Against unifying homology concepts: Redirecting the debate.

The term "homology" is persistently polysemous, defying the expectation that extensive scientific research should yield semantic stability. A common response has been to seek a unification of various prominent definitions. This paper proposes an alternative strategy, based on the insight that scientific concepts function as tools for research: When analyzing various conceptualizations of homology, we should preserve those distinguishing features that support particular research goals. We illustrate the fruitfulness of our strategy by application to two cases. First, we revisit Lankester's celebrated evolutionary reappraisal of homology and argue that his analysis has been distorted by assimilation to modern agendas. His "homogeny" does not mean the same thing as modern evolutionary "homology," and his "homoplasy" is no mere antonym. Instead, Lankester uses both new terms to pose a question that remains strikingly relevant-how do mechanistic and historical causes of morphological resemblance interact? Second, we examine the puzzle of avian digit homology, which exemplifies disciplinary differences in homology conceptualization and assessment. Recent progress has been fueled by the development of new tools within the relevant disciplines (paleontology and developmental biology) and especially by increasing interdisciplinary cooperation. Conceptual unification has played very little role in this work, which instead seeks concrete evolutionary scenarios that integrate all the available evidence. Together these cases indicate the complex relationship between concepts and other tools in homology research.