Scientific Laws as Norms of Modelling — Epoche B2
Beyond Inherence: Scientific Laws as Norms of Modelling Introduction When scientists invoke Newton's laws of motion or the conservation of energy [1] , the intuitive reading is that these statements describe objective features of the world. On this view — broadly, scientific realism about laws — a law is a regularity or a power that is already there in nature, waiting to be found, and a scientific model is an attempt to represent it accurately. This essay argues for a different reading: that a fundamental law functions less as a description of any real system than as a norm governing which models a scientist is permitted to build. The case does not rest on scepticism about science. It rests on two things that are easy to check — what the fundamental laws actually say when applied to a real system, and what scientists in fact do when a law is contradicted. The traditional view: laws as inherent properties The realist reading takes a law to state an objective regularity. Newton's third law — that every action is met by an equal and opposite reaction — is then a truth about how forces behave everywhere and always, holding independently of anyone's noticing it. Talk of laws as "inherent" means they are intrinsic features of systems, or dispositions that fix how those systems behave under stated conditions. Models, being simplified and idealised, are approximations to these prior truths, and a model is good to the extent that it is faithful to them. The difficulty is visible as soon as a law is applied. Newton's law of gravitation gives the force between two bodies as inversely proportional to the square of their separation. But no real pair of bodies is subject to gravitation alone, and the law says nothing about what happens when a charged body is also in an electric field. Written honestly the law carries a suppressed qualifier — other things being equal — and other things never are. Nancy Cartwright pressed exactly this point in How the Laws of Physics Lie (1983): the fundamental laws of physics achieve their generality by not describing any actual situation. What describes actual situations are phenomenological laws, tied to particular set-ups and far less general. There is a trade-off, and the fundamental laws sit at the end of it where explanatory reach is bought with descriptive falsehood. Popper, and why falsification does not settle the question It is tempting to reach for Karl Popper here [2] , and important not to misuse him. In The Logic of Scientific Discovery (1959) Popper argued that theories are never verified by accumulating favourable instances, and proposed instead that what marks a statement as scientific is its falsifiability: there must be some possible observation that would count against it. Science then proceeds by conjecture and refutation, and a theory that survives testing is corroborated rather than proved. Popper was a critical realist, and he did not license the practice described below. He explicitly condemned what he called conventionalist stratagems — rescuing a threatened theory by adding an assumption whose only motivation is the rescue — and treated such immunising moves as the way a research programme forfeits its scientific status. Any account that reads the protection of core laws as a Popperian insight has the history backwards. What makes the ban hard to enforce is a point Pierre Duhem had made in 1906 [3] : an experiment never tests a single hypothesis. It tests the hypothesis together with the auxiliary assumptions needed to connect it to an instrument reading, so a discordant result tells you that something in the bundle is wrong without telling you what. The logic of refutation therefore leaves open which element to revise, and it is here, in a gap that falsification cannot close, that the question of this essay lives. If the law is not what logic forces you to give up, what is it that keeps it in place? Three anomalies, and what was actually revised The historical record answers concretely, and it does not always answer the same way. Consider three cases in which a well-established law was contradicted by measurement. Uranus, tracked through the 1830s and 1840s, drifted from its Newtonian ephemeris by about two arcminutes — small, but far beyond observational error. Urbain Le Verrier and John Couch Adams held the law of gravitation fixed and revised the model of the solar system instead, positing an unseen further planet. Johann Galle found Neptune in September 1846, within about a degree of Le Verrier's predicted position. Mercury's perihelion advances by roughly 43 arcseconds per century more than Newtonian gravitation predicts. Le Verrier applied the identical move, postulating an intra-Mercurial planet he named Vulcan. Nobody ever found it. The anomaly was resolved only in 1915, when general relativity replaced the law itself and returned the 43 arcseconds without any new body. In beta decay the emitted electron carries a continuous range of energies, apparently violating conservation of energy in each individual decay. Niels Bohr was prepared to abandon strict energy conservation for such processes. Wolfgang Pauli, in 1930, instead held the conservation law fixed and postulated a new, almost undetectable particle to carry off the missing energy. Clyde Cowan and Frederick Reines detected the neutrino in 1956. Two of these three vindicated the strategy and one did not, which is precisely what makes the pattern informative rather than merely reassuring. Anomaly What was held fixed What was revised Outcome Uranus off by about $2'$ of arc the inverse-square law the census of planets Neptune found, 1846 Mercury's extra $43''$ per century the inverse-square law the census of planets (Vulcan) failed; the law itself replaced in 1915 Continuous beta spectrum conservation of energy the inventory of particles neutrino postulated 1930, found 1956 Laws as norms of modelling The common structure is that in each case the law was not treated as the hypothes