In the first three chapters we saw some of the varied roles that experiment can play in the choice between competing theories or hypotheses. In the case of parity nonconservation, the experiments of Wu, Garwin, and Telegdi and their collaborators were “crucial” experiments that decided the issue between two classes of theories: those that conserved parity and those that did not. The Princeton experiment was an example of a “convincing” experiment. It provided the evidence that persuaded the majority, though not all, of the physics community that CP invariance was violated. That episode also produced experiments refuting or eliminating most of the alternative explanations to CP violation. This was an example of a pragmatic solution to the Duhem–Quine problem.

The episode of the nondiscovery of parity nonconservation illustrated two other roles for experiment. The experiments on Mott double scattering at 0° and 180° did not lead to refutation or rejection of Dirac's theory. They did lead to various unsuccessful ad hoc modifications of the theory and to repetitions of the experiment until the discrepancy was resolved. The double-scattering experiments at 90° and 270°, which, in retrospect, should have been interpreted as showing parity nonconservation, did not seem to have much effect at all. They did not lead to any new theoretical explanations, even though it was recognized at the time that existing theory provided no explanation for their results, nor did they lead to much further experimentation. Their significance was not realized at the time; they were first neglected, and then forgotten, until 1957.