From Bas C van Fraassen Scientific Representation: Paradoxes of Perspective
p. 102-103 “Consider the rainbow. We realize pretty soon that there is no real material shining arch standing above the earth, although at first it looks that way. As a second guess we might think that certain parts of the clouds or haze are colored. But that cannot be maintained because if we move, we see the rainbow in a different location on the cloud or haze background.
In fact, we realize then that our usual way of speaking involves us in falsehood. I see a rainbow and you say you see it too. See what too? How can you be seeing the rainbow I see, when yours is located in a different place? Nor are they simply in a different place in our respective visual fields, in the way clouds are. For if that were so, we would see the colors ‘attached’ to the same part of the cloud, modulo parallax. If on the other hand I say there are two rainbows, and you agree, we are not even counting the same things. In fact, we are not counting things at all.
But thirdly, we are not hallucinating. Hallucinations are private, subjective. These rainbow observations are like hallucinations, in that they are not real things. But they are unlike hallucinations because they are public. Nature creates public hallucinations. So public, in fact, that the camera captures them as well! The observations are scientifically significant in part because they can also be made indirectly, so to speak, with the camera as instrument“.