Isaac looked at her.
"How long?"
"Three weeks, maybe."
"Then OBSIDIAN-7 doesn't have a legal problem. It has a calendar problem."
"I know."
For several seconds, only the radio and a cooling fan made sound.
Then Isaac stood, crossed to an old laptop in the corner, and powered it on.
The machine took nearly a minute to boot.
"What is that?" Eleanor asked.
"The computer I left Houston with."
He opened an archived directory dated February and March 2017.
Most files were personal notes, training documents, and permitted copies of professional records. One draft email sat in an unsent folder. Attached to it was a PDF and a native analysis file.
"I wrote myself a professional-record note before termination," Isaac said. "Never sent it because I decided the settlement wasn't worth fighting. The attachment is the final report as it existed before Cameron's methodology revisions."
Eleanor leaned closer.
"Can you legally possess that?"
"I don't know. That's your lawyer's job. Look at the metadata first."
The document history showed creation time, revision events, and internal access records embedded when the file had been exported from Nexora's approved system.
Cameron Ashford had opened the full report four times in the week before Isaac's termination.
Another access entry belonged to Giselle Hartmann's executive account.
Eleanor's face became very still.
"She read it."
"Her account opened it."
"You really won't give me an inch, will you?"
"Not on evidence."
Isaac attached the file to a message addressed to Eleanor's secure intake account.
In the subject line he wrote: Historical record relevant to current proprietary-use dispute.
In the message body, one sentence:
The nondisclosure agreement may control what I can say, but it cannot change when they accessed what I wrote.
He sent it.
Eleanor forwarded it to counsel without adding a conclusion.
By midnight, the legal problem had changed shape.
Nexora could still argue ownership.
It could no longer comfortably argue ignorance.
The emergency technical review convened the next morning at nine.
The call connected Washington, Houston, a propulsion contractor in Colorado, a flight-dynamics center in Maryland, and the improvised workstation inside Merritt Repair. The video wall showed more than two dozen faces arranged in rectangles. Some participants wore suits. Most wore badges. Isaac wore the same gray work shirt he had used the day before, a dark stain near one cuff from Raymond's truck.
Cameron joined from Washington with two Nexora engineers and the company's counsel seated just outside camera range.
He spoke first.
His presentation was polished, efficient, and built around confidence. The Nexora software workaround would reduce the excitation rate during the high-risk thermal transition, decrease measured vibration, and preserve operational control long enough to complete a revised mission profile.
The slide deck showed a twenty-five-day implementation and validation timeline.
Cameron used the word validated four times.
He used unvalidated four times when referring to Isaac.
"We need to distinguish between a compelling hypothesis and a qualified remediation," Cameron said. "Mr. Merritt's analysis is interesting, but it comes from an individual who has not worked inside the current certification environment for six years. Nexora's proposal is based on the approved system architecture and the existing software authority."
Isaac watched without expression.
Eleanor waited until Cameron finished.
"Isaac, take us through your model."
Cameron interrupted immediately.
"Before he does, I'd like the record to reflect that Nexora maintains its objection to any use of proprietary analysis developed during prior employment."
A government attorney answered from another window.
"That objection is preserved. It does not stop the technical review."
Cameron stopped.
Isaac had no slides.
Adrian shared three scanned sheets of handwritten calculations, cleaned only enough for the camera to read them.
"The problem isn't a temperature limit," Isaac began. "And it isn't a vibration limit. It's the temporary overlap between the two."
He pointed to the model. "The thermal transition changes housing stiffness. That moves the mechanical response into a resonance band during a short window each orbit. The ninety-three-minute pressure oscillation is the primary signature. The thirty-one-minute feature is the coupled harmonic."