A station reported pH around 9.8. That number was unusual enough to demand attention, but not enough to explain itself.
The easiest story was a faulty sonde. A replacement sonde was installed and behaved perfectly while the technician was present. Then the technician drove away — and the abnormal behaviour returned.
An anomalous reading is an observation, not a diagnosis.
The measurement path became the investigation
Once the replacement instrument did not end the problem, the diagnostic boundary had to widen: sensor, cable, connectors, logger configuration, communications and the receiving display. Each layer could preserve, distort or misrepresent a measurement.
The important professional lesson is not that every abnormal pH value has the same cause. It is that replacing the first suspected component does not prove a mechanism. Correlation is a clue; it is not causation.
What the evidence supported
The field evidence supported a system-level investigation and challenged the initial sensor-only diagnosis. It did not justify rewriting an unusual value into a convenient environmental explanation.
That distinction matters because environmental evidence becomes useful only when the path between environment and displayed number is understood well enough to trust.