know says that anyone with a reservoir condition should amplify immediately, but they don’t. They just keep going until they die. It doesn’t make sense.”

“And that’s what you were studying?”

A marginal nod. “Uh-huh. That’s when I found it.”

“Found what?” asked Alaric.

“Look at the statistics.” Kelly sighed, tilting her face up toward the ceiling. “The first column is population. The second column is percent of population with a known reservoir condition—type is irrelevant in this instance.”

I squinted at the numbers. I’d seen the number on the third column somewhere before. I hazarded a guess: “Column three is KA-related deaths in the last year?”

“Yeah.”

“So what’s the fourth column?”

Becks spoke, voice heavy with dawning horror. She’d managed to figure things out just a little faster than the rest of us, and she didn’t sound happy about her epiphany. “Oh, my God. It’s—that’s the number of people with reservoir conditions who died, isn’t it?”

Kelly nodded.

I squinted at the numbers. They didn’t seem to mean anything. I was about to open my mouth when George said, very quietly, Lookat column two again, Shaun.

I looked. And I understood.

“This can’t be right,” I said, suddenly cold. Reservoir conditions don’t increase the odds of viral amplification; they actually tend to reduce them, since most people who suffer from a latent form of KA wind up even more paranoid about infection than the rest of the population. People like George, who went out into the field, or Emily Ryman, who kept raising horses even after she developed retinal KA, were the exception rather than the rule.

Kelly sighed, opening her eyes for the first time since her lecture had begun. “That’s what I thought,” she said, looking right at me. “I ran the numbers over and over. I had an intern pull the census data six times. It’s all accurate.”

“But—”

“Less than eleven percent of the population suffers from reservoir conditions. Last year, they accounted for thirty-eight percent of the KA-related deaths.” Kelly’s tone was grim. Suddenly, her exhaustion was starting to make a lot of sense. “Statistically speaking, this can’t be happening.”

“Maybe it was a glitch,” suggested Dave. “Statistical anomalies happen, right?”

Becks snorted. “Yeah, and respected CDC doctors totally help their employees fake death by clone over statistical anomalies. It happens all the time.”