Priya greeted him in the reliability office with a stack of documents taller than his lunchbox. The remaining transient had appeared only under a narrow combination of operating conditions, and the original control data were not aligned with all the vibration channels. Some records had been taken in different time bases. Others had been updated after corrective changes. Their first task was not to invent a theory but to create one reliable chronology. Daniel spent two days checking timestamps, test conditions, maintenance actions, and instrument revisions. It was the kind of patient work he liked, though he knew better than to call it exciting during lunch with Sophie.
Luis found a second complication. A filtering change introduced after the third failure had altered the shape of a small event that appeared before the main vibration peak. The change was within approved settings but made direct comparisons misleading. Priya reconstructed the earlier signals using a consistent method and found that the apparent mechanical response had shifted after a control-system adjustment. The team could no longer treat all recorded events as identical recurrences. Each revision had changed the experiment in a way that mattered to the story they were trying to tell.
Arthur asked his engineers to write competing hypotheses and specify what evidence would disprove each one. A mechanical defect remained possible. A control interaction remained possible. The mounting structure's resonant behavior might amplify some conditions, and the measurement system could obscure the relationship among them. Daniel suggested that the team look at whether the engine's control transitions excited a response that the test installation amplified, rather than searching only for a damaged component that vibrated constantly. Arthur asked him what would disprove that explanation. Daniel said the transient should persist in a comparable way when the relevant control behavior changed, if his idea were wrong. Priya added several other required comparisons, and they built a formal plan.
The unexplained transient did more than delay the program. It strained trust among people who had spent years working together. A controls engineer named Peter Lang believed the revised investigation unfairly implied that his team's software had caused the mechanical failures. A test-cell engineer argued that the installation had been reviewed and approved, and he resented the suggestion that the facility was inadequate. Both men were experienced, conscientious, and exhausted by repeated criticism from departments that understood only part of their work. Arthur asked Priya to arrange a joint technical session in which every claim had to be attached to evidence rather than professional ownership.
Peter brought logs showing that the relevant control behavior remained within the limits understood at the time of the original design. He pointed out that a narrow transient could be an expected response rather than an abnormal event if considered without the test-cell amplification. The test-cell engineer, Robert Flynn, showed that the mounting structure met its own approved specifications and had not suffered a mechanical failure. The disagreement made the room tense until Luis observed that both statements could be true while the combination still produced an unacceptable measured response. Individual elements could satisfy their requirements and still interact in ways the integrated test had not fully predicted.