Four days on a single Cercis siliquastrum pod. Scarlet runner beans, actually the pod of the Judas tree, but that's the specimen. It's a decent piece technically. The drawing is right, the shadow side reads, the paper is behaving. And the green is dead. Not muddy, not too cool or too warm, just dead, the way a photograph of a leaf is dead.

I've tried the obvious corrections. Cooled it with a touch of phthalo, warmed it back with transparent yellow, dropped the value a step. All it did was move the corpse around. My suspicion is that I've been matching the local colour too faithfully and left out the thing that made me want to paint it, which was the way the light came through the thinner edge of the pod this morning, almost yellow, and the way the surface picked up a grey bloom over the top of it. Two different greens fighting, essentially. What I've painted is one green averaged out.

I've done this before and I know the answer is usually to choose. But I keep reaching for the brush to fix the colour rather than to state the light, and I'm four days in and reluctant to lift it back to the paper.

For anyone who works wet-in-wet on a large single form: do you build the two temperatures in from the start, or lay a proper base and then come over it? I've always done the latter and I'm beginning to wonder whether that's the whole problem with this one.