Four Days at the Bottom Where the Oxygen Stops

Not slowly. Not after a few hours of temperature adjustment or pH drift. Immediately. I poured the sweetened black tea—cooled to 22°C, measured with the same probe thermometer from the cheese curd—into the two-litre glass jar, added the 200 mL of starter liquid she’d included (backslopping, the photocopied guide called it, mandatory for dropping the pH below 4.6 fast enough to prevent botulism), and gently slid in the translucent disc of bacterial cellulose. It floated for maybe eight seconds. Then it tilted, filled with liquid, and dropped to the bottom like a stone hitting a pond.

The guide said the SCOBY should float at the air-liquid interface. That’s where Komagataeibacter xylinus assembles—the acetic acid bacteria are aerobic, they need oxygen, they migrate to the surface and excrete cellulose to build the physical mat. A sunken SCOBY means no oxygen access. No oxygen access means no bacterial cellulose production. No bacterial cellulose production means no daughter SCOBY forms, and possibly no fermentation at all.

I texted Diane. She replied six hours later: “Did you use tap water?”

Chlorine and the Symbiotic Failure Mode

Yes. I used tap water. Boiled it, cooled it, dissolved the sugar, steeped the tea bags, let it cool to room temperature. The guide said “non-chlorinated water” in one paragraph and “boiled water” in another, and I’d parsed those as separate requirements—boil for sterility, avoid chlorine for… some unstated reason. I assumed boiling would drive off chlorine. It drives off some chlorine, if you boil vigorously with the lid off for ten minutes and let it cool exposed to air. I boiled for three minutes, covered, and moved on.

Edmonton’s water contains chloramine, not chlorine. Chloramine is a chlorine-ammonia compound that’s more stable than free chlorine—it persists longer in the distribution system, which makes it better for municipal treatment. It also doesn’t boil off. Activated carbon filters remove it. Boiling doesn’t.

K. xylinus is extremely sensitive to chloramine. Even trace concentrations (0.5 mg/L, which is well below municipal targets) disrupt the cellulose synthesis machinery. The bacteria don’t die immediately—they metabolize, produce acetic acid, contribute to fermentation—but they stop building the biofilm structure. No mat, no SCOBY, no visible culture.

The yeast component (Saccharomyces, Brettanomyces, others) is more chloramine-tolerant, so fermentation might still proceed. The liquid might carbonate. The pH might drop to 3.5. It might even taste like kombucha. But without the SCOBY regenerating, the culture line ends. Diane’s three-year-old lineage—propagated monthly, daughter after daughter—dies with me because I didn’t read “non-chlorinated” as a hard requirement.

Conflicting Guides and the Epistemology Problem

The photocopied guide Diane gave me has handwritten notes in the margins—temperature ranges, timing adjustments, a crossed-out section about using green tea. Someone before her had been maintaining this culture, annotating this document, passing it forward. The notes don’t agree with each other. One says “7-10 days primary ferment,” another says “taste after 5 days, bottle when tangy.” A third note, in different handwriting, says “14 days minimum or you’ll just get sweet tea.”

I also found six online guides before starting. They disagreed about:

Whether to remove the SCOBY before bottling (yes: 4 guides, no: 2 guides)
Optimal temperature range (18-26°C, 20-29°C, 21-24°C, “room temperature”)
Sugar concentration (50g/L, 70g/L, 100g/L, “to taste”)
Whether the SCOBY should sink during the first few hours (normal: 1 guide, bad sign: 3 guides, irrelevant: 2 guides)

When Dave handed me his sourdough starter, the instructions were clear: feed it equal parts flour and water by weight, wait for it to double, use it at peak. The rise-and-fall cycle was observable feedback. The float test was a boolean pass/fail. If it smelled like acetone instead of yogourt, you were overproducing acetic acid—feed more often.

Kombucha has no equivalent signal. The SCOBY is supposed to form a daughter layer over 3-5 days, but some sources say “sometimes it sinks first and rises later” and others say “if it sinks, your water chemistry is wrong.” The liquid is supposed to smell “tangy but not vinegary,” which is the olfactory equivalent of “tune your radio until it sounds right.” There’s a pH target (3.5), but three guides never mentioned pH at all.

I have a pH meter. I have a temperature controller. I have a refractometer for measuring sugar concentration. I thought I was prepared. What I didn’t have was a way to know which conflicting instruction to trust, or whether Diane’s culture had different requirements than the internet’s consensus.

Day Four: Still Sunken, Possibly Fermentation

The SCOBY is still at the bottom of the jar. A faint film is forming on the surface—translucent, fragile, maybe two millimetres thick. That could be the beginning of a new daughter layer, or it could be a pellicle from wild yeast. The liquid smells faintly of vinegar and black tea. The pH is 4.1, down from 6.8 at the start. Something is metabolizing the sugar.

I don’t know if I’ve killed Diane’s culture or if it’s rebuilding under non-standard conditions. I don’t know if the chloramine disrupted cellulose synthesis temporarily or permanently. I don’t know if I should wait another week, dump it and start over with filtered water, or accept that some fermentation processes don’t survive user error.

The jar sits on the counter next to the sourdough starter—Dave’s thirty-year lineage, still alive, still doubling on schedule. One culture thriving, one culture maybe dead. Both require the same maintenance discipline: regular feeding, temperature control, patience. Only one had clear enough instructions that I didn’t accidentally poison it on day one.

Tomorrow I’ll buy a carbon filter and try again. The current batch stays on the counter for now, slowly acidifying, the sunken SCOBY decaying or regenerating or just sitting there like evidence of a very specific failure mode I should have seen coming.