This article contains a series of reflections on the nature of the lexicon in FDG inspired in large measure by the preceding articles. We start by considering how the lexicon relates to the Conceptual Component, arguing that lexemes do not label units of conceptualization but rather are associated with experientially based beliefs about their appropriate use. In our view, the Conceptual Component first develops a Message, which then influences the choice of a frame in the Grammatical Component into which appropriate lexemes are inserted. Lexemes are thus not inherently associated with frames, as was proposed in earlier work. Instead, they are marked with numerical indicators for the set of frames with which they are compatible, with coercion allowing one-off extensions of that frameset. It is a further consequence of our position that lexemes come with neither meaning definitions nor selection restrictions. We adopt Keizer's notion of partially instantiated frames to account for idiomatic expressions. We then turn to parts-of-speech, as they apply to lexemes in various language types. Lexemes are distinguished from Words: for example, the single class of Contentives in the Esperanto lexicon corresponds to Verb Words, Noun Words, etc. in morphosyntax. This leads to a discussion of derivation and compounding, where it is shown that two types of derivation are to be distinguished in FDG, lexical derivation, which uses lexical primitives ($) as its input, and grammatical derivation, which takes place after insertion of a lexeme into its frame. Three major types of compounding are differentiated and exemplified from English and Dutch: predicate-argument, nucleus-modifier, and conjunct-conjunct compounds, each of which can be either endocentric or exocentric. Turning finally to the difficulty of drawing a sharp distinction between the lexicon and the grammar, we apply Keizer's (2007) distinctions among primary and secondary lexical elements and primary and secondary grammatical elements, showing how findings from various of the preceding articles can be interpreted in this light.