Newton's ”experimentum crucis”. By rotating the prism on the left on its axis Newton could let one color at a time pass along a fixed path through the hole in the second screen. Once this color passed through the second prism, it according to Newton, retained its color and would not produce a new complete spectrum.

In the circles is shown what is actually seen when one looks through the second prism towards the light source (upper circles) and what one sees on the last screen (right) when different colors are passed through the second prism.

The experimentum crucis was criticized already in Newton's time by other scientists. Newton responded to this criticism with annoyance, explaining that the meaning of his experiment was not understood or it was simply not done right. All the conclusions he makes are, according to him, drawn straight from the ”observational facts” of the experiment. He stresses the point that he does not make hypothesis. He just reports the observations he has made. As it happens, the main objections against Newton have been precisely about the hypothetical nature of his theory, that Newton did in fact hypothesize, and that the experimentum crucis gave little if any evidence for his hypothesis – that white sun light is composed of all the colors of the spectrum, and that different spectral colors have different amount of refrangibility.

When giving a lecture for the society of natural philosophy in Helsinki on these experiments in the nineties, I noticed that the participants were at first puzzled by the questions the experiments rise. A gentleman finally said that he would do the experiments with a slit instead of a prism. No white areas there. Problem solved!

This is a typical answer from physicists when confronted with Goethe's objections. Modern physics does not need any experiments with prisms or colors. Quantum electrodynamics handles the problem quite well without any such oddities. Why then does the same physics still maintain that white sunlight is composed of all the colored lights in a spectrum? Does it follow from observations or is it the theory again?