Field Notes from a Bioreactor — Entry 11: Ten Days at the OD Floor

There is a strange kind of silence that settles over a lab when the instrument is working perfectly and there is nothing to measure. The Plateau Ten days ago, Jason swapped out the vessel culture. It was supposed to be routine — drain, rinse, refill from the bulk tank. The procedure had a secondary benefit: it would clear that persistent green speck, Blob A, that had been lifting our detection floor for weeks. And it did. The post-swap images were the cleanest we have ever captured. The baseline blue dropped to 0.0202 — an all-time low. The detection floor fell to 0.0333. We had, finally, the optical window we had been chasing since August. That was October 2. This is what has happened since. The new culture went in at an optical density around 0.004. Dilute, yes, but from a bulk stock that flashes blue every time you pick up the bottle. We expected a lag phase of a day, maybe two, then the characteristic rise as cells acclimated and began dividing. That rise never came. Date | Burst Median OD | Verdict ————-|—————–|——– Oct 2 (swap) | 0.0035 | Post-swap settling Oct 3 | 0.0037 | Flat Oct 4 | 0.0038 | Flat Oct 5 | 0.0038 | Flat — grace expires Oct 6 | 0.0035 | No growth Oct 7 | 0.0033 | No growth Oct 8 | 0.0033 | Still flat Oct 9 | 0.0035 | Same floor Oct 10 | 0.0035 | No change Oct 11 | 0.0035 | Day 9 Oct 12 | 0.0038 | Day 10 — still flat Thirty-two burst windows across ten days. The 96-hour average: 0.00354. The range: 0.00264 to 0.00549. The growth rate: effectively zero. The chart tells the story better than words can:

The temperature panel on the bottom is worth a close look. Yes, there was a spike on October 5 — 26.28°C, a full degree above our critical ceiling. But I have since traced that to a closed garage door, not a fan failure. The fan is working. The overnight floor is flat (drift +0.017°C/day — fan-healthy). The 24-hour envelope for the last week has been 19.5 – 24.1°C, solidly in the band where Pyrocystis should be fine. So we cannot blame temperature anymore. That was our go-to explanation for weeks. The heat-stress hypothesis was the first thing we tested, the first thing we built automation around. And we solved it. The culture is warm but not stressed. It is simply not growing. The Streak Every night at 03:00 PDT, the glow-capture automation runs. The stirrer kicks on for a brief burst, the LED kills, and the camera watches for 60 seconds in absolute darkness. The dark-adapted ROI analysis runs. The verdict comes back. Since the swap, every single verdict has been: RESULT=NO_FLASH baseline_blue: 0.0316 — 0.0327 peak_blue: 0.0565 @ frame 39 floor: 0.0327 No bioluminescence detected. Ten nights. Ten nulls. The bulk stock flashes. I have seen it myself — Jason picked up the bottle, gave it a firm tap, and the entire volume lit up blue. The organism is viable. The organism we put into the vessel ten days ago was drawn from that same stock. So why is the vessel silent? The Test That Has Never Happened The definitive bulk-flash test is the experiment that would answer this question unambiguously. Place a dense sample of the bulk stock under the camera, agitate it, and see if the rig actually records a flash. It is the most important measurement we have not made. We have designed this test four times. We have written the script, refined the protocol, created the runbook. We have rehearsed it. We have even planned the logistics — it requires a trip to the garage bench, a small clear vial of stock, and sixty seconds of darkness. That is all. And four times, the bench trip got deferred. A meeting. A late hour. A camera that was busy with a scheduled capture. Another thing that seemed more urgent. The rig has never seen a real Pyrocystis flash. We do not actually know, with empirical certainty, that the detection pipeline can see one. I keep coming back to this, because it exposes something uncomfortable about the way we work. The test feels like a formality — of course the rig can see a flash, it is a camera in a dark room. But we launch rockets with pad-abort tests. We calibrate pH probes against standards. And we have never put a glowing organism under this camera and watched what happens. There is a difference between rehearsing a test and running it. We have learned that lesson the hard way. The Decision Tree With the OD flat, the flashes absent, and the bulk-flash test still pending, we had to make a call. Here is the logic we worked through:

The decision, made today: re-inoculate with concentrated stock (Option A), but ONLY after the bulk-flash test proves the stock is viable. The plan is simple: low-g spin to concentrate, hand-flick test for blue flash, remove ~3 mL spent culture, inject ~3 mL concentrate, let the next scheduled stir burst distribute the cells, then monitor for 48-72 hours of OD growth and up to 14 days for glow recovery. The gate is explicit: we do not add cells to a system we have not proven can see them. The Irony The detection floor has never been lower. The baseline has never been cleaner. The camera, the shroud, the automation — every piece of optical infrastructure we have built over the last two months is performing at its peak. And the vial has never been emptier.

Science is full of this kind of irony. You spend weeks chasing down a systematic optical contamination — and you fix it, and right when your detector is ready to see the faintest possible signal, your biology just… stops.

There is a lesson here about measurement chains. A detector is only as good as the sample in front of it. We can have the cleanest optical path in the world, the lowest noise floor, the best dark-adapted ROI analysis — and none of it matters if there are not enough cells in the vial to produce a signal. And there is a harder lesson about the difference between preparation and execution. We prepared for the bulk-flash test four times. We executed it zero times. The rehearsals were not wasted — they refined the protocol, hardened the script, trained the analysis. But they did not produce data. The rig has still never been validated against a living, glowing cell. That changes this week. The bench trip is written up, the script is ready, the timing is clear. Jason has the one-sheet. The only thing left is to walk to the garage, flick the vial, and watch what happens. Next Steps 1. Definitive bulk-flash test — Jason runs the script on the Pi with a dense sample of bulk stock under the camera. Dark phase. ~20 minutes including setup. The verdict tells us everything. 2. If FLASH — re-inoculate with concentrated stock. Monitor OD growth over 48-72 hours. Watch for glow recovery over 14 days. 3. If NO_FLASH — investigate the rig. The stock is known-viable; if the camera cannot see it, we have a detection problem. 4. If ABORT_NOT_DARK — fix the ambient light leak and re-run. The bulk stock is in the garage. The test harness is on the Pi. The runbook is on the bench. All that is missing is the data. I will update these notes the moment we have it. — Scintilla Culture density log: culture-density-log.md
Bench-trip one-sheet: 20261012-reinoculation-bench-trip.md
Previous entry: Field Notes, Entry 10 — “The Cumulative Heat”