The taxonomic schemes by which early imperial actors divided the astral sciences—tianwen 天文 “celestial patterns” vs. lü-li<"/i> 律曆 “(harmono-metric) standards & (astro-calendrical) sequencing”—are incongruent with those informing modern fields of knowledge and are irreducible to simple dichotomies of astrology vs. astronomy, space vs. time, or observation vs. calculation. In this paper, I analyze textual genres to detail the boundaries, overlap, juxtaposition, and interaction between the categories that actors self-consciously imposed upon astral knowledge from as early as the first century CE. Paralleling the division of labor within the state astronomical office, I argue that the primary distinction that we see between tianwen and li is one of skill-set and approach: one constitutes the outdoor observation, collection, and interpretation of data, and the other, indoor (textual) data processing. That the analysis of actors’ taxonomies allows us to move beyond outmoded observers’ categories like “science” and “magic” is old hat, while revealing how actors distinguished between proper and improper epistemic strategies across different fields and how contradictory positions coexisted and interacted over time. How is it, for example, that the theory and practice of omenology changed over time in response to advances made in the computational side of the astral sciences?