Such an outcome was not necessarily mysterious. The early days of African exploration were incredibly hazardous. Even carefully managed expeditions lost half of their party or more. Those who did not succumb to malaria, sleeping sickness, and blackwater fever faced rivers teeming with crocodiles and hippos, jungles with leopards and suspicious, cannibalistic natives. And, for all its luxuriant growth, the rain forest provided little edible food; a number of expeditions had starved to death.
“I began,” Ross said to Elliot, “with the idea that the city existed, after all. Assuming it existed, where would I find it?”
The Lost City of Zinj was associated with diamond mines, and diamonds were found with volcanoes. This led Ross to look along the Great Rift Valley—an enormous geological fault thirty miles wide, which sliced vertically up the eastern third of the continent for a distance of fifteen hundred miles. The Rift Valley was so huge that its existence was not recognized until the 1890s, when a geologist named Gregory noticed that the cliff walls thirty miles apart were composed of the same rocks. In modern terms the Great Rift was actually an abortive attempt to form an ocean, for the eastern third of the continent had begun splitting off from the rest of the African land mass two hundred million years ago; for some reason, it had stopped before the break was complete.
On a map the Great Rift depression was marked by two features: a series of thin vertical lakes—Malawi, Tanganyika, Kivu, Mobutu—and a series of volcanoes, including the only active volcanoes in Africa at Virunga. Three volcanoes in the Virunga chain were active: Mukenko, Mubuti, and Kanagarawi. They rose 11,000— 15,000 feet above the Rift Valley to the east, and the Congo Basin to the west. Thus Virunga seemed a good place to look for diamonds. Her next step was to investigate the ground truth.
“What’s ground truth?” Peter asked.
“At ERTS, we deal mostly in remote sensing,” she explained. “Satellite photographs, aerial run-bys, radar side scans. We carry millions of remote images, but there’s no substitute for ground truth, the experience of a team actually on the site, finding out what’s there. I started with the preliminary expedition we sent in looking for gold. They found diamonds as well.” She punched buttons on the console, and the screen images changed, glowing with dozens of flashing pinpoints of light.
“This shows the placer deposit locations in streambeds near Virunga. You see the deposits form concentric semicircles leading back to the volcanoes. The obvious conclusion is that diamonds were eroded from the slopes of the Virunga volcanoes, and washed down the streams to their present locations.”
“So you sent in a party to look for the source?”
“Yes.” She pointed to the screen. “But don’t be deceived by what you see here. This satellite image covers fifty thousand square kilometers of jungle. Most of it has never been seen by white men. It’s hard terrain, with visibility limited to a few meters in any direction. An expedition could search that area for years, passing within two hundred meters of the city and failing to see it. So I needed to narrow the search sector. I decided to see if I could find the city.”
“Find the city? From satellite pictures?”
“Yes,” she said. “And I found it.”
The rain forests of the world had traditionally frustrated remote-sensing technology. The great jungle trees spread an impenetrable canopy of vegetation, concealing whatever lay beneath. In aerial or satellite pictures, the Congo rain forest appeared as a vast, undulating carpet of featureless and monotonous green. Even large features, rivers fifty or a hundred feet wide, were hidden beneath this leafy canopy, invisible from the air.
So it seemed unlikely she would find any evidence for a lost city in aerial photographs. But Ross had a different idea: she would utilize the very vegetation that obscured her vision of the ground. –
The study of vegetation was common in temperate regions, where the foliage underwent seasonal changes. But the equatorial rain forest was unchanging: winter or summer, the foliage remained the same. Ross turned her attention to another aspect, the differences in vegetation albedo.
Albedo was technically defined as the ratio of electromagnetic energy reflected by a surface to the amount of energy incident upon it. In terms of the visible spectrum, it was a measure of how “shiny” a surface was. A river had a high albedo, since water reflected most of the sunlight striking it. Vegetation absorbed light, and therefore had a low albedo. Starting in 1977, ERTS developed computer programs which measured albedo precisely, making very fine distinctions.
Ross asked herself the question: If there was a lost city, what signature might appear in the vegetation? There was an obvious answer: late secondary jungle.
The untouched or virgin rain forest was called primary jungle. Primary jungle was what most people thought of when they thought of rain forests: huge hardwood trees, mahogany and teak and ebony, and underneath a lower layer of ferns and palms, clinging to the ground. Primary jungle was dark and forbidding, but actually easy to move through. However, if the primary jungle was cleared by man and later abandoned, an entirely different secondary growth took over. The dominant plants were softwoods and fast-growing trees, bamboo and thorny tearing vines, which formed a dense and impenetrable barrier.
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 imagery—”
“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.
Now Elliot wanted a program that would distinguish between similar visual images of gorillas and chimps. Seamans could not help asking, “Why? It’s pretty obvious. A gorilla is a gorilla, and a chimp is a chimp.”
“Just do it,” Elliot said.
“Can I use size?” On the basis of size alone, gorillas and chimps could be accurately distinguished. But visual functions could not determine size unless the distance from the recording instrument to the subject image was known, as well as the focal length of the recording lens.
“No, you can’t use size,” Elliot said. “Element morphology only.”
Seamans sighed. “Thanks a lot. What resolution?”
“I need ninety-five-percent confidence limits on species assignment, to be based on less than three seconds of black-and-white scan imagery.”
Seamans frowned. Obviously, Elliot had three seconds of videotape imagery of some animal and he was not sure whether it was a gorilla or not. Elliot had seen enough gorillas over the years to know the difference: gorillas and chimps were utterly different animals in size, appearance, movement, and behavior. They were as different as intelligent oceanic mammals—say, porpoises and whales. In making such discriminations, the human eye was far superior to any computer program that could be devised. Yet Elliot apparently did not trust his eye. What was he thinking of?
“I’ll try,” Seamans said, “but it’s going to take a while. You don’t write that kind of program overnight.”
“I need it overnight, Tom,” Elliot said. “I’ll call you back in twenty-four hours.”
3. Inside the Coffin
IN ONE CORNER OF THE 747 LIVING MODULE WAS A sound-baffled fiberglass booth, with a hinged hood and a small CRT screen; it was called “the coffin” because of the claustrophobic feeling that came from working inside it. As the airplane crossed the mid-Atlantic, Ross stepped inside the coffin. She had a last look at Elliot and Amy—both asleep, both snoring loudly—and Jensen and Irving playing “submarine chase” on the computer console, as she lowered the hood.
Ross was tired, but she did not expect to get much sleep for the next two weeks, which was as long as she thought the expedition would last. Within fourteen days—336 hours— Ross’s team would either have beaten the Euro-Japanese consortium or she would have failed and the Zaire Virunga mineral exploration rights would be lost forever.
The race was already under way, and Karen Ross did not intend to lose it.
She punched Houston coordinates, including her own sender designation, and waited while the scrambler interlocked. From now on, there would be a signal delay of five seconds at both ends, because both she and Houston would be sending in coded burst transmissions to elude passive listeners.
The screen glowed: TRAVIS.
She typed back: R OS S. She picked up the telephone receiver.
“It’s a bitch,” Travis said, although it was not Travis’s voice, but a computer-generated flat audio signal, without expression.
“Tell me,” Ross said.
“The consortium’s rolling,” Travis’s surrogate voice said. “Details,” Ross said, and waited for the five-second delay. She could imagine Travis in the CCR in Houston, hearing her own computer-generated voice. That flat voice required a change in speech patterns; what was ordinarily conveyed by phrasing and emphasis had to be made explicit.
“They know you’re on your way,” Travis’s voice whined. “They are pushing their own schedule. The Germans are behind it—your friend Richter. I’m arranging a feeding in a matter of minutes. That’s the good news.”
“And the bad news?”
“The Congo has gone to hell in the last ten hours,” Travis said. “We have a nasty GPU.”
“Print,” she said.
On the screen, she saw printed GEOPOLITICAL UPDATE, followed by a dense paragraph. It read:
ZAIRE EMBASSY WASHINGTON STATES EASTERN BORDERS VIA RWANDA CLOSED / NO EXPLANATION / PRESUMPTION 101 AMIN TROOPS FLEEING TANZANIAN
INVASION UGANDA INTO EASTERN ZAIRE / CONSEQUENT DISRUPTION / BUT FACTS DIFFER / LOCAL TRIBES {KIGANI} ON RAMPAGE / REPORTED ATROCITIES AND CANNIBALISM ETC / FOREST—DWELLING PYGMIES UNRELIABLE / KILLING ALL VISITORS CONGO RAIN FOREST / ZAIRE GOVERNMENT DISPATCHED GENERAL MUGURU (AKA BUTCHER OF STAN— LEYVILLE) / PUT DOWN KIGANI REBELLION ‘AT ALL COSTS’ / SITUATION HIGHLY UNSTABLE / ONLY LEGAL ENTRY INTO ZAIRE NOW WEST THROUGH KINSHASA / YOU ARE ON YOUR OWN / ACQUISITION WHITE HUNTER
MUNRO NOW PARAMOUNT IMPORTANCE WHATEVER COST / KEEP HIM FROM CONSORTIUM WILL PAY ANYTHING / YOUR SITUATION EXTREME DANGER / MUST HAVE MUNRO TO SURVIVE /
She stared at the screen. It was the worst possible news. She said, “Have you got a time course?”