She was asleep when the swirl of iridescent liquid began to seep through Kenichi Hirai’s shroud.

It had begun with a few glistening droplets oozing through a tiny rent in the plastic, torn open when the shroud had snagged. For hours the pressure had been building, the plastic slowly inflating the contents swelled. Now the breach widened, and a shimmering ribbon streamed out.

Escaping through the pallet ventilation holes, the ribbon broke apart into blue-green droplets that briefly danced in weightless abandon before recongealing into large globules that undulated in the dimly lit cabin. The opalescent fluid continued spill forth. The globules spread, riding the gentle currents of circulating air. Drifting across the cabin, they found their way to the limp form of Jill Hewitt, who slept unaware of the shimmering cloud enveloping her, unaware of the mist she inhaled with every soft breath or of the droplets that settled like condensation on face. Only briefly did she stir, to brush the tickle on her cheek the opalescent drops slid toward her eye.

Rising with the air currents, the dancing droplets passed through the opening of the interdeck access and began to spread through the gloom of the flight deck, where three men drifted in the utter relaxation of weightless sleep.

August 8

The ominous swirl had begun to take shape over the eastern Caribbean days before. It had started as a short wave trough aloft, a gentle undulation of clouds formed from the evaporated waters of the sun-baked equatorial sea. Butting up against a bank of air from the north, the clouds had begun to rotate, spinning a calm eye of dry air. Now it was a definite spiral that seemed to grow with every new image transmitted by the geostationary GOES weather satellite. The NOM National Weather Service had been tracking it since its birth, had watched as it meandered, directionless, off the eastern end of Cuba. Now the newest buoy data was coming in, with measurements of temperature, wind speed and direction. This data reinforced what the meteorologists were now seeing on their computer screens.

It was a tropical storm. And it was moving northwest, toward the tip of Florida.

This was the sort of news shuttle flight director Randy Carpenter dreaded. They could tinker with engineering problems. They could troubleshoot multiple systems failures. But against the forces of Mother Nature, they were helpless. The primary concern of this morning’s mission management team meeting was a go-no-go decision on deorbit, and they had planned for shuttle undocking and deorbit burn in six hours’ time. The weather briefing changed everything.

“NOAA Spaceflight Meteorology Group reports the tropical storm is moving north-northwest, bearing toward the Florida Keys,” said the forecaster.

“Radar from Patrick Air Force Base Nexrad Doppler from the National Weather Service in Melbourne show radial wind velocities of up to sixty-five knots, with intensifying rain. Rawinsonde balloon and Jimsphere balloon both confirm. Also, both the Field Mill network around Canaveral as well as LDAR show increasing lightning activity. These conditions will probably continue for the next forty-eight hours.

Possibly longer.”

“In other words,” said Carpenter, “we’re not landing at Kennedy.”

“Kennedy is definitely out. At least for the next three to four days.” Carpenter sighed. “Okay, we sorta guessed that was coming. Let’s hear about Edwards.” Edwards Air Force Base, tucked into a valley east of the Sierra Nevada in California, was not their first choice. A landing at Edwards delayed shuttle processing and turnaround for the next mission because the shuttle would have to be transported back to Kennedy, piggybacked to a 747.

“Unfortunately,” said the forecaster, “there’s a problem with Edwards as well.” A knot had formed in Carpenter’s stomach. A premonition that this was the beginning of a bad chain of events. As lead shuttle flight director, he had made it his personal mission to review any mishap on record and analyze what had gone wrong. With the advantage of hindsight, he could usually trace the problem backward, through a succession of bad but seemingly innocuous decisions. Sometimes it started back at the factory with a technician, a miswired panel. Hell, even something as big and expensive as the Hubble Telescope lens had started off screwed up from the very beginning.

Now he could not shake off the feeling that he would later think back to this very meeting and ask himself, What should I have done differently?

What could I have done to prevent a catastrophe?

He asked, “What are the conditions at Edwards?”

“Currently they’re looking at a cloud ceiling at seven thousand feet.”

“That’s an automatic no-go.”

“Right. So much for sunny California. But there’s the possibility of partial clearing within the next twenty-four to thirty-six hours. We might have reasonable landing conditions if we just wait it out. Otherwise, it’s off to New Mexico we go. I just checked MIDDS, and White Sands looks good. Clear skies, head winds at five to ten knots. No adverse weather forecast.”

“So it’s down to a choice,” said Carpenter. “Wait till Edwards clears up. Or go for White Sands.” He looked around the room at the rest of the team, seeking opinions.

One of the program managers said, “They’re fine up there right now. We could leave them docked to ISS as long as we need to, until the weather cooperates. I don’t see the necessity of rushing them home to a less than optimal site.” Less than optimal was an understatement. White Sands was little more than an isolated landing strip equipped with heading alignment cylinders.

“There’s the matter of getting the corpse back as soon as possible,” said Todd Cutler. “While an autopsy’s still useful.”

“We’re all aware of that,” said the program manager. “But weigh it against the negatives. White Sands is limited. Civilian medical backup just isn’t there, if we have any problems on landing. In fact, all things considered, I’d suggest we wait it longer, till Kennedy’s clear. Logistically, it’s the best thing program. Quicker orbiter turnaround, get her right back on the pad for the next mission. In the meantime, the flight crew can stay as a hotel for the next few days.”

Several other program managers nodded. They were all taking the most conservative approach. The crew was safe where they were, the urgency of bringing home Hirai’s corpse paled in light of all the problems of a White Sands landing. Carpenter thought of the ways he could be second-guessed should there, God forbid, be a catastrophic landing at White Sands. He thought of the questions he would ask, were he reviewing the decisions of another flight director.

Why didn’t you wait out the weather? Why did you hurry them home?

The right decision was the one that minimized risk, yet met mission goals.

He decided to choose the middle ground.

“Three days is stretching it out too long,” he said. “So Kennedy’s out. Let’s go for Edwards. Maybe we’ll get clear skies tomorrow.” He looked at the forecaster. “Make those clouds go away.”

“Sure. I’ll just do a reverse rain dance.” Carpenter glanced at the wall clock. “Okay, crew’s wake-up call is in four hours. We’ll give ‘em the news then. They can’t come home quite yet.”