Dose is the product, everything else is input
UV disinfection works by damaging microorganisms' DNA with 254 nm light as water passes the lamp. The measure of protection is dose — intensity × exposure time, in mJ/cm². The working targets: 30 mJ/cm² minimum for drinking-water duty, 40 mJ/cm² where validated performance to international protocols is required. Each pathogen has its own sensitivity, and the spread is wider than most product literature admits. At 40 mJ/cm² the bacterial and protozoan targets are comfortably covered — the EPA's validated dose table puts 4-log Cryptosporidium and Giardia, both of which chlorine struggles with, at about 22 mJ/cm² — and so are most waterborne viruses. Adenovirus is the exception, and it is a large one: it is exceptionally UV-resistant, and the EPA's 4-log virus credit is 186 mJ/cm², more than four times a typical unit's rating. So a 40 mJ/cm² system is an excellent bacterial and protozoan barrier and a good viral one. It is not a validated four-log virus barrier, and it should not be sold as one.
Because dose = intensity × time, everything that changes either input changes the protection:
Flow: the time input
Double the flow and each parcel of water spends half as long in the chamber — half the dose. A UV unit therefore has no single "capacity"; it has a dose at a flow. The same chamber might be rated 40 L/min at 40 mJ/cm² and 60 L/min at 30 mJ/cm². Quote the real peak flow honestly — the moment two outlets run together, an optimistically rated unit is delivering an under-spec dose with no outward sign whatsoever. If peak flow is genuinely uncontrollable, a flow restrictor sized to the rated flow is cheap honesty.
UVT: the transparency input
UV transmittance (UVT) is the percentage of 254 nm light surviving passage through 1 cm of the water. Clear municipal water runs 90 %+; borehole waters with organics, iron or colour can run far lower — and intensity falls off steeply with distance in low-UVT water, starving the outer edges of the chamber. Manufacturers rate systems at a stated UVT (often 95%). If your water's UVT is lower — and any coloured, iron-bearing or surface-influenced water deserves an actual measurement — the same unit delivers a materially lower dose, and the sizing must be corrected. UVT is the most ignored number in small-system UV, and the most common reason a "correctly sized" unit fails a water test.
End of lamp life: the worst-day input
Lamp output decays from the day it is switched on; a low-pressure lamp is typically specified for ~9 000 hours (a year of continuous duty) with output falling toward 80 – 85% of new by then. Honest ratings therefore quote dose at end of lamp life — the design covers the lamp's worst legal day, not its first. Two operational consequences: replace lamps on schedule, not on failure (a glowing lamp is not a performing lamp — output is invisible to the eye), and replace on the lamp maker's stated life, which for the low-pressure lamps in this duty is normally annually — that is the ~9 000 hours above, and it is stated as a calendar interval because the lamp ages in storage and in service alike. Where a manufacturer states something different for its own lamp, that figure governs. Frequent switching also ages lamps disproportionately; continuous operation is kinder than cycling on small systems.
The remaining input is cleanliness: intensity assumes light actually reaches the water, which is the quartz sleeve's job and the fouling article's subject — hard water scales sleeves, and a scaled sleeve throttles dose exactly like a tired lamp.
Rule of thumb: a UV specification is honest only when it names all three conditions — this dose, at this flow, at this UVT, at end of lamp life. If a quoted unit names only a flow, assume the missing conditions are optimistic, and check the UVT of your actual water before trusting it.