Long-term history of science: on the flexibility and fragility of scientific disciplinesMost scientific disciplines, such as chemistry, biology and physics, are now about two centuries old. Using physics as a case study the present paper aims to account for this longevity. What kept the physics discipline together from the early nineteenth century onwards? Literature on the rise of physics suggests that the discipline was formed around energy, the ether, or other theoretical notions. Yet the twentieth-century revolutions in physics showed that the discipline could prosper without some of its most 'fundamental' concepts. Some scholars conclude that internal factors are therefore irrelevant and disciplinary identity and continuity are purely institutional. Drawing on the work of Thomas Kuhn, Peter Galison and Andrew Warwick, this paper defends a different point of view. Although there is no intellectual core of disciplines, the prolonged existence of disciplines cannot be explained without some degree of internal continuity. If there is a revolution of a theoretical level, there may still be continuity on the level of experimental practices (and vice versa). It is this flexibility that accounts for the fact that disciplines may adapt to different circumstances. In addition, an educational tradition is required to transmit knowledge from one generation to the next.