PAINT — layer 4 of seven
Why Some Colours Fade and Others Don't
The real difference is in the particle — and whether light can break it apart.

Two Kinds of Colour, Two Kinds of Fate
Every colour you see in a painting sits on the surface in one of two fundamentally different forms. The first is a pigment: a solid particle, insoluble, physically present in the paint layer like grit suspended in a binder. The second is a dye: a molecule that dissolves and stains its host material at a molecular level. This distinction — particle versus molecule — is the root cause of differential fading in old pictures.

Pigment particles tend to be chemically stable. They sit encased in oil or resin, physically shielded. A particle of red ochre, for instance, is essentially iron oxide — the same chemistry as rust, already oxidised, with almost nothing left to react to. Light falls on it and bounces. The colour holds.
Dye molecules are a different problem. They bond to their substrate chemically rather than sitting in it mechanically, and many of them absorb ultraviolet light in a way that unpicks their molecular structure. The bonds break. The molecule changes shape, and when a molecule changes shape, it absorbs different wavelengths of light. The colour shifts or disappears entirely. This is why a painted red rose in a seventeenth-century still life might still be vivid while the green leaves beside it have faded to near-nothing: the red was lead white mixed with vermilion, both pigments; the green may have been made with a copper-based lake colour — part dye — or simply a fugitive natural extract.
The word lake names a particular category: a dye that has been precipitated onto a powdery base — chalk or alumina — to make it behave like a pigment. It looks stable. It isn't. The dye molecule is still present and still vulnerable; the particle form is cosmetic. Madder lake, carmine, and many yellow lakes have faded significantly in works across every major collection because of this.
What makes fading uneven across a single canvas is the mixture. Two adjacent passages might carry very different proportions of fugitive dye versus stable particle, and light treats each according to its chemistry, not its position in the composition. The painter usually had no way of knowing which colours would outlast the others. The result is a surface that no longer reads as intended — not damaged, exactly, but shifted. What you are looking at may be a shadow of a shadow.