The term mimicry was introduced to biology in 1862 by Henry Walter Bates in his evolutionary explanation of deceptive communication in nature, based on a three-part interaction system of a mimicked organism or object (called model), a mimicking organism (called mimic), and one or more organisms as selecting agents. Bates gave two incongruous definitions of mimicry: one from the viewpoint of a natural agent that selects for, and in consequence is deceived by, the close resemblance of a toxic model's warning signal and the similar appearance of a palatable mimic, and another one from the viewpoint of a human taxonomist who under an evolutionary aspect focuses on convergent resemblance between model and mimic. Later definitions of Müllerian (F. Müller), arithmetic (A. Wallace) and social (M. Moynihan) mimicry abolish deception in the natural selecting agent, rely on the convergence criterion alone, fuse the roles of model and mimic but have to accept a mix of homologous and convergent resemblance amongst them for a functional explanation. The definition of vocal mimicry (E. Armstrong) refers to a learned resemblance between mimic and heterospecific model by character duplication (no convergence), so far without known (deceived or not deceived) natural selecting agents. It excludes Batesian vocal mimicry. The functional ethological understanding of mimicry as a tripartite communication system (W. Wickler) is consistent with Bates' concept and accepts deception as key element of Batesian mimicry beyond homologous and convergent resemblances. Deception is seen as caused by the divergence between a sign and its meaning for the natural selecting agent. This understanding covers mimicry in all behaviour domains, provides a generally applicable definition of mimic and model so far missing in any mimicry concept, and it distinguishes – still in line with Henry Bates – cultural from genetically determined model-mimic-resemblance; this applies to vocal mimicry in particular. Convergently evolved model-mimic-resemblance, not essential in Batesian mimicry but mandatory for its alternatives, marks a fundamental distinction between Batesian mimicry (including Mimesis) and all other conceptualized mimicries and accounts for the non-existence of a unified meaning of the term mimicry. However, character convergence does not help to explain the mere existence of mimicry phenomena and is irrelevant for their permanence in nature. I therefore propose to remove the convergence argument from any mimicry definition.