The phenomenon of superconductivity has so for defied all attempts of explanation since it was first discovered in 1911. Although two phenomenological theories have been put forward and proved very successful, yet no atomic theories based on quantum mechanics have proved adequate. In the present thesis a new theory is proposed which will be shown to account for most of the important properties of a superconductor. In this part we briefly review the experimental facts and the earlier theories respectively ~1 and ~2.

The central aim of this thesis is to present a new kind of realism that is driven not from the traditional realism/anti-realism debate but from the practice of physicists. The usual debate focuses on discussions about the truth of theories and how they relate with nature, while the real practices of the scientists are forgotten. The position I shall defend is called "phenomenological realism". The realist doctrine was recently undermined by the argument from pessimistic meta-induction, also known as the argument from scientific revolutions. I argue that phenomenological realism is a new kind of scientific realism that can overcome the problem generated by the pessimistic meta-induction, and which reflects scientific practice. The realist has tried to overcome the pessimistic meta-induction by suggesting various types of theory dichotomy. I claim that the different types of dichotomy normally presented by realists do not overcome the problem, for these dichotomies cut through theory vertically. I argue for a different kind of dichotomy, one that cuts horizontally, between high-level and low-level theoretical representations. I claim that theoretical forms in physics have two distinct types depending on the way they are built. These are theoretical models that are built depending on a top-down approach and phenomenological models that are built depending on a bottom-up approach. I argue that for the most part only phenomenological models are the vehicles of accurate representation. I present two case studies. The first case study is from superconductivity, where I contrast the BCS model of superconductivity with the phenomenological model of Landau and Ginzburg. The other case study is a fresh look at the Bohr-Einstein debate.