I paint the view from my window; one particular spot, determined by its position in the architecture of the house, I paint ochre. I say this is the colour I see this spot. That does not mean that I see the colour of ochre here, for in these surroundings this pigment may look lighter, darker, more reddish, (etc.). "I see this spot the way I have painted it here with ochre, namely as a strong reddish-yellow".

But what if someone asked me to give the exact shade of colour that I see there? — How should it be described and how determined? Someone could ask me to produce a colour sample (a rectangular piece of paper of this colour). I don't say that such a comparison would be utterly uninteresting, but it shows us that it isn't from the outset clear how shades of colour are to be compared and what "sameness of colour" means.

Imagine a painting cut up into small, almost monochromatic bits which are then used as pieces in a jig-saw puzzle. Even when such a piece is not monochromatic it should not indicate any three-dimensional shape, but should appear as a flat colour-patch. Only together with the other pieces does it become a bit of blue-sky, a shadow, a high-light, transparent or opaque, etc. Do the individual pieces show us the real colours of the parts of the picture?

We are inclined to believe the analysis of our colour concepts would lead ultimately to the colours of places in our visual field, which are independent of any spatial or physical interpretation; for here there is neither light nor shadow, nor high-light, etc.,etc..

The fact that I can say this place in my visual field is grey-green does not mean that I know what should be called an exact reproduction of this shade of colour.