Read Chapter 5: A VERY AMERICAN PROBLEM

Élodie listened without interrupting while Luc explained the altered addendum.

When he finished, she asked, “Who benefits?”

“Everyone in the short term.”

“That is usually the most dangerous answer.”

He stood at the kitchen sink, staring at the dark backyard. “Asterion avoids redesign. NASA avoids delay. Managers avoid questions. The program keeps moving.”

“And in the long term?”

“Someone may die.”

She closed her book. “Then why are you asking me?”

“I am not asking whether I should act.”

“What are you asking?”

“How.”

The problem was more difficult than simple exposure. Luc had seen enough aerospace politics to know that an accusation made without proof could disappear with the accuser. He was a foreign-born engineer on a visa tied to employment. He had no union protection, no political allies and no appetite for martyrdom.

The next day he approached Harold Mercer, the NASA representative, carefully.

“I would like access to source qualification records for the DCP relay.”

Mercer frowned. “You have the contractor file.”

“I would like the source records.”

“Why?”

“To resolve uncertainty.”

“What uncertainty?”

Luc had prepared a neutral answer. “There are inconsistencies in document history.”

Mercer studied him for a long moment. “Do you think Asterion falsified something?”

Luc chose each word. “I think the evidence is incomplete.”

Mercer closed the office door.

“Moreau, I respect your work. But you need to understand where we are. Every major subsystem in this program has problems. Some are serious. Some only look serious because people stare at them too long. If we freeze hardware every time a document chain is imperfect, we will never leave Earth.”

“Then the Moon is easy. Paperwork is difficult.”

Mercer did not smile.

“I’ll request the source file,” he said. “Quietly.”

Two weeks later the answer came back: the relay vendor, MidSouth Electromechanical, had suffered a records-room fire the previous year. Original environmental qualification logs were unavailable.

Luc read the memo twice.

Reed said, “That is convenient.”

“Too convenient.”

They drove to Galveston on a Saturday and found the vibration lab closed. Reed knew a technician there, Charlie Voss, who met them at a seafood place along the seawall.

Voss remembered the Asterion campaign.

“Ugly test,” he said. “That little box hated Z-axis.”

Luc leaned forward. “What do you mean?”

“Intermittent dropout. Couple of millisecond spikes. We reran with extra fixture damping.”

“Was that in the report?”

Voss laughed. “Not the final one.”

“Why?”

“Client said fixture resonance invalidated the run.”

“Did it?”

“Maybe some of it. But the dropout tracked the unit, not the fixture.”

“Do you still have raw data?”

Voss’s expression changed. “Officially? No.”

“And unofficially?”

He looked at Reed.

“You two trying to get me fired?”

“No,” Luc said. “We are trying to determine whether a spacecraft power system has an intermittent failure.”

Voss was silent.

Finally he said, “I keep duplicate strip charts when clients argue about tests. Habit.”

“Where?”

“At home.”

Luc felt his pulse quicken.

Voss pointed a finger at him. “This is a very American problem, Frenchman.”

“What is?”

“We built a machine so big nobody wants to be the man who stops it.”

Luc looked through the restaurant window toward the Gulf.

“In France,” he said, “we would stop it just to prove we could.”

Voss laughed.

But none of them believed the joke.

Before driving to Galveston, Luc and Reed spent two evenings reconstructing the official test chronology. They used conference-room walls because no desk was large enough.

Run 41: full qualification, X-axis, pass.

Run 42: Y-axis, pass with minor fixture resonance.

Run 43: Z-axis, aborted after instrumentation issue.

Run 44: Z-axis repeat, pass.

That was the official sequence.

The maintenance log told a different story. No recorder fault was documented after Run 43. No fixture repair occurred before Run 44. But the shaker amplitude calibration changed.

“Someone reduced the drive and called it a repeat,” Reed said.

“Probably.”

“Could be legitimate if the fixture amplified response.”

“Yes.”

Reed looked at Luc. “You’re supposed to say no.”

“I am suspicious, not religious.”

They modeled the fixture resonance using the lab’s calibration sheets. Even allowing for the highest plausible amplification, the reduced Run 44 under-tested several frequency bands.

Luc wrote the result on the board.

Reed stared at it.

“So even their justification does not justify the rerun.”

“No.”

“Now you may become religious.”

That careful reconstruction mattered later. It prevented their case from depending on intuition. Luc knew that suspicion without calculation was merely style.

At Galveston, Charlie Voss would give them the missing traces. But before that, Luc already understood what he was looking for: not proof that the fixture had been perfect, but proof that the response could not honestly be called qualification.

The difference was subtle enough to survive lawyers.

It was also the difference between an argument and evidence.

Luc's first months in Houston also taught him how much of Apollo existed outside the famous centers. Small machine shops, subcontract laboratories, plastics suppliers and family-owned electronics firms fed hardware into the program from towns that rarely appeared in newspapers. Each carried a small piece of the Moon.

That distributed structure made the qualification problem harder. The relay had been manufactured in Alabama, purchased through an Asterion supplier office, integrated in Texas and tested near the Gulf. Responsibility crossed corporate lines faster than accountability.

During one procurement meeting, Luc asked whether anyone at Asterion had audited MidSouth's combined-environment test process.

A buyer answered, “They are an approved vendor.”

“That describes status, not evidence.”

“We cannot audit every supplier.”

“Then how do we know approval still means anything?”

The buyer sighed. “Moreau, the space program is not a monastery. We buy parts from industry.”

Luc almost replied that monasteries kept better records, but restrained himself.

Instead he requested the supplier-history file. It contained glowing delivery assessments, cost charts and one buried quality note describing spring-temper variability in an earlier relay family.

That note did not concern lot 62-B directly, yet it changed Luc's interpretation. The failure was not an isolated mystery. It fit a known manufacturing sensitivity.

He showed the note to Reed.

“So our little twitch has ancestors,” Reed said.

“Possibly.”

“You know, normal people would call this confirmation.”

“Normal people do not testify under oath.”

Reed rolled his eyes.

Luc's caution annoyed even his allies. But he had begun to understand that the stronger the accusation, the more carefully every supporting fact had to be separated from assumption. He wanted a case that could survive people who desperately needed it to be wrong.