Natural kind terms like 'water,' 'gold,' and 'tiger' refer to categories defined by underlying properties rather than apparent properties. The meaning and reference of natural kind terms depends on scientific discoveries about the kinds' essences, not on stereotype or appearance.
From your study of Putnam's semantic externalism you know that meaning "ain't in the head" — what our words refer to is determined partly by facts about the world and our environment, not purely by our internal mental states or descriptions. From your study of Kripke's causal theory of naming you know that proper names are rigid designators: they pick out the same individual across all possible worlds, with reference fixed by an initial dubbing and transmitted through a causal-historical chain. Natural kind terms extend this apparatus from individuals to kinds. When your ancestors first pointed at water and said "water," they fixed the reference of that term not to a description (the clear, drinkable liquid) but to whatever natural kind instantiated the stuff in front of them — which turned out to be H₂O.
The key consequence is that essential properties of natural kinds are discovered, not stipulated. Before chemistry, no one knew that water was H₂O. But H₂O is not merely how we recognize water — it is what water fundamentally is. In every possible world where something is water, it is H₂O. A possible world with XYZ (another substance, structurally different) filling the lakes and running from taps would have something that looks and behaves like water but is not water. This is Putnam's Twin Earth thought experiment: the inhabitants of Twin Earth who use "water" to refer to XYZ are talking about a different substance, even though their inner lives — perceptions, beliefs, descriptions — are indistinguishable from ours. Reference tracks the underlying nature; the rest is stereotype.
This means natural kind terms support a distinctive kind of modal claim: a posteriori necessities. "Water is H₂O" is necessary — true in every possible world — but it was only discovered through empirical investigation, not through conceptual analysis. Similarly, "Gold has atomic number 79" is not a definition that exhausts what we mean by "gold"; it is a scientific discovery that reveals gold's essence. Once discovered, the identity holds necessarily. Compare this with a description theory of meaning, on which "water" means "the clear, drinkable liquid" — then "Water is H₂O" would be contingent (there could have been worlds with clear, drinkable XYZ), and the term would shift reference in different possible worlds. Kripke and Putnam argue the description theory gets this wrong.
The scope extends beyond chemistry. Biological kinds like "tiger" and "human" are natural kinds whose essential properties are determined by biology — genetic structure, evolutionary lineage, developmental biology — not by the appearance or behavioral stereotype we use to identify them. A natural kind in this sense is any category that "carves nature at its joints": a real division in the world that supports inductive inference, causal explanation, and scientific law. Not every grammatically natural predicate picks out a natural kind — "jade," as noted, picks out two different minerals with similar appearances; "is an electron" plausibly does pick out a natural kind. Distinguishing natural kinds from mere categories or artificial kinds, and understanding what makes a property essential to a kind rather than merely typical of it, sits at the intersection of philosophy of language, metaphysics, and philosophy of science.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.