Arguing with the Azande

Azande beliefs about witchcraft were rich and complex, but it does not take more than a brief summary to get to the heart of the diffi­culty I want to address.

The Azande believed that mangu—which is the word that Evans­Pritchard translated as “witchcraft”—was a substance in the bodies of witches. Mangu produced a spiritual power that could cause ill health or other misfortune to its victims, even without the conscious intention of the witch. Mangus physical manifestation was supposed to be a black substance—perhaps in the gallbladder—which could be detected at autopsy, and this substance was passed on from males to males and females to females.

Witches were supposed to do their evil in two major ways. Sometimes the “soul” of a witch traveled through the air—visible in the daytime only to other witches but at night visible to all as a flame—and devoured the “soul of the flesh” of the victim. On other occasions, witches projected “witchcraft things” into their victims, causing pain in the relevant place, but this substance could be removed by the professional healers and seers whom Evans­Pritchard called “witch-doctors.”

These witch-doctors were experts in the use of various kinds of Zande magic, but most ordinary Zande people knew many spells and rituals that were intended to help them control their world by, for example, bringing rain, curing disease, ensuring success in hunt­ing or in farming, or guaranteeing the fertility of men and women.

Witchcraft, for the Azande, was involved in the explanation of all those unfortunate happenings that do people harm.

If you asked the Azande what evidence they had for the existence of witchcraft, they would point, first, to many of the misfortunes of human life, and ask how else they could be explained. But they would also tell you that they had a number of ways of discovering more precisely how witchcraft operated: and these various ways of finding out about witchcraft they called “soroka,” which Evans­Pritchard translated as “oracles.”

The Zande used many kinds of oracles—ways of finding out what was going on in the world of spirits, in general, and witchcraft, in particular. They regarded dreams about witchcraft as oracles, for example. But the highest in the hierarchy of oracles, in terms of reli­ability, was their “poison oracle,” and they used it regularly in their attempts to discover who had bewitched them.

The oracle involved administering a special poison to young chicks; questions were put to it, and whether the chicken died determined the answer. In a typical case, an Azande man—and, in Zandeland, it always was an adult male—would administer the poison to a chicken and ask the oracle whether so-and-so had bewitched them. If the fowl died, the accusation was confirmed, but the question had now to be put the other way round, so that, on the second test, it was the fowl's survival that confirmed that there had been witchcraft.

Even given this little sketch of some Zande beliefs, you might think that you had enough to begin to persuade a reasonable Zande person that they were wrong. After all, surely on many occasions the oracle would give contradictory answers. Suppose someone put the two questions I just suggested to an oracle and the chicken died both times? Wouldn't that show the oracle was unreliable?

Unfortunately, things are not so simple. Like many traditional people, the Azande believed that there were many taboos that should be observed in every important area of their lives, and the oracle was no exception. If the operator had broken a taboo—for example, by eating certain prohibited foods—the oracle was sup­posed to lose its power. So if an oracle proved unreliable, they could say that one of the operators had broken a taboo. But they also believed that powerful witchcraft could undermine the working of the oracle; that would be another possible explanation for the fail­ure. In short, when an oracle failed, the Azande had plenty of resources within their theories to explain it.

Evans-Pritchard noticed this feature of Zande thought, and he said that the reason why they didn't notice that their oracles were unreliable was that they were able to make these explanatory moves, which he called “secondary elaborations.” Evans-Pritchard observed, “The perception of error in one mystical notion in a par­ticular situation merely proves the correctness of another and equally mystical notion.” The problem is that it is not so clear that the Zande were being unreasonable in making these secondary elaborations.

As Evans-Pritchard noticed, the system of witchcraft, oracles, and other kinds of magic formed a coherent system of mutually sup­porting beliefs.

Death is proof of witchcraft....

result by itself but is associated with empirical action that does in fact produce it—e.g. a prince gives food to attract followers and does not rely on magic alone.

And he also gave many more examples of the ways in which they can explain failures when they occur.

Consider, for the sake of comparison, what you would say if you did a simple experiment in chemistry that came out differently on two successive occasions. You would say, quite reasonably, that you had probably not done the experiment quite the same way both times. Perhaps, for example, one of your test tubes wasn't quite clean, perhaps you hadn't measured the reagents quite carefully enough, and so on. In other words, it would take systematic obser­vation, experimentation (where possible), and thought.

Now, why shouldn't an Azande say to you that your explanation here is just as much a case of defending one mystical notion—the idea of chemical reactions—in terms of another—the idea that there is an invisible quantity of some reagent in the test tube? Your theory, too, constitutes a set of “mutually supporting beliefs,” and that—far from being an argument against it—seems to be a point in its favor. Nevertheless, unless you already have some faith that the world is made of atoms and molecules that react according to defi­nite rules, there is no obvious reason why a few experiments should persuade you of this general theory. And, similarly, there is no rea­son why the failure of even a good number of experiments should make you give it up.

At this point you may recall something I said in the chapter on science.

In practice, then, we should have to do more than point to a few cases where the oracle seemed to give inconsistent results if we were to persuade a reasonable Azande person that his or her theory was wrong. What more would it take?

The answer, surely, is that it would take the collection of a lot of data on oracles; examining carefully the question whether anyone had broken a taboo; looking to see if we could find grounds to sup­port the claim that witchcraft was interfering in those cases where the oracle failed and no one had broken a taboo; checking to see that the reason one chicken died and the other did not was not that dif­ferent quantities of poison had been administered; and so on. (This is the sort of way we should set about evaluating a medical proce­dure in our own society; the medical journals suggest that establish­ing effectiveness and ineffectiveness can be quite difficult.)

Notice that we could do all this while still using the language of the Azande to describe what we were doing. We would not need to assume our own theories were correct. We could use our theories in order to see if we could construct cases where the oracle would fail, but we would still leave it up to the actual experiments to decide whether we were right.

In the long run, after much experimentation of this kind, some Azande might come to give up their theory. But there is no guaran­tee that this would happen. Just as it is always possible for us to explain away experimental results by supposing that something— though we are not sure what—went wrong, so this move is open to the Azande also.

Nothing I have suggested presupposes that it has to be we who raise doubts about Azande beliefs. Because the problem of consis­tency with the evidence can be put without presupposing that Zande theory is false, it would have been open to them to carry out these experiments. So, perhaps, if the Azande were wrong, they could have found it out for themselves.

I shall return in section 9.5 to the question of whether we should expect the Azande to come, after experiment and systematic thought, to agree with us, and not simply to assume a development of their own witchcraft theory. But it is worth spending a little time first to consider why it is unlikely that the Azande would have done either of these things if they had been left alone. For even if the Azande of Evans-Pritchard's day had started to worry about their beliefs, they would have been severely limited in their ability to the­orize about them and to carry out these sorts of experiments—not because they were not clever enough, but because they lacked at least one essential tool. For the Azande did not have writing. And, as we shall see, much of what we take to be typical of formal philos­ophy derives in large measure from the fact that formal philosophy, unlike folk philosophy, is written.