What Was Supposed to Happen
The plan was clean. I mean genuinely clean — the kind of plan where you read it back to yourself three times and every time it makes more sense than the last. I was going to run a full rainwater collection system off the back slope of my roof, feed it into two 275-gallon IBC totes, filter it through a three-stage setup I'd sourced piece by piece, and have potable water completely independent of the county line. No meter. No bill. No guy in a polo shirt reading a box on the side of my house every thirty days.
I'd been thinking about what real independence actually requires for a long time before this, and water was always the piece I kept circling back to. Food you can grow, power you can generate, but water is the one they really have you on. Cut the water and you're done in three days. That's not paranoia — that's biology.
I had the gutters rerouted. I had the totes sitting on cinder block platforms, leveled with a torpedo level. I had first-flush diverters installed to knock out the initial dirty runoff before it hit the storage. Sediment filter, carbon block, UV sterilizer in line. I even ran the numbers on average annual rainfall for Clairfield and calculated that two totes plus a third overflow tank would cover me for any dry stretch I was likely to see. On paper, this thing was airtight. I genuinely believed that by the following spring I would never need to think about water again.
A plug-and-play spreadsheet that tracks sales, expenses and profit for any side hustle, so you can see what you really keep.
What Actually Happened
The totes filled up fast. First good rain, maybe three weeks after I finished the install, and I was watching clean-ish water flow exactly where I told it to flow. That part felt good. That part felt like proof.
I let it run through the full filter stack and pulled a sample. Looked clear. Smelled fine. I did not send it out for a lab test because I had a UV sterilizer and I trusted the UV sterilizer. That's the first place I could point to now, but it didn't feel like a mistake at the time. It felt like confidence.
What I didn't account for — and I want to be specific here because vague post-mortems are useless — was the roof itself. My roof is old. Not ancient, but old. And somewhere in the shingle composite, something was leaching. I don't know exactly what. I know that after about six weeks of running the system and drinking maybe a third of my water from it, I started getting headaches that I couldn't place. Dull, persistent, behind the eyes. I blamed the weather. I blamed screen time. I blamed a bad run of sleep.
Then I had a buddy over — not Jeff, not anyone from outside Clairfield, just a guy I trust — and he took one look at the water in my glass and said it had a color to it. Faint. Yellow-gray. I hadn't noticed because I'd been looking at it every day and your eyes adjust to gradual changes. That's a documented thing. It's also, I'd argue, one of the more unsettling things about why you have to keep questioning even the systems you built yourself.
I sent the sample out. The results came back with elevated levels of zinc and a couple of other things I won't pretend I fully understood. Not immediately dangerous, but not nothing either. I drained both totes. Six weeks of collection, gone in an afternoon. The filter stack I'd been so proud of had done exactly what it was rated to do — and it wasn't rated for what was coming off that roof.
Where It Turned
The turning point was not the leaching roof. The roof was just a fact, a material reality sitting there waiting to matter. The turning point was the moment I decided the UV sterilizer was the finish line.
I remember the specific logic I used. I'd read enough about UV sterilization to know it handles biological contamination — bacteria, viruses, protozoa. I knew my sediment and carbon stages handled particulates and some chemical load. I looked at that three-stage stack and I thought: that's a serious system. And because it looked serious, I stopped asking what it didn't cover.
That's the part that's hard to explain to someone who hasn't done it. The system wasn't incomplete in an obvious way. It would have been perfectly adequate for municipal water running through it. The problem was source-specific, and I had evaluated the filter stack without fully evaluating the source. I'd thought about rain. I hadn't thought about what rain picks up on the way down the roof.
It looked fine right up until the glass looked yellow-gray in someone else's hand. And the reason it was invisible to me is that I had already decided it was solved. The decision to trust the UV sterilizer as a ceiling rather than a component — that's where the whole thing was already over. Everything after that was just waiting for the consequence to show up.
I've thought about this in the context of every other system I've built, because the pattern is identical every time: the failure hides inside the last thing you felt certain about. Not the thing you were worried about. The thing you stopped worrying about.
What Changes
I'm not abandoning rainwater collection. That would be the wrong takeaway and I don't do wrong takeaways just because something went sideways. What I'm doing is treating the source and the filter as two completely separate problems that each require their own verification, and neither one signs off on the other.
The roof is getting replaced — not because of this specifically, but this accelerated the timeline. In the meantime, I've added a fourth stage to the stack: an inline heavy metals filter rated for the specific compounds that showed up in my test results. I'm also sending out a water sample after every significant change to the system. New filter cartridge, new sample. Seasonal shift, new sample. It costs money and it costs time, and I used to think that was overkill. I don't anymore.
The bigger change is procedural. Before I declare any system operational, I'm now running what I call a source audit separately from a filter audit. What is the actual input — not the ideal input, the actual one? What is it carrying that I haven't tested for? Independence has a real price, and part of that price is the testing you skip when you're tired and the thing looks like it's working.
I also bought a decent handheld TDS meter and a basic water test kit that I can run myself before anything goes into a glass. Not a replacement for lab results, but a first-pass check I can do in two minutes. If the number moves in a direction I don't like, I stop and ask why before I drink anything.
The totes are full again. The system is running. I'm not going to tell you it's perfect, because that's exactly the kind of thing I should stop saying.