The unit that governs every lighting decision on earth was fixed by a small number of people sitting in darkened rooms, turning a dial until a flickering patch of light stopped flickering.

That is not a metaphor. To compare the brightness of a red light against a green one, you cannot simply look; the eye has no reliable ruler for it. Alternate the two rapidly, though, and any difference in brightness betrays itself as flicker. Turn the dial until the flicker disappears, note the setting, and you have measured how much a watt of that particular colour is worth to a human observer. Those measurements, gathered from a handful of laboratories and averaged across a few dozen mostly young, mostly European subjects, were assembled into a curve. An international committee adopted it, and it has lived inside every lumen since — including the one printed on the box of the last bulb you bought.

Julian Ashworth treats that as the most peculiar arrangement in metrology, and builds the book around it. Photometry is simply radiometry multiplied by that curve, which makes a lumen a measure not of light but of light weighted by an averaged eye. The candela is defined by fixing a single number at a single green frequency and letting the agreed curve carry the definition outward to every other wavelength, phrased in strictly physical language so that no observer ever has to be consulted again.

Realising it is stranger still. A national laboratory absorbs a beam of light and measures how much it warms an absorber near absolute zero, then reproduces that same warming with a known electrical current. From that comparison, and a silicon detector fitted with a filter shaped to imitate the curve, the whole photometric scale descends. The filter never quite matches, and that mismatch is where most of the genuine uncertainty in lighting measurement lives.

Then the findings that complicate everything. Let the eye adapt to darkness and the weighting shifts, so a lumen at midnight is not the same quantity as a lumen at noon — the reason a red bicycle outshines a blue one in daylight and loses to it at dusk. More awkwardly, two patches of identical measured luminance can look plainly unequal in brightness, the more saturated one winning. Photometry knows this and tolerates it, because a quantity that can be added and calibrated is worth more to engineering than one that matches experience.

The book then follows the unit into commerce: spheres built to destroy directional information and instruments built to recover it, tests that measure a finished lamp rather than the chip inside it, a colour rendering index that worked until manufacturers began optimising against it, and the recent discovery of a receptor in the eye with nothing to do with seeing — which required an entirely parallel system of quantities.

This edition carries two dozen bespoke diagrams, and forty references — standards, national laboratory publications, patents and peer-reviewed work — close it. The constants that define the candela appear exactly; peak wavelengths, sample counts and test durations do not, because the documents carrying them were not consulted, and each omission is stated where the figure would have gone.