Just ran it again with MMO anode (iridium-tantalum titanium) and the cell voltage is around 0.3V lower 
With minimal media (high iron), half filled, with MMO anode:
| Intensity | Current | Earlier electrodes (cell V) | MMO anode (cell V) | MMO lower by |
|---|---|---|---|---|
| 0% | 0mA | 0.03V | 0V | — |
| 3% | 5.7mA | 2.85V | 2.58V | 0.27V |
| 5% | 7.8mA | 2.98V | 2.70V | 0.28V |
| 10% | 13mA | 3.25V | 2.94V | 0.31V |
| 15% | 18mA | 3.44V | 3.11V | 0.33V |
| 20% | 23.3mA | 3.60V | 3.30V | 0.30V |
| 25% | 28.5mA | 3.78V | 3.43V | 0.35V |
Here is the case for switching from platinised to MMO (iridium-tantalum-coated titanium) anodes. IrO₂ is a much better OER catalyst at neutral pH, so it reaches the same current at a lower voltage - i.e. less overpotential, less ROS. My measurements bear this out: across matched runs, MMO sat ~0.3 V below the previous electrodes at every current.
One caveat on thresholds: I’ve been using ~2.5 V cell voltage as a rough “stay below this” figure based on our own “no growth above 2%” observation, but I want to be clear that’s a working hypothesis for our setup, not a literature constant. The MMO swap clearly reduces overpotential and therefore ROS; whether it’s enough to get us below the actual inhibition threshold is something we’ll only confirm with culture data.