Definitions of the term gene typically superimpose molecular genetics onto Mendelism. What emerges are persistent attempts to regard the gene as a unit of structure and/or function, language that creates multiple meanings for the term and fails to acknowledge the diversity of gene architecture. I argue that coherence at the molecular level requires abandonment of the classical unit concept and recognition that a gene is constructed from an assemblage of domains. Hence, a domain set (1) conforms more closely to empirical (...) evidence for genetic organization of DNA regions capable of transcription and (2) has ontological properties lacking in the traditional unit definition. (shrink)