At the first blast of machine-gun fire, Munro dived for cover, the glassware on the table splattered, one of the girls screamed, and Elliot and Ross threw themselves to the marble floor as the bullets whined around them, chipping the plaster overhead, raining plaster dust down upon them. The blast lasted thirty seconds or so, and it was followed by complete silence.

When it was over, they got up hesitantly, staring at one another.

“The consortium plays for keeps.” Munro grinned. “Just my sort of people.”

Ross brushed plaster dust off her clothes. She turned to Munro. “Five point two against the first two hundred, no deductions, in Swiss francs, adjusted.”

“Five point seven, and you have me.”

“Five point seven. Done.”

Munro shook hands with them, then announced that he would need a few minutes to pack his things before leaving for Nairobi.

“Just like that?” Ross asked. She seemed suddenly concerned, glancing again at her watch.

“What’s your problem?” Munro asked.

“Czech AK-47s,” she said. “In your warehouse.”

Munro showed no surprise. “Better get them out,” he said. “The consortium undoubtedly has something similar in the works, and we’ve got a lot to do in the next few hours.” As he spoke, they heard the police Kiaxons approaching from a distance. Munro said, “We’ll take the back stair.”

An hour later, they were airborne, heading toward Nairobi.

IT WAS FARTHER ACROSS AFRICA FROM TANGIER TO Nairobi than it was across the Atlantic Ocean from New York to London-3,600 miles, an eight-hour flight. Ross spent the time at the computer console, working out what she called “hyperspace probability lines.”

The screen showed a computer-generated map of Africa, with streaking multicolored lines across it. “These are all timelines,” Ross said. “We can weight them for duration and delay factors.” Beneath the screen was a total-elapsed-time clock, which kept shifting numbers.

“What’s that mean?” Elliot asked.

“The computer’s picking the fastest route. You see it’s just identified a timeline that will get us on-site in six days eighteen hours and fifty-one minutes. Now it’s trying to beat that time.”

Elliot had to smile. The idea of a computer predicting to the minute when they would reach their Congo location seemed ludicrous to him. But Ross was totally serious.

As they watched, the computer clock shifted to 5 days 22 hours 24 minutes.

“Better,” Ross said, nodding. “But still not very good.” She pressed another key and the lines shifted, stretching like rubber bands over the African continent. “This is the consortium route,” she said, “based on our assumptions about the expedition. They’re going in big-thirty or more people, a full-scale undertaking. And they don’t know the exact location of the city; at least, we don’t think they know. But they have a substantial start on us, at least twelve hours, since their aircraft is already forming up in Nairobi.”

The clock registered total elapsed time: 5 days 09 hours 19 minutes. Then she pressed a button marked DATE and it shifted to 06 21 790814. “According to this, the consortium will reach the Congo site a little after eight o’clock in the morning on June 21.”

The computer clicked quietly; the lines continued to stretch and pull, and the clock read a new date: 06 21 79 1224.

“Well,” she said, “that’s where we are now. Given maximum favorable movements for us and them, the consortium will beat us to the site by slightly more than four hours, five days from now.”

Munro walked past, eating a sandwich. “Better lock another path,” he said. “Or go radical.”

“I hesitate to go radical with the ape.”

Munro shrugged. “Have to do something, with a timeline like that.”

Elliot listened to them with a vague sense of unreality: they were discussing a difference of hours, five days in the future. “But surely,” Elliot said, “over the next few days, with all the arrangements at Nairobi, and then getting into the jungle-you can’t put too much faith in those figures.”

“This isn’t like the old days of African exploration,” Ross said, “where parties disappeared into the wilds for months. At most, the computer is off by minutes-say, roughly half an hour in the total five-day projection.” She shook her head. “No. We have a problem here, and we’ve got to do something about it. The stakes are too great.”

“You mean the diamonds.”

She nodded, and pointed to the bottom of the screen, where the words BLUE CONTRACT appeared. He asked her what the Blue Contract was.

“One hell of a lot of money,” Ross said. And she added, “I think.” For in truth she did not really know.

Each new contract at ERTS was given a code name. Only Travis and the computer knew the name of the company buying the contract; everyone else at ERTS, from computer programmers to field personnel, knew the projects only by their color-code names: Red Contract, Yellow Contract, White Contract. This was a business protection for the firms involved. But the ERTS mathematicians could not resist a lively guessing game about contract sources, which was the staple of daily conversation in the company canteen.

The Blue Contract had come to ERTS in December, 1978. It called for ERTS to locate a natural source of industrial-grade diamonds in a friendly or neutralist country. The diamonds were to be Type IIb, “nitrogen-poor” crystals. No dimensions were specified, so crystal size did not matter; nor were recoverable quantities specified: the contractor would take what he could get. And, most unusual, there was no UECL.

Nearly all contracts arrived with a unit extraction cost limit. It was not enough to find a mineral source; the minerals had to be extractable at a specified unit cost. This unit cost in turn reflected the richness of the ore body, its remoteness, the availability of local labor, political conditions, the possible need to build airfields, roads, hospitals, mines, or refineries.

For a contract to come in without a UECL meant only one thing: somebody wanted blue diamonds so badly he didn’t care what they cost.

Within forty-eight hours, the ERTS canteen had explained the Blue Contract. It turned out that Type JIb diamonds were blue from trace quantities of the element boron, which rendered them worthless as gemstones but altered their electronic properties, making them semiconductors with a resistively on the order of 100 ohms centimeters. They also had light-transmissive properties.

Someone then found a brief article in Electronic News for November 17, 1978: “McPhee Doping Dropped.” It explained that the Waltham, Massachusetts, firm of Silec, Inc., had abandoned the experimental McPhee technique to dope diamonds artificially with a monolayer boron coating. The McPhee process had been abandoned as too expensive and too unreliable to produce “desirable semi conducting properties.” The article concluded that “other firms have underestimated problems in boron monolayer doping; Morikawa (Tokyo) abandoned the Nagaura process in September of this year.” Working backward, the ERTS canteen fitted additional pieces of the puzzle into place.

Back in 1971, Intec, the Santa Clara microelectronics firm, had first predicted that diamond semiconductors would be important to a future generation of “super conducting” computers in the 1980s.

The first generation of electronic computers, ENIAC and UNIVAC, built in the wartime secrecy of the 1940s, employed vacuum tubes. Vacuum tubes had an average life span of twenty hours, but with thousands of glowing hot tubes in a single machine, some computers shut down every seven to twelve minutes. Vacuum-tube technology imposed a limit on the size and power of planned second-generation computers.