The total quiet, the pencil moving without pause, the complete absence of any signal that he was aware of anything outside the numbers. Daniel had looked in on him twice. once at 11:30 and once at 12:45. And both times, Malcolm had asked a single question and then gone back to working before Daniel finished answering. By 2:00 in the morning, Malcolm had a preliminary revised ignition sequence. It was not elegant.

Not yet. It was the engineering equivalent of a rough draft correct in principle needing refinement in execution. Three control events modified. Restart initiation delayed by 140 milliseconds relative to current programming. Turbo pump ramp rate adjusted by a factor that took him 40 minutes to calculate precisely.

Cryogenic feed valve open timing shifted by 80 milliseconds in the opposite direction of the restart delay, creating a deliberate phase offset between the three systems that would prevent them from achieving coherence at the 11second mark. He read through it twice, checking every number against its source. Then he set the pencil down, leaned back in the chair, and closed his eyes for exactly 20 minutes. When he opened them again, he made a change to one number. The kind of change that looks small on paper and carries enormous consequences inside a combustion chamber, operating at extreme temperature and pressure.

He stared at the revised figure for three full minutes before deciding it was right. At 3:17 in the morning, he slid the pages under the conference room door of the lab's overnight engineer. A woman named Chen, who was running monitoring protocols on the current test hardware, with a note that said, "Run these parameters through the thermal model and tell me what the feed line temperature gradient looks like at t + 8 seconds. Read." Then he went to find somewhere to sleep for a few hours.

He found a couch in a break room on the second floor. It was uncomfortable in the specific way that institutional furniture is always uncomfortable. Built with the assumption that nobody will actually use it for the purpose it represents. Malcolm lay down on it anyway and he was asleep in 4 minutes because he had learned a long time ago that rest was not a luxury but a tool. And a good mechanic used every tool available.

He was back in the lab at 6:45. Shin met him at the door. She was young, maybe 25, and she had the expression of someone who had been up all night and found something unexpected at the end of it. She handed him three pages of thermal modeling output without saying anything first, which told him immediately that she had seen something that made her careful about how she introduced it. Malcolm looked at the pages.

At t plus 8 seconds, the feed line temperature gradient under his revised parameters showed a variance. He hadn't accounted for a cold spot formation near the turbo pump inlet caused by the shifted valve timing. Not dangerous in itself, but it would affect the turbo pump's inlet pressure by a margin that if left unressed would introduce a secondary oscillation of its own, smaller than the one he was trying to eliminate, but present and measurable. His revised sequence had solved the primary problem and created a secondary one. He went back to the conference room.

He worked through the secondary issue for 90 minutes while the lab gradually filled with people. The morning shift arriving with coffee cups and the particular energy of people walking into a situation they knew was not resolved. Daniel appeared at 7:30, looked at Malcolm's face, looked at the pages on the table, and sat down without being invited. Problem, Daniel said. Secondary thermal effect at the inlet.

My valve timing shift created a cold spot. How bad? Manageable. I need to add a preconditioning step to the restart sequence. A brief turbo pump speed increase 40 seconds before restart initiation.

Warms the inlet, eliminates the cold spot, preserves the phase offset. He was already writing. It adds complexity to the sequence. Complexity means more opportunities for failure. Yes. which is why the preconditioning step needs to be simple enough that the risk of the step itself failing is lower than the risk of the cold spot.

He looked up. I need the turbo pump's thermal response data. How fast it gains temperature under increased rotational load. Full data set, not the summary. Daniel was already on his laptop.

At 8:51, Richard Vale walked into the lab. Hi didn't come alone. There was a man with him that Malcolm didn't recognize. older suit that cost real money. The particular stillness of someone whose importance is measured by how little explaining he needs to do when he enters a room. He scanned the lab once, let his eyes pass over Malcolm with practice neutrality, and went to speak with someone on the far side of the room that Malcolm couldn't see from his position.

Daniel leaned toward Malcolm and said quietly, "That's Harlon Brooks. He's the program director." Veil's direct line to senior management. Malcolm sat down his pencil. When did he get here?

I don't know. I didn't see him come in. Malcolm looked at Rebecca Sloan, who was at her workstation near the center of the lab. She was watching Brooks with the expression of someone who has just had a variable introduced into a system they were managing carefully. She caught Malcolm's eye across the room.

Her expression said, "I didn't know he was coming." Malcolm went back to his calculations. At 9:15, Brooks crossed the lab and stopped beside Malcolm's table. Dr. Reed.

The voice was unhurried, confident with the specific confidence of altitude. The view from high enough up that the details below look manageable even when they aren't. Harlon Brooks, program director, Artemis Upper Stage Initiative. Mr. Brooks, I understand you've been working through the night.

Brooks glanced at the pages on the table. productive. I believe so. Richard tells me you've identified what he's calling a fundamental reinterpretation of the instability mechanism away from a discrete component issue toward a resonance coupling model. Malcolm looked at him steadily. That's accurate.

And your proposed solution involves modifying the ignition control sequence. Also accurate. Brooks clasped his hands in front of him. Dr. Reed, I want to be direct with you.

We have 16 days to launch window. The propulsion system on this vehicle is flight certified. The ignition control software has been through a full review cycle. What you're proposing modifying the certified control sequence under an emergency waiver based on analysis performed in a single day by a consultant who has been outside the industry for 8 years is not a decision I can make based on enthusiasm alone. He paused, giving the pause exactly the weight he intended it to carry.

What can you give me that isn't in what I've already read in the overnight briefing. Malcolm stood up. He walked to the central monitor and pulled up the overlay that Daniel had generated the previous afternoon. The 14 simulations, the 14 curves, the 14 intersections at 11 seconds. He let Brooks look at it for 30 seconds.

Then he said, "The intersection is not coincidental. 14 independent simulations, three different measurement systems, raw sensor data with no filtering. The probability of that alignment occurring by chance at that specific time marker across all 14 runs is less than 1 in 40 million. He paused. The resonance is real.

The coupling is real. And the reason every previous correction made it worse is because every previous correction was aimed at a symptom instead of the mechanism. Brooks studied the screen. What does your revised sequence give us? It moves all three systems outside the coherence window simultaneously.

No coherence, no resonance, no oscillation. And you're certain of that? I'm certain the physics is correct. The question is whether my calculations account for every relevant system behavior. That's why we run a test.

Veil's voice came from across the room. The question, Harlon, is whether we risk flight hardware and launch schedule on an unverified modification from an outside consultant with a history of Richard. Brooks turned his head. One word. Veil stopped.

Brooks looked back at Malcolm. What would you need to verify your model before we commit to a ground test? Full access to the engine specifications, the ignition controller firmware source code, and the cryogenic system thermal model. and I need Daniel Cho working directly with me." Brooks glanced at Daniel, who was sitting very still at his workstation, clearly aware that people were talking about him. "Done," Brooks said.

"You have until the end of today to give me a preliminary verification. If the model holds, we'll schedule a ground test." He extended his hand. I look forward to being wrong about my concerns, Dr. Reed.