Poincare discovered the mathematical foundations of chaos in the 1890s. Lorenz rediscovered sensitivity to initial conditions in weather models in 1963. The gap is seventy years. The mathematics existed the entire time.
The paper (arXiv:2603.07684) argues that the gap was not accidental. It was philosophical. Positivism — the strict form of empiricism that dominated physics in the early 20th century — provided explicit reasons to dismiss Poincare's chaotic mathematics as meaningless. Hadamard and Duhem, contemporaries who understood the mathematics, articulated why it should be ignored: it lacked grounding in observable experience. Mathematical structures that could not be tied to measurable phenomena were not physics. They were, at best, mathematical curiosities. At worst, they were meaningless manipulations of symbols.
This was not ignorance. It was a principled philosophical position applied consistently. Positivism had been enormously productive — it helped purge physics of metaphysical speculation, sharpened the distinction between science and philosophy, and demanded that theoretical claims be answerable to experiment. The demand for empirical grounding was the discipline's greatest strength. And it was the discipline that exiled chaos for seventy years.
The mathematics of chaos describes systems where arbitrarily small differences in initial conditions produce arbitrarily large differences in outcomes. This behavior is real — weather systems, turbulent flows, population dynamics all exhibit it. But the behavior cannot be demonstrated through the kind of controlled experiments that positivism demands. You cannot set up two weather systems with initial conditions differing by one part in a billion and measure the divergence. The chaos is visible only through the mathematics or through computation, neither of which positivism trusted as a source of physical knowledge.
The through-claim: a philosophy of science that succeeds by constraining what counts as knowledge inevitably constrains what knowledge can be discovered. The constraint is the method. The method worked — until it encountered phenomena whose signatures exist in the mathematics rather than in the measurement. The same filter that removed metaphysical noise removed a real signal. The exiled mathematics was correct. The exile lasted as long as the philosophy that imposed it.
Note (2026-11-11): checked against the full text of the paper (Park, Studies in History and Philosophy of Science 116, 2026, arXiv 2603.07684). The thesis is as the essay says: Hadamard and Duhem understood the mathematics of Poincare's chaos and rejected its physical significance on positivist grounds, calling it meaningless or useless. Three points differ. (1) The essay puts the gap between Poincare in the 1890s and Lorenz in 1963. The paper puts it between Poincare and Stephen Smale in the 1960s: it says Poincare's results were reintroduced into physics through Smale, by pure mathematics, and it describes the revival as the convergence of two initially independent threads, Lorenz's numerical work (1962 and 1963, which it says was initially distrusted for its numerical origin) and the pure mathematics of Smale, Anosov and others. (2) The paper's account of positivism is two ideas: positivists rejected unobservable causes, and understood laws as summaries of observed regularities, which a sensitivity to unobservably small differences contradicts; the essay's version, that chaos cannot be shown by controlled experiments and that positivism did not trust mathematics or computation as sources of physical knowledge, is the essay's own gloss. (3) The paper explicitly says that the standard explanation, the lack of computing power, is 'by itself insufficient' and gives the example that Smale's results came by pure mathematics, a point the essay's closing lines do not mention. The essay's closing claim that a philosophy that succeeds by constraining what counts as knowledge inevitably constrains what can be discovered is its own.