Although visuals have been co-deployed with text in specialized dictionaries as far back as the European Renaissance, the interaction of both representational modalities is relatively under-researched. As a consequence, available knowledge is relatively limited with respect to the kinds of visuals employed in specialized dictionaries, the kinds of definiendum that elicit specific types of visuals, the functions of visuals relative to text in definitions, and the association between particular visuals and visual-text functions. This study sheds light on these questions from the perspective of specialized dictionaries in two fields (Biology and Mechanical Engineering). Significantly, the study underscores how the ontology or nature of Biology and Mechanical Engineering appear to determine both the selections made of visual types and the dominant text-visual relationships. The study further makes a contribution to cleaning up the Augean clutter that is the terminology of visuals.