Chapter 3 articulates a proposal about how syntax encodes the fundamental parameters of noun lexical semantics: entity type, part structure, atomicity, and numerosity, represented at different syntactic loci. It is claimed that the naming (not describing) of an entity type is what makes a noun a noun—more precisely, what turns a root into a noun. An ‘entity type’ is understood as an e-type element in the speaker’s ontology, not yet determined as denoting over kinds or their object instances. Higher projections turn this into a predicate, and further grammatical specifications may restrict the part structure of its denotation domain. Whether the reference is to kinds or to instances depends jointly on all these specifications. In particular, plurality in some cases (waters) enforces an instance reading where the partition it determines is based on spatiotemporally situated parts, in the absence of other criteria for individuality. Besides determining what types of plurals may not license a kind reading, this view of the division of reference analyses fake mass nouns (furniture) as a count structure inside a mass shell, relating their semantics to their morphology. The same analytical framework, with no additional stipulations, accounts for the properties of non-countable plural nouns. Finally, a type of mixed-gender DP which cannot be pluralized suggests that the attribution of a gender value can strongly constrain the determination of part structure.