In this post, I argued that, contra Kripke, when the baptismal origin of a natural kind term has none of the properties we associate with that term, the term no longer refers to a kind that includes its baptismal origin. I then promised that I argue that, when objects not “of the same kind” (in a realist sense) as the baptismal origin of a natural kind term have all of the properties associated with a kind term, the kind term does (or at least, Kripke didn’t show that it doesn’t) include those objects in its extension. Here’s where I’ll do that.
Kripke’s Original Argument
Kripke’s example is as follows: Consider English as it was spoken pre-Lavoisier. The term “water,” at that point, clearly did not have associated with it the property of having the chemical formula H2O. The set of properties associated with it, at most, involved those properties accessible to people at the time. Now, suppose there was another compound (which Kripke calls XYZ), such that it had all of the properties of water that were accessible to people before the start of modern chemistry.
Kripke takes it as obvious (and I’m pretty much inclined to agree) that XYZ shouldn’t count as water. But should it count as water as the term was used before the start of modern chemistry? Kripke thinks that’s pretty obvious, too. Of course, he claims, the meaning of “water” hasn’t actually changed since the 17th century. So XYZ couldn’t have counted as “water” then, either, even though it had all the properties that were associated with the term.
But the claim that the meaning of “water” hasn’t changed since the 17th century is a much bigger leap than Kripke supposes. (1970s-era Putnam had an actual argument for this claim, which I’ll address when I post about said Putnam’s various arguments for natural kinds realism. For now, I’ll just present an argument against it.) For one thing, consider the following two counterfactual situations.
- Situation 1: In this situation, which, I imagine, is the situation Kripke is imagining, the identity of water as H2O was first claimed, as in actuality (well, approximately) by Henry Cavendish in about 1781. Then, relatively recently (in, let’s say, 1973), the substance XYZ was discovered. I think that, in this situation, it’s pretty clear that the vast majority of people would agree (as Kripke and I in fact do) with the statement “XYZ is not water.”
- Situation 2: In this situation, as above, the identity of water as H2O was first claimed by Cavendish in 1781. Then, in 1782, XYZ is discovered. In this situation, I’m pretty sure that the general reaction would have been, “It turns out Cavendish was wrong! Not all water is H2O; water actually comes in two forms: H2O and XYZ,” and that belief would hold up to the present day.
So in Situation 1, the term “water” would exclude XYZ; in Situation 2, it would include “XYZ.” The term would have different meanings in the two situations!
What explains this difference? Well, it’s something rather like the following: In Situation 1, the identity “water = H2O ” had a considerable amount of time to hang around in the public consciousness, to the point that the actual meaning of water shifted to exclude anything but H2O. In Situation 2, the identity never really got a chance to take hold; XYZ was discovered at a time when the meaning of water was not yet tied to its chemical composition.
Note that that explanation actually does involve a shift in the meaning of the term water between the pre-chemistry days and 1973. So “water” really would have included XYZ in its extension (had XYZ existed), before the time when water’s chemical composition was a property associated with it.