Books on Amazon
Theory/Duhem: Thesis: The goal of physical theory formation is not the knowledge of reality, because it would move in a metaphysical field of concepts, but the design of a formal system that has optimal order properties for the world of phenomena.
The structure of science is a holistic context, not a combination of single sentences whose truth values are determined in isolation. Only the whole of science can be compared with the totality of facts and examined.
Observations are "theory-laden".
Theories are not inductively gained knowledge of experience, but designs of the human mind, agreements of a formal nature, whose empirical utility is found only in practice. Conventionalist science-perception, nevertheless Poincaré: it is not the definitive status of the fundamental laws, which deprives them of the revision, such revisions, even of the fundamental laws, can become necessary and meaningful, they cannot be forced experimentally.
Delimitation from metaphysics, which does not allow refutation by observation. (Also Popper, 1934) autonomy of physics.
Today one would say: theory is supposed to explain phenomena. The performance of theories or laws can be characterized by the logical link with verifiable statements. For Duhem (in contrast to today's language usage): "Description".
"Explanation"/Duhem: reserved for the metaphysicians' claim for knowledge.
Metaphysics: (DuhemVs): Premises must be distinguished ontologically, they must concern "things in themselves". Observable properties are "explained" in such a way that they are attributed to another (unobservable) reality layer. E.g. "ether", heat: atomic movement. This is the Cartesian concept of explanation, which Duhem does not grant physics.
His physics conception is antimetaphysical, unlike Mach, but he does not consider metaphysics to be pointless.
Definition Physical Theory/Duhem: A physical theory is not an explanation. It is a system of mathematical theorems that are derived from a small number of principles and have the purpose of presenting a group of experimental laws as simple as they are complete and accurate.
A correct theory does not explain reality, but presents a group of experimental laws satisfactorily.
A false theory is a set of equations that do not agree with the experimental laws. (Comparison with reality would be metaphysics).
Revision of a theory: the relationships remain, but their nature is understood differently.
Theories/Duhem: consist of 2 parts
The explanatory part is by no means the adequate basis of the descriptive. It is not the seed or the root.
Ziel und Struktur der physikalischen Theorien Hamburg 1998