"It should," Laurel admitted.
She could have requested a stop then. Wesley later thought about that conversation more often than he thought about the failure itself, because the choice was not between having information and lacking it. Both of them had enough information to insist on clarification; neither had acted as forcefully as the situation required. Laurel returned to her office promising to put her warning in writing. Wesley maintained the run, telling himself that the automated protection system would stop the test if the assembly departed its limits.
The prototype continued into the night. Visitors had left, lights in the main assembly buildings had dimmed, and the laboratory staff had been reduced to a small crew. Wesley sat at the console with a paper cup of tea and watched a fine pattern of variation emerge in the tail-assembly data. It was subtle, a change in response rather than a dramatic alarm, but it matched the timing of the noise he had marked earlier.
At the nineteenth hour, the part failed. The fastener fractured through its load-bearing section, the bracket shifted, and the monitored tail assembly moved seven-tenths of a degree outside its calibrated position. That deviation was enough to trigger the chamber's protective shutdown. The equipment stopped in an orderly sequence, the display turned red, and a silence so sudden it felt physical settled over the room.
Wesley was on his feet before the technicians finished reading the alert. He confirmed the chamber was secure, checked that no one had been injured, and ordered the assembly left untouched until the required evidence could be collected. The damage was contained inside a structure designed to withstand exactly such events. No aircraft had crashed, nobody had been harmed, and the failure had occurred where it should have been discovered: under controlled test conditions, before qualification or deployment.
That fact brought Wesley no comfort when he reviewed the camera feed. The recording showed a small defect developing gradually in the replacement fastener, growing with the repeated loading until the remaining metal could no longer carry the force. The assembly's final movement was almost anticlimactic, just a slight shift followed by the shutdown. The meaning was not small. It was visible evidence that the unapproved configuration could not be treated as equivalent to the original.
He secured the failed component under the laboratory's evidence procedure and placed the footage under preservation hold. He had the technician who assisted him verify the label, the location, and the data file identifier. Then he made two calls: one to the engineering duty officer and one to the executive escalation number. Preston arrived before Laurel, wearing a shirt he had evidently thrown on while driving and an expression that grew colder as Wesley explained what had happened.
"Could be a fixture problem," Preston said. He was standing in front of the graph when he made the suggestion, although the fixture data showed no corresponding fault. Wesley told him the fractured fastener was the initiating event indicated by the video and sensor records. Preston asked whether the report had been finalized. When Wesley said no, Preston told him to wait until management could review the wording.