Kinds and (In)Commensurability

  • Jed Z. Buchwald
Part of the Boston Studies in the Philosophy of Science book series (BSPS, volume 151)


For several decades many historians of science have not felt comfortable with philosophers of science, because contemporary philosophy has not often seemed to provide much that would be useful in historical practice. History wants pragmatic value from philosophy. Philosophy has until recently been unable to provide much of it. “Until recently” seems to imply that philosophy of science is or is about to become useful. To the extent that normative concerns persist, philosophy remains without much importance to historians.3 It must instead try to penetrate what characterizes science in a way that captures something historically essential about it, something that can for that reason be put to practical use by historians in their work. I believe that something like this may soon come into being; it is, moreover, something that in a vastly less formal way many historians have long used.


Scientific Practice Taxonomic Tree Cloud Chamber Sorting Mechanism Material Device 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 2.
    Both Tom Kuhn and Ian Hacking have been more than generous in providing me with their recent thoughts on the subject of kinds. Their work on it motivates and underlies my considerations here, which do not claim to provide philosophical novelty. My own interests here are more practical than theirs, in that I am concerned with the doctrine of kinds almost entirely for its usefulness in understanding the behaviour of groups of scientists, and in particular their creation and use of instruments.Google Scholar
  2. 3.
    Although I am going to be discussing something that does look as though it might be used to make normative decisions about past science, I do not think that it can be because it concerns only one aspect of science history — an important, indeed central, one to be sure, but not one that provides a Royal Road, as it were, to good science. See Note 17 for a brief discussion.Google Scholar
  3. 4.
    See Kuhn’s “The Presence of Past Science”, “The Shearman Memorial Lecture”, University College, London, 1987; “Possible Worlds in History of Science”, Proceedings of Nobel Symposium 65, ed. S. Allén, Berlin: de Gruyter, 1989; “An Historian’s Theory of Meaning”, talk to Cognitive Science Colloquium, UCLA, April 26, 1990; “The Road Since Structure”, Presidential Address to the Philosophy of Science Association, 1990.Google Scholar
  4. I. Hacking, “Working in a New World: the Taxonomic Solution”, for a volume containing essays in honour of T. S. Kuhn read at the Massachusetts Institute of Technology in May, 1990, forthcoming from MIT, ed. P. Horwich.Google Scholar
  5. 6.
    Chemical structure was not important for metals because it was precisely their failure to decompose when carrying a current that distinguished them as a class from electrolytes.Google Scholar
  6. aaaaaaaaaaaaaaFor present purposes no distinction will be made between objects such as chairs or even tachyons, and objects like waves of light, even though one commonly says that objects of the former kind may produce objects of the latter kind as effects. Such a relationship would presumably takes its place in an appropriate taxonomy, in which objects of the latter kind sort objects of the former kind.Google Scholar
  7. 8.
    A particular object might fall into kinds that belong to completely different areas of inquiry. A liquid might for example be considered solely in respect to its viscosity for determining what kind of hydrodynamic entity it is; the same liquid might be considered in respect to its effect on the phase of reflected light in considering what kind of optical object it is. These two groupings of kinds — hydrodynamic and optical — might very well have nothing at all to do with one another, in which case one might have a hydrodynamic object of type h that has reflection properties of type ra and another hydrodynamic object of the same type h that nevertheless has optical properties of type rb. (In fact, during the late nineteenth century one did have something like this, since aniline dyes are rather viscous, like most solutions, but they are nearly unique in exhibiting marked anomalous dispersion in the visible spectrum, with accompanying oddities in reflection spectra.) What one cannot have is hydrodynamic object that falls simultaneously into hydrodynamic kinds that are not nested. Of course it is certainly possible that optical and hydrodynamic behaviour might eventually be brought together (by, e.g., linking molecular structure to both hydrodynamic and to optical properties), but if this does happen then the kinds will have to be reconstructed to prevent overlap.Google Scholar
  8. Hacking pointed out to me an important, and I think related, objection to the non-overlap condition [see Hacking, “Working in a New World”]. Consider arsenic and hemlock as, respectively, kinds of minerals and kinds of vegetables. Both are also kinds of poison — and there is a forensic science of poisons — which accordingly overlaps as a category both minerals and vegetables. Hacking resolves this mundane impasse by distinguishing poisons as a category relative to us from minerals and vegetables, which are not relative to us. This problem, as well as the one concerning hydrodynamic and optical kinds, refers back to one of the central issues underlying the entire doctrine, namely to distinguish properties that are essential to an object’s being this or that kind of thing from those that are not. It seems to me that, among practising scientists, instruments are the incarnated repositories of these sorts of questions, which are for the most part rarely brought directly to the surface and addressed forthrightly. I will accordingly rely quite strongly in my detailed example on the function that instruments serve in putting objects into this or that category.Google Scholar
  9. 9.
    The tree has a deeply historical character because its topology is fabricated by practice, and its distinctions are activated by specific devices that sit at the nodes. The tree is formed by the pragmatic activity of scientists as they try to make things cohere on paper and in the laboratory — which, it is critical to say, certainly is not at all the same thing as asserting that scientists are free to do anything they want to do at any time. I will return to this point through examples below.Google Scholar
  10. 10.
    Which in taxonomic terms corresponds to the grafting of one scheme onto another, followed by the separate and perhaps autonomous development of the graft — so long as it does not disrupt the rest of the structure).Google Scholar
  11. 11.
    Which might be taxonomic changes that occur very far up the tree.Google Scholar
  12. 12.
    I distinguish here between assimilation and grafting for reasons that will be made clear through example below. The distinction corresponds to the difference between being able to understand something new in existing terms (assimilation), which leaves the taxonomy unaffected, and having to add new terms to the taxonomy (grafting).Google Scholar
  13. 13.
    Increasing the power of a telescope may reveal things not seen before, but it does not do anything qualitatively different — the kind of effect that is being examined (light used to produce an image) remains the same. Using that light in a spectroscope is indeed doing something essentially different, as is using a radio-telescope, because the effects involved are entirely novel (absorption and emission spectra or long-wave emanations). Then the several effects can even be played off against one another in a sort of romp of devices.Google Scholar
  14. 14.
    See S. Shapin and S. Schaffer, Leviathan and the Air-Pump,Princeton: Princeton University Press, 1985 for this unusual way of deploying the word “technology”. I thank Andrew Warwick for discussion about “technologies”.Google Scholar
  15. 15.
    One can envision a device that would automatically sort smudges of celestial light into the appropriate objects. Such a thing, it seems to me, would be rather like a polarimeter despite the obvious difference that crystal slices sit in the polarimeter whereas, e.g. comets travel through the heavens. Ian Hacking (“Extragalactic reality: the case of gravitational lensing”, Philosophy of Science 56 (1989), pp. 555–81) argues that the absence of the object from the laboratory, with one’s concomitant inability to manipulate it, constitutes a fundamental distinction (though Hacking’s argument aims at grounds for scientific realism, with which I am not concerned).Google Scholar
  16. No doubt the ability to do something to something and see what happens as a result may rapidly produce confidence in what the thing is (i.e. can sort it); being able only to examine what it does as a result of humanly-uncontrollable influences is not so felicitous a situation for the investigator. This is obvious: if you must find an appropriate natural object-stimulator rather than make one yourself then you cannot try to force the type of responses that are interesting when you want them. You must look around for an appropriate natural stimulus. But such stimuli often do exist, and if there are enough of them then you may still feel confidence in saying that the object is such-and-such a thing. Control of the object lies at the heart of laboratory science, but it is not perhaps essential for sorting-activity. For the latter, the issue is rather what the object is than whether it is an object at all.Google Scholar
  17. 16.
    There is an obvious caveat that the stuff which feeds into a theoretical sorting mechanism must be produced by something else, often a device, whereas material technologies may both produce and sort. This distinction is important, but it seems to me that the issue revolves rather about historical substance than philosophical absolutes, because successful theoretical technologies often become embedded in physical devices. A related difficulty concerns the kind of device that produces the stuff that a “theoretical technology” may work on. There are intricate devices that may take years of training to work; other devices may require an hour to become skilled with. Some devices may be extremely complicated in construction and yet simple in operation; others may be the reverse. It is for example much easier to push a button on a radio-rangefinder than to make a careful triangulation, even though the former device is much more intricate than the latter. If what you’re interested in is the range, then it may not make much difference how you find it. You might even pace it out. But if something funny happens when the range is subjected to your “theoretical technology” then you may start to wonder about the device that gave it to you.Google Scholar
  18. 17.
    Moreover it does not provide normative criteria that can generally be used in retrospect to assert that this rather than that scheme should (or should not) have been pursued; it depends upon the context. If the major contemporary desiderata revolved about the behaviour of certain kinds of apparatus, and about the production of new kinds, then it may indeed be possible to say that, in this single respect, taxonomy x is weak and taxonomy y is strong. It is possible to do so in the case of early nineteenth-century optics. But there are usually many other factors at work as well, and it would be deeply misleading to ignore them.Google Scholar
  19. Two other sorts of factors are worth mentioning because of their pertinence for our example. First, scheme x may be able to produce all sorts of clever new processes, but it may have trouble dealing at all with some older ones that scheme y could at least account for qualitatively. Indeed, just this will almost always be a major element in the critiques produced by y-adherents, such as that the universe should not be filled with stuff. These beliefs can be just as important to scientific work as success with devices because they often underpin the reasoning, covert as well as overt, that produces a novel taxonomy. Divorcing the taxonomy from its belief-structure may very well rob it of something essential to its subsequent vitality.Google Scholar
  20. Belief-structures and arguments over whether this or that process must be taken into account cannot be evaluated normatively, and yet they are always irremediably present in the development of science, which makes it otiose to set up comparative evaluations of schemes during periods of intense controversy — that is, during the only times when it is philosophically interesting to do so. Over time this can change, though it is usually difficult to mark a single point at which one can say that y-adherents have ceased being rational, primarily because the devices that x-adherents claim for their own have by then formed an entire universe. When a remaining y-adherent spends all of his time adapting to x-devices and generating nothing new then most community members will conclude that the time for dissent has irreversibly passed.Google Scholar
  21. 18.
    Another question that I have encountered is not so much an objection as puzzlement over what the taxonomic tree is built out of. Where in it, someone asked me, are say Maxwell’s equations? The answer is I think reasonably simple: to the extent that Maxwell’s equations are considered to specify the essential properties of fields, then to that extent they sit essentially in the devices that sort fields into this or that category. Is there a rapidly-growing but small magnetic field here? Bring to bear a device that can respond to electromotive forces and you will find out, say Maxwell’s equations. No response? Perhaps it is a small but rapidly-growing electric field. Bring in a device that is sensitive to induced magnetomotive forces, say Maxwell’s equations. The quantitative structure of a subject generally becomes part of the devices with which it sorts things as the subject stabilizes. New mathematics might destabilize relations between categories, or perhaps even the categories themselves, by calling into question the behaviour of previously-closed devices.Google Scholar
  22. 19.
    See N. R. Hanson, The Concept of the Positron, Cambridge: Cambridge University Press, 1963 and P. Galison, How Experiments End, Chicago: The University of Chicago Press, 1987, pp. 90–93.Google Scholar
  23. 20.
    Note that under these circumstances lots of things are suddenly up for grabs since changes in the laboratory taxonomy have the potential to destabilise previously-secure results. To avoid that possibility requires creating a very high-order kind that, by virtue of its height in the scheme, has no effect on equal-level categories and their subkinds. In general, one might say that a new kind always has the potential for establishing at least the kinds that contain it.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1994

Authors and Affiliations

  • Jed Z. Buchwald
    • 1
  1. 1.University of TorontoCanada

Personalised recommendations