IN THE DAYS before the rise of the social sciences, the persons who wrote about magic were usually practitioners or prosecutors, and what they wrote were practical manuals. From ancient Egypt and the classical world to Aleister Crowley and George Gurdjieff there have been documents that told you how to attain great powers, or do your enemy in, or prevent him from doing you in.

But when anthropology and sociology arose, it soon became evident that magic was not necessarily something practiced by evil people or overambitious Faustians (or fraud, conscious or half-conscious). It could be an important part of general human behavior. It was an attempt to control the uncontrollable, often for the sake of safety. Instead of being outlaw power, magic was often, more importantly, a way to help the universe to do things correctly. For example, the Polish-British anthropologist Bronislaw Malinowski showed, in a series of classic studies, that magic permeated and sustained almost all aspects of life in the culture of the Trobriand Islanders.

Questions and problems soon arose with this modern recording and theorizing. Does one distinguish magic from the technique of religion, which often tries to compel the gods? Are magic and religion the same, or opposites? Are they phenomena on a continuum, with magic the shady elder sibling? And when studies shift from primitive to civilized cultures, is one justified in calling magic a nonrational act aimed at a goal? One can argue these questions interminably, as the theorists do.

An important school of thought that concerned itself with magic in this sense was that of the French sociologist Emile Durkheim. He and his followers, notably Marcel Mauss, decided that magic and religion were connected. According to them, religion is a projection of the social organization, and the sort of social and political world one lives in will determine the form of one's religion. Religion thereupon gives birth to magic, although the exact details of this parturition vary according to individual theorists.

O'Keefe's Stolen Lighting is based frankly on the work of Durkheim and Mauss, and attempts to codify it more rigorously, to expand it into a broader synthesis, and to bring it up to date. O'Keefe works through the classical position thoroughly, making small changes, smoothing out inconsistencies, and generally shoring up the system. Then, from the early psychoanalytical literature (Freud, Reik, R,oheim) he borrows the idea that magic is intended to be the protection of the individual against society (the superego) and combines this concept with Durkheim-Mauss. O'Keefe also places great importance on the role of magic in creating the modern individual, as opposed to the earlier group man of French sociological theory.

Other elements that are brought into the scheme include the work of Mircea Eliade (heavily drawn on), Wittgenstein, Max Weber, and a host of other authorities on society, the individual, and matters in between, even to modern politics. Unfortunately, O'Keefe is sometimes undiscriminating, equating, for example, Castaneda with the great names of the social sciences.

All this is very ambitious, an enormous amount of work. Creating a synthesis out of this hodgepodge of seemingly disparate material is an achievement, and O'Keefe manages it ingeniously. On the positive side his restatements of French social theory are useful in picking up items that may not be generally known in America, and his summaries of classical material are usually apt. He has many insights and makes many cross-cultural linkages that are thought-provoking. His own ideas (the role of magic in creating the individual, feedback mechanisms, and the concept of magic "in the weak sense") fit well onto the Durkheimian torso.

On the negative side, O'Keefe is very doctrinary, and it soon becomes obvious that despite the ambitious statements of purpose and elaborate footnotes, his working area is really much smaller than it seems to be. References tend to be repeated, and areas that might have been or should have been included are skimped or slid over lightly. For example, in anthropology, which should be the heart of any systematic theory of origins (if one cares about observed data and is not obsessed with aerial theory), only Evans Pritchard is consistently used. There is very little reference to any other school of thought, or other collections of data, and the Azande of Evans Pritchard are taken to be valid for the entire world of primitive and civilized man. Most of the other anthropological references are to semipopular compilations. The same narrowness is to be seen in psychology. Surely others than Freud, Reik, and R,oheim have worked here.

Two areas of importance are underdeveloped. These are history of religions, where the author is obviously not at home, and European thaumaturgy. The author's concern is theory, it is true, but if he could desert theory long enough to follow attenuated trails into politics, he surely could have treated religion and thaumaturgy more thoroughly.

Social scientists will probably find this updating and expansion of Durkheim-Mauss thought-provoking, although they will ultimately judge it, I would guess, much as they happen to judge R,oheim and Durkheim. R,oheim has never been taken very seriously by most American anthropologists and historians of religion. Durkheim has certainly been taken seriously, but many criticisms have been made of his work and his successors': stretching definitions to the point of loss of meaning; ripping material out of context; and being unhistorical -- despite perpetual talk about origins.

O'Keefe's presentation, however, creates difficulties, and I wish that he had imitated his French prototypes more in this respect. Levy-Bruhl, for example, may write nonsense, but one can still admire his deftness. But in Stolen Lightning the development is very repetitive; the style often very turgid; and the book is much too long for what it has to say. All this was not unavoidable, for O'Keefe, as many examples within the text show, can turn a very neat, amusing phrase when he wants to.

His publisher has not helped him, for the book cries out for an index. The bad proofreading will leave the reader wondering what happened on page 448, and is likely to send the reader searching vainly for such magical names as Van Gennup, Kluckholn, Chandrasheka, Goldenweisser, and Minayana Buddhism.