A History of Color
Thomas Young proposes that the retina does not carry a receptor for every hue: it samples the spectrum with just three receptor types tuned to broad, overlapping bands. Hermann von Helmholtz quantifies the idea in the 1850s, and it becomes the Young–Helmholtz theory of trichromatic vision.
It is a claim about biology, but it reads like an engineering spec: if the eye reduces every spectrum to three signals, then three well-chosen records, and three well-chosen primaries, are all a color technology needs. It is also why two different spectra can look identical (metamerism), the loophole every capture and display device lives in.
Two centuries in 30 stops
From bitumen on a pewter plate to an absolute HDR transfer function: every stop on this line added a piece of machinery color science still uses. Photography learned to capture, cinema learned to reproduce, television learned to do it live, and in 1931 the CIE gave all three a way to prove it with numbers. Click along the track.
Three threads, one problem+
Photography, cinema, and television hit the same wall in sequence: capture the world, reproduce it, then prove the reproduction matches. Photography solved capture (1826–1907), cinema industrialized reproduction (1895–1935), television made it live (1926–1953), and none of them could define "correct" without a measurement system.
The 1931 CIE Standard Observer is where the three threads knot. From that point a film stock, a phosphor, and a printing ink could be compared as coordinates instead of adjectives. Every other module in this series lives downstream of that knot.
The trichromatic bet+
Young's 1802 guess, three receptor types, not one per hue, is the wager under everything here. Maxwell's 1861 projection proved three records suffice; Vogel's sensitizers made the records honest; Autochrome's starch-grain mosaic anticipated the Bayer filter by seven decades; and every three-gun, three-filter, three-subpixel display since is the same bet, cashed again.
The fine print: three numbers are only enough for one fixed observer. Individual variation and eye-versus-camera mismatch are the residual error terms: the reason TLCI and SSI exist (see the Measuring Light Quality module) and the reason colorimeters need per-display profiles.
Color before measurement was craft+
For a century, color quality was enforced by institutions, not instruments: women hand-tinting daguerreotypes frame by frame, Pathé's stencil-color production lines, Technicolor shipping its own cameras, its own crews, and Natalie Kalmus's Color Advisory Service dictating set and costume palettes by memo, color grading before grading existed.
Craft control works while one company owns the whole chain. It collapses the moment devices multiply. NTSC in 1953 is the turning point: the first time "correct color" for a mass medium was written down as coordinates a meter could check, and the moment the calibrator's profession became inevitable.
1931: why then?+
Not an accident of date. Wright's and Guild's independent color-matching datasets, 17 observers between them, agreed well enough to standardize; the National Physical Laboratory wanted a national reference; and industry was drowning in match-by-eye disputes: textile dye houses, paint and ink makers, lamp manufacturers, and the new film and broadcasting businesses all needed color specified on paper.
The CIE fused the datasets into one Standard Observer and one chromaticity diagram everyone could share. How those experiments actually worked (the bipartite field, the 2° patch, the imaginary XYZ primaries) is the next module in this series.
Standards all the way down+
NTSC (1953) → PAL/SECAM (1967) → Rec. 709 (1990) → sRGB (1996) → DCI-P3 (2007) → BT.2020 and BT.2100 (2012–2016): each is a set of CIE coordinates plus a transfer function, which is to say, each is a promise a device can drift away from.
Calibration is not improvement. It is returning a display to its published numbers, verified by a probe that traces back to the same 1931 observer this timeline builds toward. If it isn't measured, it isn't calibrated.
IF IT ISN'T MEASURED, IT ISN'T CALIBRATED. · Color Volume Explorer · Measuring Light Quality