In this chapter, we discuss various forms of formational (phonological/phonetic) encoding of signs, that is transcription (e.g. HamNoSys), tagging (e.g. SignStream), and lexical databases (e.g. SignPhon). We point out that different goals are served by these systems, and how differences in the set of categories used follow from these goals. In all cases the designer of a system is faced with what we call "the database paradox". This term refers to the problem that the system is used to find out which linguistic distinctions are relevant, whereas the design of the system already requires such knowledge. For many tagging systems (such as SignStream) this problem is not so acute, since they leave it up to the user to define the necessary categories. Thus, the database paradox is transferred to the user in case the system is used for phonetic/phonological research. For an overview of different notation systems, see Miller & Radutzky (this volume). We describe in some detail the system that we developed (SignPhon), and discuss how it deals with the database paradox. Finally, bearing in mind the different functions of SignPhon and HamNoSys, we compare both systems with respect to the formational categories that are used.