In the 1990s, physicists started looking beyond their disciplinary boundaries by using their methods to study various problems usually thrown up by financial economics. This dissertation deals with this extension of physics outside its disciplinary borders. It seeks to determine what sort of discipline econophysics is in relation to physics and to economics, how its emergence was made possible, and what sort of knowledge it produces. Using a variety of evidence including bibliometric analysis Chapter 1 explores the field’s disciplinary identity as a branch of physics even though its intellectual heart is better seen as the re-emergence of a 1960s research programme initiated in economics. Chapter 2 is historical: it identifies the key role played by the Santa Fe Institute and its pioneering complexity research in the shaping of methodological horizons of econophysics. These are in turn investigated in Chapter 3, which argues that there are in fact three methodological strands: statistical econophysics, bottom-up agent-based econophysics, and top-down agent-based econophysics. Viewed from a Lakatosian perspective they all share a conceptual hard-core but articulate the protective belt in distinctly different ways. The last and final chapter is devoted to the way econophysicists produce and justify their knowledge. It shows that econophysics operates by proposing empirically adequate analogies between physical and other systems in exactly the ways emphasised by Pierre Duhem. The contrast between such use of analogy in econophysics and modeling practices implemented by financial economics explains why econophysics remains so controversial to economists.