The production of effects in the world by actions, often ritualized, whose source of power is not open to observation; or by words, especially by incantation: chants of formulae, which may sound nonsensical to the outsider, may summon the relevant power, or may themselves affect the consequence: they may also be apotropaic. Attempts to define the relation of magic to religion have formed part of the modern study of religion since its inception at the end of the 19th century. J. G. Frazer argued for a complete separation of the two: religion is oriented towards transcendental beings (gods, spirits, etc.), but the magician "supplicates no higher power: he sues the favor of no fickle and wayward being.... He can wield his power only as he strictly conforms to the rules of his art, or to what may be called the laws of nature by him"
Thus the worker of magic is a would-be scientist who misunderstood the true nature of causality. From Frazer derives the division of magic into major categories, of imitative magic (achieving results through mimicry), contagious magic (using materials which have been in contact with the object of the magic), and sympathetic magic (using items which symbolize the intended object).
E. Durkheim accepted the radical distinction between magic and religion, claiming that religion is a communal matter (indeed, it is society being itself in externalized form), whereas magic is an affair between a practitioner and a client: the magician has clients but no congregation. This division was contested by M. Mauss (A General Theory of Magic, 1902), who insisted that magic is a social phenomenon, making use of available power in a variety of different ways. In other words, he went behind the conduits through which this power flowed to identify the nature of that power. This he thought he had located in the Melanesian concept of mana, which R. H. Codrington (The Melanesians, 1891) had introduced: "Mana is power, par excellence, the genuine effectiveness of things which corroborates their practical actions without annihilating them. This is what causes the net to bring in a good catch, makes the house solid and keeps the canoe sailing smoothly: in the farms it is fertility, in medicine it is either health or death."
This view of magic as pervasive power was the exact opposite of the view of Bronislav Malinowski (1884-1942), an anthropologist who based his arguments on his time in the Trobriand Islands. Since it is obvious that magic is not always or automatically successful, he maintained that magic must be understood primarily as a matter of psychology: magic operates in areas where knowledge or technology is wanting, and by isolating hidden forces at work, it thereby offers psychological relief -- such-and-such a situation is beyond human competence. This was in strong contrast to the view of E. Evans-Pritchard (Witchcraft, Magic and Oracles among the Azande, 1937), who held that magic belonged to an interactive world in which it is possible to ask questions which a Westerner would not ask. Thus the Azande are interested in 'cause-and-effect', but also ask why events have happened to one person rather than another: magic is a means of interrogation as well as of finding answers.
More recently, this view has been developed further, seeing magic as embedded in religion, where it acts as an organization of context and meaning. Thus for Mary Douglas (Purity and Danger, 1966), magic offers the framing of experience in a local context; and for Susanne Langer (Philosophy in a New Key, 1964) "magic is not a method but a language: it is part and parcel of that greater phenomenon, ritual, which is the language of religion; ritual is a symbolic transformation of experiences that no other medium can adequately express."
In this perspective, magic offers the transformation of circumstances without guaranteeing effects: one consequence or its opposite will still be a demonstration that magic 'works', because it confirms the entire context in which a person lives.