I am trying to think of an example of scientific explanation whose scope was in fact more limited than we initially thought. The idea would be the following:

Initially, we used H (the explanation) to explain a certain phenomenon (call it x) and we took a range of phenomena to be relevantly similar to x in the sense that H would also apply to them. We latter discovered that the phenomena we took to be relevantly similar to x were not so and that another explanation was needed to explain them. We did not however discover that H did not apply to x.

Basically, I am trying to think of a historical example of such a situation in science; I am convinced it must exist.

Thanks in advance for your help!


3 Answers 3


Quasicrystals seem to be a good example, even if that might need some technical details. In a nutshell: crystals were defined as materials producing sharp diffraction spots; it was thought that translational symmetry does the trick. However sharp diffraction spots arranged in a fivefold pattern were discovered and this type of symmetry does not allow for translation. Translation came to be replaced/expanded by a weaker notion of long range order: classical crystals were understood to be simply periodic while quasicystals are almost periodic, which, strictly put, is "aperiodic".

Actually the distinction order vs. disorder which was considered to be a matter of logic and quality came to be seen as a matter of degree. But(!) in this case it was not a theory that was found to be approximately true: nature turned out to be more subtle. Translational symmetry is still a good explanation for crystals, even if now they would be better named "classical crystals".

  • Great, thanks a lot for this example!
    – Philo102
    Commented Nov 26, 2020 at 14:57

This question is interesting, because it points up the fact that a scientific theory can experience a reduction in its scope and explanatory power without being rejected as completely wrong. In addition to the answer given by sand1, here are some other examples that might fit the bill.

Dalton's theory of atomism. According to Dalton, all matter is composed of atoms of the chemical elements. This theory has considerable explanatory power. It succeeded in accounting for the chemistry that was known in Dalton's day, such as the fact that substances can be reproducibly decomposed into the same elements, and that elements combine in fixed proportions to make compounds, etc. Dalton's theory was that atoms are indivisible and the elements are immutable, and that all observable changes are the result of atoms separating and combining. The latter turned out to be incorrect. Atoms are divisible and elements can turn into other elements by radioactive decay. Nevertheless the core idea remains that atoms are the fundamental particles that constitute chemical elements, and chemical changes can be explained in terms of atoms separating and combining. We need other theories to explain nuclear changes.

Conservation of mass. Classically it was thought that matter was conserved. There was strong empirical support for this, and it appeared to hold universally. Later it was shown that in relativistic settings the energy associated with the mass of a body can be converted into other forms of energy. The principle is still useful, however, just not universal.

Charge, parity and time symmetry. It used to be thought that all of these forms of symmetry held independently. Later we learned that there are exceptions to each of them, but the combination of all three appears to be symmetric. This means we still have a working theory of symmetry, but it has less scope and is weaker than having three separate ones.


Well take for example:

  • statistical methods in social science

  • qualitative vs quantitative and merging them

  • any mathematical theory that starts as abstract and later becomes to explain something real, such as statistical models

These, I would say, start as "formal ideas about how it would be nice to see things". Then they're "verified" by using them succesfully in empirical studies.

What's the role of philosophy of science here? Well, because fundamentally it's about "how to see things".

While linear models may be still usable, it'd be intutive to say that stochastic models are a revolution, since they allow "seeing between only nice shapes". Similarly as irrational numbers could be seen to revolutionalize rational numbers.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .