The Aim and Structure of Physical Theory

Summary

The Aim and Structure of Physical Theory (La Théorie physique: son objet et sa structure) was published in Paris in 1906 by Pierre Duhem, professor of theoretical physics at Bordeaux, after appearing as a series of articles in the Revue de philosophie. A second edition (1914) added two essays, 'The Physics of a Believer' and 'The Value of Physical Theory'. The French original is in the public domain, but the standard English translation, by Philip P. Wiener (Princeton, 1954, with a foreword by Louis de Broglie), is still under copyright, so the library provides this guide without a stored text and without quotations. Part I asks what a physical theory is for. Duhem rejects the view that physics explains, that is, strips away appearances to reveal the underlying reality of matter. If physics aimed at explanation, it would depend on metaphysics and share its disputes, since Aristotelians, atomists and Cartesians each explain differently. A physical theory is instead a system of mathematical propositions, deduced from a few principles, that represents a body of experimental laws as simply, completely and exactly as possible. Such a theory economizes thought and orders laws, and as it grows more perfect it tends toward a 'natural classification' that reflects, without proving, a real order in things. Duhem contrasts the 'abstract' theories favoured by continental physicists with the mechanical models favoured by British physicists such as Kelvin and Oliver Lodge, and links the contrast to Pascal's distinction between minds that are strong but narrow and minds that are ample but weak. Part II examines the structure of theory. Physics turns qualities into quantities by measurement; mathematical deduction must allow for approximation, since a deduction that is exact for ideal data may be useless for real ones, as Hadamard's study of paths on a curved surface showed. An experiment in physics is not simply the observation of facts but their interpretation by means of theory, so that the report of an experiment is already theoretical. Physical laws are symbolic, approximate and provisional. From this follows Duhem's best-known argument: an experiment can never condemn a single hypothesis in isolation, but only a whole group of hypotheses taken together, and the physicist must decide by 'good sense' which to change. For the same reason a 'crucial experiment' that settles between two rival hypotheses, such as Foucault's measurement of the speed of light in water, is impossible in physics. The last chapter argues that hypotheses should be chosen with the history of physics in view, and traces the long development of the theory of universal gravitation.

Top Lessons

• Representation, not explanation

  Duhem separates physics from metaphysics. A physical theory summarizes and orders experimental laws in mathematical form; it does not claim to reveal what matter ultimately is. This keeps physics free of the disputes among philosophical schools and keeps metaphysics free of the changing fashions of physics.

• Hypotheses are tested in groups

  Every experiment in physics relies on theories of the instruments and on many auxiliary assumptions. When a prediction fails, logic shows only that some member of the group is false, not which one. This is the thesis later called the Duhem thesis, or, in a broadened form, the Duhem-Quine thesis.

• No crucial experiment

  Foucault's experiment was hailed as deciding between the emission and wave theories of light. Duhem argues that it refuted the whole system of assumptions surrounding the emission theory, not one hypothesis, and that a crucial experiment would require listing every possible hypothesis, which cannot be done.

• Good sense and tradition

  Since logic alone does not tell the physicist which hypothesis to give up, Duhem appeals to good sense, a trained judgment that weighs reasons logic cannot formalize, and to the history of science, which shows how theories grow by continuity rather than sudden invention.

Verified Quotes

Medical Lens

A work on the philosophy of physics, but Part II, chapter 6 compares testing in physics with testing in physiology and medicine. Duhem discusses Claude Bernard's rules for the experimental method and argues that a physiologist can often test one idea at a time, while a physicist's instruments and interpretations already involve whole theories. He compares the physicist to a physician who must diagnose an illness from the state of the whole body, rather than to a watchmaker who can take apart a mechanism. These passages bear on the history of medical method and on later discussion of how clinical hypotheses are tested together with background assumptions.

• Part II, ch. 6, opening sections

  Duhem contrasts Claude Bernard's account of experiment in physiology with the logical situation of experiment in theoretical physics.

• Part II, ch. 6, on the impossibility of isolating a hypothesis

  The comparison of the physicist with a physician who diagnoses the whole organism rather than a watchmaker who dismantles a watch.