Mary Douglas: Purity and Danger (1966): Zande Poison Oracle. This page looks at Douglas's references to the Zande poison oracle in Chapter 5.

The Zande Poison Oracle (Chapter 5)

Hits since 28-01-12:    10681

Links to Other Chapters:
Purity and Dangerch 1ch 2 |  3 |  ch 4ch 5 |  6 |  7 |  8 |  9 |  ch 10

Extended Comments Pages:
Ndembu Doctor (ch 4)Zande Poison Oracle (ch 5)

Douglas refers repeatedly to the Zande poison oracle in Purity and Danger, Chapter 5. Indeed, her key argument in this chapter on primitive thought relates to this ethnographic example. However, comparison between Douglas's account of this example and her stated source shows significant discrepancies between the two.

This is particularly surprising. Douglas gives as her source E.E.Evans-Pritchard: Witchcraft, Oracles and Magic among the Azande [1937]. For one thing, Evans-Pritchard taught Douglas. For another thing, she later wrote a book on him: Evans-Pritchard [1980]. And again, she tells us in the Preface to her Leviticus as Literature [1999]:

for students of my generation the main text was Evans-Pritchard's Witchcraft, Oracles and Magic (1937) [p v].

Those who are unhappy with what follows are invited to do a little spadework themselves. Check what Douglas says and what Evans-Pritchard says: for both I provide quotations with page references to the full edition of the latter's book, not to the abridged version.

I start with an intrinsically very minor point and work through to the most important.

Azande, Zande

On p 90 Douglas uses Azande as a singular noun in an Azande. But Evans-Pritchard has made the distinction, op.cit. p 13, [t]he Azande (singular, Zande), and applied it throughout the 550 or so pages of his book. Douglas's error suggests the possibility of her not having checked the Evans-Pritchard source while writing Purity and Danger.

Any reasonably attentive person looking at the source text could hardly have missed the distinction Zande (singular noun) ~ Azande (plural noun). Zande is also used throughout as an adjective. The plural noun prefix A- is also seen in references in the source to the component tribes of the Azande. What is more, the territory of the Azande was called Zandeland, a term Douglas herself uses on p 87, where on the very next line she uses the term Azande. How can she not have been aware that there was a linguistic issue here.


A more important error by Douglas had already appeared in her first reference to the Zande poison oracle. She says:

The Azande themselves brew their poison from bark. [p 87].

However, according to her named source, the poison was neither brewed or derived from bark.

On p 260 of Evans-Pritchard's book, we learn that

The poison used is a red powder manufactured from a forest creeper and mixed with water to a paste.

The mixing of the powder with water is done during the actual ceremony, as we learn on p 295. There is absolutely no question of brewing.

We also learn on that page 260 that previous writers on the Azande had suggested that the poison came from the leaves of a plant or bush, the root of a shrub, the root of a liana. The third suggestion is correct. Not even the erroneous suggestions mention bark, so Douglas's mistake cannot have come from one of them.

Evans-Pritchard has a 10 page chapter on Gathering of Poison. Here we are repeatedly told that the poison is derived from the roots of a creeper. Thus on p 277 we read:

The roots of the poison creeper are dug up, cleaned, and scraped to the inner wood. These scrapings, which are like a red powder, are then placed in the sun for three days to dry.

Thus, assuming, as we must, that Evans-Pritchard got his facts right, we have moved from a linguistic error to one of fact. Still, in itself, saying that the poison was brewed from bark rather than a powdered from a root and mixed with water seems unimportant. However, given that the correct information is made so clear in the book, you must get to wondering if there may be more serious errors to contend with in Douglas's account. Closer examination reveals that indeed there are.

Talking to the poison

In the same passage Douglas suggests that there is something special in the Azande addressing their poison oracle:

[its] limited response to questioning radically modifies its thing-status in the Azande universe. It is not an ordinary poison ... [pp 87-88]

Douglas's remark makes it appear that the Azande had ordinary poisons with thing-status, but that the oracle poison was not one of them. But this construction is not sustainable in terms of Evans-Pritchard's text.

Firstly, Evans-Pritchard indicates that the Azande did not have any conception of poisons as such, making no distinction between poisons and medicines: see op.cit pp 315-318. Therefore to suggest that the oracle poison was not any ordinary poison for the Azande makes no sense: in their understanding, they didn't have poisons, ordinary or otherwise. The label poison oracle is not a literal translation of some indigenous term, but merely a rough approximation we should use with extreme caution.

Secondly, Evans-Pritchard tells us that the Azande spoke to their medicines generally, and to other things besides, such as to the stones used in rain prevention ceremonies. For example, he asks:

Do addresses to medicines and attributions of psychic action to them justify us in concluding that Azande personify them? [op.cit. p 465]

He suggests not, remarking a little later:

Magical action is sui generis, and is not explained by the presence of spirits residing in medicines nor by the attribution to them of personality and will. [op.cit. p 465]

Consequently, the poison oracle was just one of lots of things that the Azande spoke to without regarding them as persons: and got a reply from in the form of something that happened or didn't happen. The poison oracle was not an escapee from thing-status, as Douglas suggests, but a member in good standing of things-we-talk-to-and-get-a-yes/no-response-from status.

The significance of this error by Douglas becomes apparent when we consider her key argument about intellectual consistency.

Intellectual consistency

On p 90, Douglas tells us:

The fact that [a Zande] addresses the poison oracle in words does not imply any confusion whatever in his mind between things and persons. It merely means that he is not striving for intellectual consistency and that in this field symbolic action seems appropriate. He can express the situation as he sees it by speech and mime, and these ritual elements have become incorporated into a technique ... like programming a problem through a computer.

Two points arise here. Firstly, Douglas's claim that the Azande were not looking for intellectual consistency when they addressed the poison oracle makes sense only if they had the same intellectual categories as ourselves. But if we consider Evans-Pritchard's evidence rather than rely on Douglas, we find they did not.

For we have seen in the previous section on Talking to the poison that the Azande did not differentiate between medicines and poisons and that they talked to medicines generally and to other things besides. This suggested that there was a Zande category of things-we-talk-to-and-get-a-yes/no-response-from. The existence of such a category would in turn suggest that in their terms the Azande were being perfectly consistent intellectually when they spoke to the poison oracle.

So Douglas's key argument about primitivethinkers not being primarily concerned with intellectual consistency is not demonstrable in terms of her chosen instance.


Secondly, (re the p 90 passage most recently quoted), Douglas suggests that the Azande had ritualised consultation of the poison oracle for themselves. However, this is extremely unlikely because, if Evans-Pritchard is correct, they did not originate the poison oracle themselves, but had borrowed it relatively recently from another people.

Evans-Pritchard says:

it may be assumed that the Ambomu [the ruling tribe of the Azande] ... learnt the oracle from some other people and imposed its use on the tribes they conquered. [p 271]

He then gives convincing reasons for coming to this conclusion.

It seems well-nigh impossible that the poison oracle was introduced without a ready-made technique for its use being part of the package. For as well as his chapters on the poison oracle, Evans-Pritchard has a 30 odd page chapter on Other Zande Oracles, in which we learn about three other oracles, employing respectively termites, three sticks and the rubbing board.

We are told that before the poison oracle was introduced, the Azande regarded their termite oracle as the most important. This too involved addressing to the oracle questions that it answered positively or negatively, this time through the behaviour of termites. Therefore, we must suppose that when they were given the poison oracle, the Azande simply transferred to its consultation the questioning technique they were already familiar with from working with their termite oracle.

Failing this, we must suppose that the poison oracle came with instructions on its use, that the princes who imposed it did not do so without demonstrating how it was to be used. Whatever the case, we can be sure that Douglas's scenario of the Azande somehow spontaneously generating a ritual to go with their oracular administration of a poison to fowls is totally mistaken.

Zande oracle poison

Douglas first refers to the Zande poison oracle in the following:

the poison used for the oracular detection of witches in Zandeland [p 87].

But to write thus massively understates the significance of the poison oracle for the Azande.

The Evans-Pritchard text has over 90 pages, pp 258 - 351, specifically devoted to the poison oracle. The author makes it clear that this oracle was the basis of the pre-colonial legal system, that direct access to it was related to social status, that the exclusion of women from access to it was related to their subjugation, that access to it enabled established men to subordinate the young men, that the Azande consulted this oracle regarding all the important decisions in their lives and in many cases regarding lesser matters, that the Azande were prepared to run grave personal risks to obtain the poison, that consultation of the oracle involved deviousness and secrecy as well as taboos ...

In fact, on Evans-Pritchard's evidence, we may conclude that in the period just prior to colonialisation the maintenance of the whole Zande political and social order had depended on the poison oracle.

Witches were certainly a constant and absolutely fundamental concern for the Azande, with the poison oracle as the most important means of detecting them. But to refer to it as Douglas does just will not do.

Poison ordeal

Douglas had previously referred to the Zande poison oracle in a 1963 article, Techniques of sorcery control in Central Africa, which later became Chapter 4 of her book, Implicit Meanings [1975/1999]. The reference came in Douglas's discussion of the use of poison ordeals for the detection of witches, on p 64 in the book.

We get the impression in the paragraph Douglas devotes to it that the Zande practice, like others discussed in the piece, was an ordeal, that is, a poison was administered directly to the person accused of witchcraft. But of course this does not correspond to what she was later going to tell us in Purity and Danger: that the poison was administered to fowl.

Now, as in the 1966 book, in her 1963 article Douglas cites Evans-Pritchard, likewise without pages references. In the Evans-Pritchard index we do find re the poison oracle,

formerly sometimes drunk by human beings, or administered to boy captives, 309-12 [p 553].

This gives us a mere 2½ pages on the poison oracle as an ordeal. Evans-Pritchard says,

in the past people sometimes, though very rarely, actually drank poison themselves [p 309]


Oracle poison was also occasionally administered to boy captives in important cases involving princes [p 310].

We gather that this exceptional use of the poison as an ordeal had happened before colonialisation, that is, pre-1905.

However, overall, Douglas's paragraph in her 1963 article just cannot be justified in terms of her stated source.     [Back to Article]


(c) John C Durham, 2006

Links to Other Chapters:
Purity and Dangerch 1ch 2 |  3 |  ch 4ch 5 |  6 |  7 |  8 |  9 |  ch 10

Extended Comments Pages:
Ndembu Doctor (ch 4)Zande Poison Oracle (ch 5)