The share of light a surface sends back, measured wavelength by wavelength.
Spectral reflectance runs from 0 (absorbs everything) to 1 (returns everything) at each wavelength. A surface's colour under a given light follows from its reflectance multiplied by that light's spectrum.
Reflectance is a property of the surface; colour is what results when a particular light and a particular observer meet it.
A red paint has high reflectance above about 600 nm and low reflectance elsewhere.
Spectrophotometers measure reflectance because it is the one description of a surface that holds under every light.
The CIE defines reflectance as reflected radiant flux divided by incident flux, a pure ratio. Instruments report a closely related quantity, the reflectance factor: the light a sample returns compared with what a perfect white diffuser would return in the same geometry, wavelength by wavelength. The instrument is calibrated on a white tile of known values. Because the comparison depends on the angles of lighting and viewing, the geometry (45°/0° or sphere, gloss included or excluded) is part of the result. Reflectance, transmittance and absorptance at any wavelength add up to one.
| Colour | 450 nm | 550 nm | 650 nm |
|---|---|---|---|
| Brick Red #a62f20 | 3% | 4% | 24% |
| Sky Blue #87ceeb | 85% | 61% | 16% |
| Sage Green #9caf88 | 26% | 45% | 19% |
The brick red returns 24% of the light at 650 nm and 3% at 450 nm: it is red because of what it absorbs. Sage green's three figures (26%, 45%, 19%) span a far narrower range than Sky blue's, the flatter curve of a greyed colour. A vivid colour has a steep curve and a muted one a shallow curve, whatever the hue.
How this was calculated: Each curve is a smooth spectrum fitted so that it reproduces the catalogue HEX value under D65. It is one of many spectra that would, and stands in for a spectrophotometer reading rather than replacing one.
A replacement material is matched to an original using only CIELAB values under one illuminant.
The two can share those values with different reflectance curves and separate under other lights. Comparing the curves themselves is the only check that covers every light at once.
A translucent or thin sample is measured on whatever surface is to hand.
Light passes through, reflects off the bench and returns, so the reading includes the backing. The same sheet over white and over black gives two different curves.