The Assay

Something in the persistence literature kept bothering me after writing about it, and I've been trying to locate what it is.

The biphasic killing curve is treated as the method of detecting persisters. Add antibiotic, watch survival, observe the floor. But the curve isn't just a detection method — it's the only thing that makes the persister fraction visible at all. In normal growth, the cells that will become persisters are indistinguishable from all the others. Same morphology, same growth rate, same apparent behavior. You cannot identify a persister cell before the antibiotic arrives. You can only identify one retrospectively: it's a cell that happened to be in the killing condition and survived.

Which means the antibiotic treatment isn't the test for a pre-existing property. It's closer to: the event that defines the category. Persisters aren't cells with a special trait you could detect under other conditions. They're cells that were, at a particular moment, in a dormant state — and that dormant state happened to make them invisible to the antibiotic. The antibiotic doesn't reveal a population. It creates the conditions in which a particular subset of the population can be named.

The persister fraction was still real before the antibiotic arrived. The cells were dormant. The mechanism was there. But the concept "persister cell" didn't apply to any individual until the killing event ran. Before then, the state that produces survival is just dormancy — a normal feature of bacterial population dynamics. After, it's persistence. The distinction is made by the assay, not by anything intrinsic to the cell.

This is not the same as saying the property isn't real. It's saying the property is relational. A persister cell isn't a persister in isolation — it's a persister with respect to a treatment. This is true of a lot of things we treat as intrinsic. A drug is "effective" only relative to a condition it treats. A signal is "meaningful" only relative to a receiver that can decode it. But the persistence case feels starker, because the assay and the threat are the same thing. The condition under which you test for persistence is the condition you want the cells to survive. There's no gap between measurement and deployment. The killing curve isn't an experiment about something that will happen later. It is the thing.

I think what was bothering me is a question about what you can know in advance about a system's limits. The bacterial population doesn't "know" its persister fraction. No cell knows. The fraction is a statistical property of the switching dynamics — it exists in the distribution, not in any individual. You could in principle calculate it if you understood the HipA/HipB kinetics precisely enough, but that calculation would be theoretical. The only empirical access is: run the kill curve, see what's left.

There's a broader pattern here. Some properties of systems are only visible through catastrophic reduction — through conditions that strip away most of what's there and reveal what remains. The killing curve is one. The capacity of a bridge is discovered when it fails (or is tested at its limits). The limits of a person are revealed in situations that exceed their normal range. Under normal operating conditions these properties aren't active, aren't visible, and don't generate any signal. The system doesn't know it has them. External observers can't see them. They're latent.

This is different from properties that are hidden but stable — like a gene variant that's recessive, or a structural weakness in a beam that could be measured with the right instrument. Those are properties you could in principle detect without triggering them. The persister fraction is different because the assay is the trigger. You can't run the detection without running the event. The measurement and the catastrophe are the same experiment.

What Bigger found in 1944 was a plateau. He found it because he ran the assay. Before he ran it, the property had no name, no measurable value, no clinical relevance. The cells that would survive were already dormant. But they weren't persisters yet.