At that instant the entire Newton spacecraft lurched sideways. Nicole was thrown backward, against the wall. Janos fell forward, smacking his head against the operating table. His outstretched fingers landed on the control box and then slowly released as he slumped to the floor. General O’Toole and Franceses were both thrown against the far wall. A beep, beep from one of the inserted Hakamatsu probes indicated that someone in the room was in serious trouble physically. Nicole checked briefly to see that O’Toole and Sabatini were all right and then struggled against the continuing torque to regain her position next to the operating table. With great effort she pulled herself across the room on the floor, using the anchored legs of the table. When she was beside the table she steadied herself, still holding on to the legs, and stood up.

Blood spattered Nicole as her head crossed the plane of the operating table. She stared with disbelief at Borzov’s body. The entire incision was full of blood and RoSur’s scalpel!hand was buried inside, apparently still cutting away. It was Borzov’s probe set that was going beep, beep, despite the fact that Nicole had inserted, by command, significantly wider emergency values just before the operation.

A wave of fear and nausea swept through Nicole as she realized that the robot had not aborted its surgical activities. Holding on tight against the powerful force trying to push her against the wall again, she somehow man­aged to reach over to the control box and switch off the power. The scalpel withdrew from the pool of blood and restowed itself against a stanchion, Nicole then tried to stop the massive hemorrhaging.

Thirty seconds later the unexplained force vanished as suddenly as it had appeared. General O’Toole clambered to his feet and came over beside the now desperate Nicole. The scalpel had done too much damage. The com­mander was bleeding to death before her eyes. “Oh, no. Oh, God!” O’Toole said as he surveyed the wreckage of his friend’s body. The insistent beep, beep continued. Now the life system alarms around the table sounded as well. Francesca recovered in time to record the final ten seconds of Valeriy Borzov’s life.

It was a very long night for the entire Newton crew. In the two hours immediately after the operation, Rama went through a sequence of three more maneuvers, each, like the first one, lasting one or two minutes. The Earth eventually confirmed that the combined maneuvers had changed the attitude, spin rate, and trajectory of the alien spaceship. Nobody could ascer­tain the exact purpose of the set of maneuvers; they were just “orientation changes,” according to the Earth scientists, that had altered the inclination and line of apsides of the Rama orbit. However, the energy of the trajectory had not been changed significantly — Rama was still on a hyperbolic escape path with respect to the Sun.

Everyone onboard the Newton and on Earth was stunned by the sudden death of General Borzov. He was eulogized by the press of all nations and his many accomplishments were lauded by his peers and associates. His death was reported as an accident, attributed to the untimely motion of the Rama spacecraft that had taken place during the middle of a routine appendec­tomy. But within eight hours after his death, knowledgeable people every­where were asking tough questions. Why had the Rama spacecraft moved at exactly that time? Why had RoSur’s fault protection system failed to stop the operation? Why were the human medical officers presiding over the procedure not able to switch off the power before it was too late?

Nicole des Jardins was asking herself the same questions. She had already completed the documents required when a death occurs in space and had sealed Borzov’s body in the vacuum coffin at the back of the military ship’s huge supply depot. She had quickly prepared and filed her report on the incident; O’Toole, Sabatini, and Tabori had all done the same. There was only one significant omission in the reports. Janos failed to mention that he had reached for the control box during the Raman maneuver. At the time Nicole did not think his omission was important.

The required teleconferences with ISA officials were extremely painful. Nicole was the person who bore the brunt of all the inane and repetitious questioning. She had to reach deep inside herself for extra reserves to keep from losing her temper several times. Nicole had expected that Francesca might hint at incompetence on the part of the Newton medical staff in her teleconference, but the Italian journalist was evenhanded and fair in her reportage.

After a short interview with Francesca, in which Nicole discussed how horrified she had been at the moment she had first seen Borzov’s incision filled with blood, the life science officer retired to her room, ostensibly to rest and!or sleep. But Nicole did not allow herself the luxury of resting. Over and over she reviewed the critical seconds of the operation. Could she have done anything to change the outcome? What could possibly explain RoSur’s fail­ure to stop itself automatically?

In Nicole’s mind there was little or no probability that RoSur’s fault protection algorithms had a design flaw; they wouldn’t have passed all the rigorous prelaunch testing if they contained errors. So somewhere there must have been a human error, either negligence (had she and Janos, in their haste, forgotten to initialize some key fault protection parameter?) or an accident during those chaotic seconds following the unexpected torque. Her fruitless searching for an explanation and her almost total fatigue made her extremely depressed when she finally fell asleep. To her, one part of the equation was very clear. A man had died and she had been responsible.

18

POSTMORTEM

As expected, the day after General Borzov’s death was full of turmoil. The ISA investigation into the incident expanded and most of the cosmo­nauts were subjected to another long cross-examination. Nicole was interro­gated about her sobriety at the time of the operation. Some of the questions were ugly and Nicole, who was trying to husband her energy for her own investigation of the events surrounding the tragedy, lost her patience twice with the interrogators.

“Look,” she exclaimed at one point, “I have now explained four times that I had two glasses of wine and one glass of vodka three hours before the operation. I have admitted that I would not have drunk any alcohol prior to surgery, !! I had known that I was going to operate. I have even acknowl­edged, in retrospect, that perhaps one of the two life science officers should have remained completely sober, But that’s all hindsight. I repeat what I said earlier. Neither my judgment nor my physical abilities was in any way im­paired by alcohol at the time of the operation,”

Back in her room, Nicole focused her attention on the issue of why the robot surgeon proceeded with the operation when its own internal fault protection should have aborted all activities. Based on the RoSur User’s Guide, it was evident that at least two separate sensor systems should have sent error messages to the central processor in the robot surgeon. The acoel-erometer package should have informed the processor that tbe environmen­tal conditions were outside acceptable limits because of the untoward lateral force. And the stereo cameras should have transmitted a message indicating that the observed images were at variance with the predicted images. But for some reason neither sensor set was successful in interrupting the ongoing operation. What had happened?

It took Nicole almost 6ve hours to rule out the possibility of a major error, either software or hardware, in the RoSur system itself. She verified that the loaded software and data base had been correct by doing a code comparison with tbe benchmark standard version of the software tested extensively dur­ing prelaunch. She also isolated the stereo imaging and accelerometer telem­etry from tbe few seconds right after the spacecraft lurched. These data were properly transmitted to the central processor and should have resulted in an aborted sequence. But they didn’t. Why not? The only possible explanation was that the software had been changed by manual command between the time of loading and the performance of the appendectomy.