"Something wrong?"
"I hope not."
"Kenneth."
I looked back. "Three houses reporting odd water. I'm checking the main."
Her expression tightened. She knew enough to understand that three houses meant more than one plumbing problem. "Call me when you know."
The Hendersons let me sample from an outdoor spigot downstream of their house shutoff. The water looked faintly brown against the white collection tray. It smelled vegetal, almost sweet, with a second odor beneath it I didn't like. I checked another point on Sycamore and then a sample location nearer the clubhouse. The results weren't consistent in every detail, but they told the same broad story: something was wrong beyond the controlled equipment at the pump house.
I spent the next hours checking every obvious cause. I inspected the chlorine feed, the solution tank, the injection fitting, the pressure recorder, and the upstream line. I examined the last few days of logs. I tested a fresh reagent against a known standard. I checked for an instrument failure. I could see chlorine where it was being introduced, yet the protection that should have remained in the distribution water was vanishing.
That distinction matters. Water can leave a treatment point in good condition and lose its protection if it encounters contamination somewhere else. A disinfectant residual is not a magical guarantee that everything is safe, but it is one of the early warnings an operator watches. If the number suddenly disappears where it should not, you investigate as though the warning is real until you can establish otherwise.
At 4:12 that afternoon I held a clear sample vial up to the light in the pump house. I had just run a field test from a downstream monitoring tap that had always given dependable results. The reagent should have produced at least a trace of color. It didn't.
Zero point zero.
I rinsed the vial, opened a fresh reagent packet, and repeated the test. I thought about every previous summer when we'd had unusually high demand, every small leak and maintenance outage, every storm that had interrupted power. In nineteen years, I had never measured a complete loss of residual at that point under normal operating conditions.
Zero point zero again.
I checked the chart recorder. Around the time of the first complaints, there had been a pressure disturbance followed by a change in the pattern of distribution flow. It was not proof of a particular source, but combined with the odors and the tests, it was enough to make a cross-connection or intrusion a serious concern. There was no sensible reason to continue sending that water into occupied houses while I tried to identify the cause.
I called the county's general water office and reached an automated menu. It gave me a number for after-hours emergencies. I called and described the odor reports, the loss of residual, and my intention to isolate the distribution side. The duty receptionist took my name and said she would route the message. I wrote the time in my notebook. What I failed to understand then was that leaving a message at a county line did not satisfy every formal reporting obligation to the state. It would become one of the most expensive mistakes of my career.