As feature terms are the basic building blocks of UB-grammars, we start by discussing how they can be implemented in Oz. We then turn to the lexicon. Because feature-terms contain a lot more information than atoms for categories, the emergence of unification-based theories of grammar went hand-in-hand with a stark lexicalisation trend: information that earlier was part of the grammar proper (i.e. the syntactic rules and filters) got integrated in the lexicon. Practically this means that the lexicon contains a lot of information. To avoid redundancies and support maintenance, it therefore becomes important to structure the lexicon in some way. This is the object of the second section.