# Examples of Kuhn loss?

A Kuhn loss is:

a success, empirical or theoretical, of a prior theory – or paradigm as Kuhn would have preferred – that does not carry over to the theory or paradigm that replaced it. [Midwinter and Janssen, see below.]

Kuhn introduced the concept in The Structure of Scientific Revolutions (p. 99–100, 3rd ed.). The examples he gives there come from the phlogiston theory, which explained:

• Why metals are so much more alike than their ores: All metals contain phlogiston.
• Why burning something in a closed flask reduces the pressure of the air: The phlogiston released by combustion “spoils” the elasticity of the air, just as fire “spoils” the elasticity of a steel spring.

I have also run into two examples from Cartesian vortex theory, which explained:

• Why all the planets revolve around the sun in the same direction: They are all carried by a common vortex.
• Why there is gravity: The vortex “pushes” bodies “downward”, due to differential impact of the vortex stream (see Aiton’s book The Vortex Theory of Planetary Motions for more details).

Finally, the article by Midwinter and Janssen, "Kuhn Losses Regained: Van Vleck from Spectra to Susceptibilities" gives a lengthy account of a technical example in the theory of electric and magnetic susceptibilities. Briefly, there is a constant C in that theory. According to classical physics, C = ⅓. The old quantum theory (1913–1925) gave much larger values, the exact value depending on the model and the way quantum conditions were imposed. In the new quantum theory (i.e., modern quantum mechanics), the experimentally correct value of ⅓ is restored.

Since posting this, another near-example has occurred to me: Kepler's 1609 prediction of the sun's rotation, confirmed shortly by the discovery of sunspots. Kepler's prediction stemmed from his theory of a whirlpool-like force from the sun sweeping the planets around; in Newton's system, the rotation of the sun is just an unexplained fact. But this example doesn't quite qualify, since Kepler's whirlpool force was never widely accepted; also, the prediction was not fully successful. (Kepler predicted the wrong period and the wrong axis of rotation.)

I should mention that the concept of Kuhn loss remains controversial, but these examples seem reasonable at first glace. Note that a Kuhn loss can later be regained with a yet newer theory. Electron theory, for example, explains the similarities of metals.

Taxonomy. Kuhn loss comes in several varieties. Both explanations and predictions can be lost. Most examples concern lost explanations. Subsequent history can take several courses. (a) Changing standards of scientific explanation can turn the old issue into a non-problem. For example, Newton's gravity was eventually accepted as fact, not needing a mechanical cause. (b) A yet newer theory can provide a new explanation, as in the metal example. (Here, the electron theory of metals does not refute Lavoisier's "chemical revolution"; on the other hand, the new quantum theory replaced its immediate predecessor. So two versions of this case.) (c) The orphaned phenomenon (deprived of its old explanation) can be treated as an anomaly, something still requiring resolution. Example: geocentric theories explained the lack of stellar parallax, and for almost 300 years, detecting it was an outstanding problem. When Bessel finally detected it in 1838, heliocentricity had long been accepted.

To answer a question in a comment, then:

Doesn't this also mean that they must have been "rediscovered" during a later theory/paradigm shift, since they are defined in terms of previous successes? Or do all things that were believed to be successes in the previous paradigm but later discarded count, even though they were due to e.g. biased observation methods or other misconceptions?

All these count as Kuhn loss. The C-constant example is of type (b); the stellar parallax example was of type (c) until 1838, transitioning then to type (b); the metal example transitioned from type (a) to type (b).

Last clarification: you don't have to accept Kuhn's relativism to be interested in Kuhn loss.

What other examples of Kuhn loss are there?

• The whole Kuhn theory is highly controversial. On my own opinion he just did not understand enough of science whose history he tried to study. This applies to most professional philosophers. Nov 9, 2014 at 17:19
• So you are asking about features of a theory, that can in hindsight be considered "successful" in the sense that the previous theory provided useful predictions on the issue, but that were lost in a paradigm shift? Doesn't this also mean that they must have been "rediscovered" during a later theory/paradigm shift, since they are defined in terms of previous successes? Or do all things that were believed to be successes in the previous paradigm but later discarded count, even though they were due to e.g. biased observation methods or other misconceptions? Nov 20, 2014 at 12:47
• [continued]: However, the C-constant example seems to imply the former (only "true" successes count as Kuhn losses), but maybe you can clarify this. Nov 20, 2014 at 12:48
• @Michael Weiss: Of all he wrote I only read the Structure of scientific revolutions, and on my opinion his understanding of how science develops and scientific revolutions is wrong. Nov 20, 2014 at 23:08
• Actually the term Kuhn loss was introduced by Heinz Post in his 1971 article on Correspondence and heuristics. For example see the article by Votsis. Post claims there are no Kuhn losses. (Post was my PhD advisor) Mar 25, 2018 at 19:20

## 1 Answer

One example sometimes given is the lack of an explanation for hybrid sterility in Darwin's original theory of evolution by natural selection, something that was explained by theistic evolution. Another is the lack of an explanation for dispersion and selective absorption in the wave theory of light, something that was explained by the emission theory. However, like many of Kuhn's examples, neither of these was much of a 'loss' in practice, as neither of the displaced theories had genuine predictive power.

Perhaps a better example is the classical description of fluid dynamics given by the Poiseuille equation, which cannot (yet) be derived in the modern quantum physics paradigm. However, its usefulness means that it continues to be taught alongside the quantum model, despite the fact that it can no longer be 'justified' in the accepted paradigm.

• You are right that general relativity is incompatible with modern quantum physics, but GR has not yet been "lost" in any sense. GR has passed every experimental test given it, to many significant digits. Proposed replacements such as M theory has not even reached the status of theory, and if GR is thereby replaced so is quantum theory. GR is taught alongside quantum theory because neither has yet been replaced. GR is therefore not an example of Kuhn loss. Apr 21, 2015 at 19:45
• I agree that GR is not a loss in any real sense. However as far as I understand it it does qualify as a Kuhn 'loss' as it is a success of an outgoing paradigm (classical mechanics) that does not (yet) carry over to the new paradigm (quantum mechanics). Apr 21, 2015 at 20:28
• You seem to lump GR together with classical mechanics, and you also seem to consider quantum mechanics alone as the new paradigm. How do you justify either opinion? As I see it, most scientists consider both GR and QM to be the current paradigm, both to be incompatible with Newtonian mechanics (and with each other), and that the next paradigm will probably replace both. GR is indeed closer to Newton than QM is, but is that enough to place QM as new and GR as not? Apr 21, 2015 at 20:49
• @UriZarfaty Do you have more info about the Poiseuille equation? Why is it incompatible with QM? And does that make the empirical success of the P. equation a puzzle ("anomaly", in Kuhn-speak) for QM? (Like Feynman's remark in his Lectures that ferromagnetism has never been explained with QM, though I don't know if that's still true.) Jun 27, 2015 at 14:37
• @MichaelWeiss The word incompatible is too strong and the result of careless editing of my answer. I think the point is that many fluid flow properties cannot yet be derived using QM; this example is given in books.google.co.uk/… Jun 27, 2015 at 15:34