Congo

But Ross was not concerned about any aspect of the jungle except its albedo. Because the secondary plants were different, secondary jungle had a different albedo from primary jungle. And it could be graded by age: unlike the hardwood trees of primary jungle, which lived hundreds of years, the softwoods of secondary jungle lived only twenty years or so. Thus as time went on, the secondary jungle was replaced by another form of secondary jungle, and later by still another form.

By checking regions where late secondary jungle was generally found – such as the banks of large rivers, where innumerable human settlements had been cleared and abandoned – Ross confirmed that the ERTS computers could, indeed, measure the necessary small differences in reflectivity.

She then instructed the ERTS scanners to search for albedo differences of .03 or less, with a unit signature size of a hundred meters or less, across the fifty thousand square kilometers of rain forest on the western slopes of the Virunga volcanoes. This job would occupy a team of fifty human aerial photographic analysts for thirty-one years. The computer scanned 129,000 satellite and aerial photographs in under nine hours.

And found her city.

In May, 1979, Ross had a computer image showing a very old secondary jungle pattern laid out in a geometric, gridlike form. The pattern was located 2 degrees north of the equator, longitude 3 degrees, on the western slopes of the active volcano Mukenko. The computer estimated the age of the secondary jungle at five hundred to eight hundred years.

"So you sent an expedition in?" Elliot said.

Ross nodded. "Three weeks ago, led by a South African named Kruger. The expedition confirmed the placer diamond deposits, went on to search for the origin, and found the ruins of the city.,,

"And then what happened?" Elliot asked.

He ran the videotape a second time.

Onscreen he saw black-and-white images of the camp, destroyed, smoldering. Several dead bodies with crushed skulls were visible. As they watched, a shadow moved over the dead bodies, and the camera zoomed back to show the outline of the lumbering shadow. Elliot agreed that it looked like the shadow of a gorilla, but he insisted, "Gorillas couldn’t do this. Gorillas are peaceful, vegetarian animals."

They watched as the tape ran to the end. And then they reviewed her final computer-reconstituted image, which clearly showed the head of a male gorilla.

"That’s ground truth," Ross said.

Elliot was not so sure. He reran the last three seconds of videotape a final time, staring at the gorilla head. The image was fleeting, leaving a ghostly trail, but something was wrong with it. He couldn’t quite identify what. Certainly this was atypical gorilla behavior, but there was something else. –

He pushed the freeze-frame button and stared at the frozen image. The face and the fur were both gray: unquestionably gray.

"Can we increase contrast?" he asked Ross. "This image is washed out."

"I don’t know," Ross said, touching the controls. "I think this is a pretty good image." She was unable to darken it.

"It’s very gray," he said. "Gorillas are much darker."

"Well, this contrast range is correct for video."

Elliot was sure this creature was too light to be a mountain gorilla. Either they were seeing a new race of animal, or a new species. A new species of great ape, gray in color, aggressive in behavior, discovered in the eastern Congo.

He had come on this expedition to verify Amy’s dreams – a fascinating psychological insight – but now the stakes were suddenly much higher.

Ross said, "You don’t think this is a gorilla?"

"There are ways to test it," he said. He stared at the screen, frowning, as the plane flew onward in the night.

2.B-8 Problems

"YOU WANT ME TO WHAT’?" TOM SEAMANS SAID, cradling the phone in his shoulder and rolling over to look at his bedside clock. It was 3 A.M.

"Go to the zoo," Elliot repeated. His voice sounded garbled, as if coming from under water.

"Peter, where are you calling from?"

"We’re somewhere over the Atlantic now," Elliot said. "On our way to Africa."

"Is everything all right?"

"Everything is fine," Elliot said. "But I want you to go to the zoo first thing in the morning."

"And do what?"

"Videotape the gorillas. Try to get them in movement. That’s very important for the discriminant function, that they be moving."

"I’d better write this down," Seamans said. Seamans handled the computer programming for the Project Amy staff, and he was accustomed to unusual requests, but not in the middle of the night. "What discriminant function?"

"While you’re at it, run any films we have in the library of gorillas – any gorillas, wild or in zoos or whatever. The more specimens the better, so long as they’re moving. And for a baseline, you’d better use chimps. Anything we have on chimps. Transfer it to tape and put it through the function."

"What function?" Seamans yawned.

"The function you’re going to write," Elliot said. "I want a multiple variable discriminant function based on total im?agery"

"You mean a pattern-recognition function?" Seamans had written pattern-recognition functions for Amy’s language use, enabling them to monitor her signing around the clock. Sea-mans was proud of that program; in its own way, it was highly inventive.

"However you structure it," Elliot said. "I just want a function that’ll discriminate gorillas from other primates like chimps. A species-differentiating function."

"Are you kidding?" Seamans said. "That’s a B-8 problem." In the developing field of pattern-recognition computer programs, so-called B-8 problems were the most difficult; whole teams of researchers had devoted years to trying to teach computers the difference between "B" and "8′ ‘ – precisely because the difference was so obvious. But what was obvious to the human eye was not obvious to the computer scanner. The scanner had to be told, and the specific instructions turned out to be far more difficult than anyone anticipated, particularly for handwritten characters.