One of the problems with defining magic is that at various points in history, and in many cultures, magic has not only overlapped the contemporary notions of religion, science and art, but even completely encapsulated them. Greek medicine and magic were virtually the same phenomenon, and medieval optics were profoundly occultic– as examples.
Our solution is not to merely understand the logical underpinnings of how magic is supposed to have worked, but also investigate the etymology and historical context of words associated with magic. Thus, the search for the origins and nature of magic becomes not only an investigation of history and anthropology, but a study of language.
When the Greek term goés is translated into sorcerer in English and its Germanic forebears, for example, what is it about the Greek practices of this type which was so immediately recognizable to the Early English, as distinct from other terms such as “witch” and “magician” and “wizard” and their antecedents? Because doing this set up a link between the perceived practices of one category of miracle-worker in a far-off land with a more recognizable native one, and thus can teach us what was actually being done– or at least what was perceived to be done by the makers of language and translators.