At the heart of the book are historical studies of four sets of experiments in modem physics: Millikan's oildrop experiments, which established the quantization and magnitude of the charge on the electron; the experiments that in 1964 and 1965 established the CP (charge-parity) noninvariance of the decays of K particles; the 1957 experiments that established the nonconservation of parity; and the measurements made between 1928 and 1930 by R.T. Cox and Carl T. Chase, which might have also established the nonconservation of parity but didn't.
The traditional neglect of experiment is legitimized by the assumption that experimentation is somehow trivial and unproblematic, that scientists just set up their apparatus according to principles laid down in advance, and that incontestable facts simply pop out. Franklin's studies expose the naiveté of this view. They exemplify the complexity and uncertainty, the space for doubt, inherent in the performance and interpretation of experiments. They also exemplify some of the strategies scientists routinely use to render doubt manageable and findings convincing&3151;checks, calibrations, interventions, repetitions, variations in techniques and so on. But the case of the "nondiscovery" of parity non-conservation shows that such strategies can fail.
Cox and Chase detected a left-right asymmetry in the double scattering of electrons that would be interpreted today as evidence for the nonconservation of parity. In 1930 this interpretation occurred to nobody. Contemporary interest in double-scattering experiments came quickly to revolve around Mott's new calculations of the process in Dirac theory. Throughout the 1930s a whole sequence of experiments threatened the calculations and hence the theory. But neither the calculations nor the theory were abandoned; instead the experiments were repeated and developed until, in 1940, experimenters succeeded in getting the predicted results. In this instance, then, theory stood firm against experiment and was eventually confirmed by it. Along the way, the potentially revolutionary findings of Cox and Chase were forgotten. As Franklin notes, theoretical context is itself an important factor in the evaluation of any experiment and its findings.
Scientists will find these studies thought-provoking. They could serve students as partial surrogates for first-hand experience of the uncertainties of real research. Both scientists and students may be puzzled, though, by the frame in which the studies are set. All are couched as refutations of the highly implausible view that experiment plays little or no role in the production of scientific knowledge.
In short philosophical interludes Franklin formalizes his counterargument, and criticizes those who would oppose him. But he fails to understand the positions of his supposed adversaries&3151;Thomas Kuhn, Paul Feyerabend and others. In granting a role for theoretical context in the assessment of experiments, Franklin is closer to his villains than he thinks, and his case studies are all the more interesting for it.