ARTIFICIAL WASTELAND · A FLAME-TEST BENCH

The color you see is a wavelength you can read

A red firework is glowing strontium; a green one is barium; a blue one is copper. Each metal salt, heated in the burst, emits light at fixed wavelengths set by its atoms and molecules. Pick a salt below — the color in the swatch isn't chosen by hand, it's computed from that salt's real emission lines through the response curves of your own eye. Then turn up the heat and watch why a true blue is the hardest color to make.

Its emission spectrum

400 nm · violet550 nm · green700 nm · red

Show the check — the color, derived

Each emission line excites the eye's three cone responses (the CIE 1931 standard observer). Sum them across the spectrum → a tristimulus XYZ → sRGB. No color was picked; this table is the whole arithmetic.

Why blue is the hardest color

Red and green ride on rugged emitters — strontium and barium bands that shine brighter the hotter the flame. Blue is different. The only practical blue emitter is copper(I) chloride, CuCl, and CuCl falls apart above about 1200 °C. But a firework star burns at 1500–2000 °C. So a blue star has to thread a needle: hot enough to burn and ignite, cool enough that the CuCl survives long enough to glow. That narrow, contradictory window — not pigment, not rarity — is why a deep, saturated blue is the connoisseur's test of a fireworks maker. Select Copper above and drag the heat past 1200 °C: the blue dies in your hand. Do the same with strontium and the red only brightens.

The honest footnote the color charts skip: the reds, greens and blues are mostly molecular emission — strontium monochloride, barium monochloride, copper monochloride — not bare metal atoms. Only sodium, lithium and potassium here glow as clean atomic lines. The spectrum plot marks which is which.

How this page is built — and checked

The wavelengths are real and sourced. Atomic lines (Na, Li, K) are from the NIST Atomic Spectra Database; the molecular band heads (SrCl, BaCl, CuCl, CaCl) are from pyrotechnic-flame spectroscopy (Douda; Conkling, The Chemistry of Pyrotechnics). Every swatch color is computed from those wavelengths by the CIE 1931 2° observer (Wyman, Sloan & Shirley 2013, an analytic fit to the standard color-matching functions). An offline verifier re-derives all of it from scratch and asserts each color lands in the bin the chemistry predicts, and that copper's blue collapses past CuCl's decomposition point — research/fireworks-colors/verify.mjs, 28 checks. The page and the verifier import the same spectra.mjs, so the color on screen is the color that was checked. The temperature fall-off is a labelled pedagogical model of a real fact (CuCl decomposes >~1200 °C), not a measured intensity curve.