This paper explores the semantic features of the so-called “neuter” in the demonstrative system of Modern Spanish and presents a diachronic analysis of the semantic as well as morphophonological changes which have taken place from Latin to Spanish. We show that the semantic features commonly assumed as being associated with gender (or classification), and more particularly, with the “neuter”, (e.g., [(in)animate]), are not able to capture the semantic difference between “neuter” and feminine / masculine, neither in Latin nor in Spanish. For Latin, we argue that the relevant difference for this classification is based on the fact that the neuter is underspecified for a feature [discrete] (vs. presence of the feature [discrete] for masculine / feminine) and elaborate a feature geometry for demonstratives which captures this fact. As the opposition between [discrete] and [non-discrete] is strictly speaking not a matter of classification, i.e., one of gender, but a specification of the operation of individuation, this leads ultimately to the reduction of the Latin classification-node in the geometry and to the Modern Spanish feature geometry. There, the absence of [individuation] results in what mistakenly is called “neuter”, i.e., in expressions whose referent does not have to be individuated (vs. feminine/masculine, with a specified feature [individuation]). We present a detailed morphophonological analysis of the Latin pronominal morphology, which is based on a realizational approach and which uses case feature decomposition and the morphological schemes proposed by Wiese (Zur lateinischen Nominalflexion: Die Form-Funktions-Beziehung, IDS Mannheim, 2003). This analysis leads us to the conclusion that there are no specific morphological schemes for the neuter in the Latin demonstratives (Vd, i.e., /ud/, being the mere default in our analysis). The most intriguing fact here is the absence of genuine neuter endings in the plural, both in Latin and in Modern Spanish. We do not consider this a mere coincidence, but as a hint at the fundamental semantic change mentioned above, the feature [individuation] being superordinate to [group] (for plural). Finally, we describe the important morphological change in the pronominal system from Latin to Spanish, i.e., the reduction from a five-case-system to a two case-system, in detail and argue, based on the notions of underspecification and default, that Spanish /o/ just preserves the default status of Latin /ud/, being thus no “neuter” gender marker at all.