出事之后总有人问「为什么是我」。医生把病理讲得清清楚楚，你还是会在心里追一句：那为什么偏偏是我、偏偏这一年。这句追问在今天通常没人接——它被归到情绪，归到运气，然后就没下文了。世界上有过很多套体系专门接这句话，巫术是被研究得最透的一套。
反直觉的地方在于：它并不否认白蚁，也不代替白蚁，它补的是另一个位置。这一页拆三件事：这套说法到底回答了什么、它答错时凭什么还站得住、以及为什么被指控的人翻来覆去总是那几个。
1926 到 1930 年间，一个叫 Evans-Pritchard 的英国人几次去当时英埃共管的苏丹南部，跟 Zande（阿赞德）人住在一起。这群人今天分布在南苏丹、刚果民主共和国和中非共和国，是活着的人，不是标本。他做田野那几年当地在殖民行政之下，他能走那些路、住那些村子，靠的也是那套行政——观察可以是准的，同时它被采集的位置得摆上桌面。
他记下的第一件事就把「他们不懂因果」这个预设顶翻了。Zande 人的谷仓是架空的，底下阴凉，中午常有人坐在里面聊天。白蚁蛀空木支柱，谷仓塌下来砸到人。问他们谷仓为什么塌，回答是白蚁蛀的——跟你我一模一样。可你接着说「那就是碰巧」，他们不接受。白蚁解释了谷仓为什么会塌，解释不了它为什么在这几个人坐在下面的那一刻塌。两条链子各自都清楚，它们为什么在此时此地交叉，没有人管。巫术管的正是这一段。
Zande 人打猎有个规矩：一头兽倒下，第一支矛和第二支矛都算数，肉归两人分。Evans-Pritchard 借了这个说法——巫术是第二支矛，不取代白蚁，叠在白蚁上面。
他们说的巫术（mangu）是一样具体的东西：长在人肚子里、随血脉传下来、人死后剖开可以验。带着它的人自己不一定知道，也不一定存心害谁。这跟另一样东西要分开：ngua，人特意配出来用的巫药与法术，那是行为，不是体质。前者算倒霉，后者才是罪。指控一个人是巫，不等于说他坏，是说他身上那样东西碰巧朝你发作了。
证据强度：这是 1930 年代的单点民族志，出自一个人的观察，没有第三方数据可比对；它经得起讲，是因为记录够细、后来有人回去接着做。
EN Between 1926 and 1930 Evans-Pritchard worked among the Zande of what was then the Anglo-Egyptian Sudan; their descendants live today in South Sudan, the Democratic Republic of the Congo and the Central African Republic. Ask why a granary collapsed and the answer is termites — the same answer you would give. But say the people sitting beneath it were merely unlucky and the answer is refused: termites explain why a granary falls, not why it fell at the moment those particular people were under it. Two causal chains are each perfectly clear, and nothing explains their intersection. Witchcraft, in his phrase borrowed from Zande hunting, is the second spear: it does not replace the termites, it is laid on top of them. Mangu, the witchcraft substance, is concrete — carried in the belly, inherited, detectable at autopsy — and its bearer need not know he has it or intend harm, which separates it sharply from ngua, deliberately prepared magic. Evidence strength: a single-site ethnography from the 1930s, gathered under colonial administration.
要判断是谁干的，Zande 人问神谕。最有分量的一种叫 benge：林中藤本磨成的红色毒粉，和水调成糊灌进小鸡嘴里，问题要问成「如果是某某，就让这只鸡死」。有个细节值得停一下：要紧的事必须问两轮，第二轮把问题反过来问——第一轮鸡死了，第二轮就必须鸡活，答案才作数。这不是没有检验，是一套内建的一致性检查。在英国人接管之前，神谕的结论在王庭上有法律分量。
那么检查没通过怎么办？Evans-Pritchard 把 Zande 人给出的理由收成一串，大致八类：毒采错了品种、有人犯了禁忌、有巫术在旁边干扰、采毒那片林子的主人发了怒、毒放得太久、鬼魂发怒、有人下了法术、操作不当。每一条单独看都不荒唐。合起来的效果是：神谕永远不会被推翻，而且每一次「原来是犯了禁忌」都让人更确信神谕是准的——失败被回收成了证据。他写过一句常被引用的话，大意是：网里每根线都挂在别的线上，人出不去，因为那是他唯一知道的世界；网不罩在他外面，网就是他思想的纹理本身。
现在到了最容易被跳过的一步。你大概想说：科学不一样，科学会被推翻。这里得慢一点。物理学家 Duhem 和后来的哲学家 Quine 指出过：任何假说都不是单独受检的，它总拖着一堆辅助假设一起上场——仪器是准的、样本没污染、背景条件没变。结果不对时，逻辑上你永远可以改其中一条、把核心保住。Lakatos 索性给这圈可改的东西起了个名字叫保护带。所以「出了错还有说法」不能当分界线，两边都有。
那差别在哪？一个经得起用的版本是看改外圈要不要付代价：改完之后得交出一个新的、可以拿去查的预测，而且这次改动记在账上，别人能翻旧账。按这个标准，各种体系落在一条连续的轴上，而不是分成「理性」和「不理性」两堆。这一条是哲学论证，不是实验结论，得这么标着用。
学界为此吵过很久：1964 年哲学家 Winch 主张不同社会的理性标准本来就不一样，拿外面那把尺子去判是越界；1967 年人类学家 Horton 强调两边共享同一种解释世界的冲动，差别在体系开不开放。这个至今还有争论，但两边都承认：Zande 人在自己那套里推理得一点都不含糊。
EN To find out who is responsible, the Zande consult oracles, the weightiest being benge: a red poison ground from a forest creeper, mixed with water and fed to chickens, the question posed so that the fowl's death or survival answers it. Serious matters require two rounds with the question reversed — if the first fowl dies the second must live — which is a built-in consistency check, not an absence of testing. When the check fails, Evans-Pritchard recorded roughly eight kinds of reason: wrong variety of poison, a breached taboo, interfering witchcraft, the anger of the owners of the forest, poison too old, angry ghosts, sorcery, faulty handling. Failure is recycled as evidence, and the belief is strengthened rather than shaken. The tempting reply is that science differs because it can be refuted — but Duhem and later Quine showed that no hypothesis is tested alone; auxiliary assumptions can always be adjusted to save the core, and Lakatos named that adjustable ring the protective belt. So "there is always an excuse" cannot be the dividing line. A more usable line is whether adjusting the belt costs something: a new checkable prediction, and a public record others can hold you to. Winch argued in 1964 that standards of rationality differ between societies; Horton argued in 1967 for a shared explanatory impulse differing in openness. The debate continues; both sides grant that Zande reasoning is rigorous given its premises.
如果巫术只是一套解释倒霉的说法，那谁被点名应该挺随机。真实的分布完全不随机。Evans-Pritchard 记录的指控几乎全发生在住得近、身份相当的人之间：同一片邻里、同一个丈夫的几位妻子。往上指控是行不通的——跟贵族过不去，输的是自己。能被指控的人，先得是你得罪得起的人。
更硬的对照来自另一个大陆。1970 年历史学家 Macfarlane 翻了英格兰 Essex 一郡 1560 到 1680 年的巡回法庭档案，五百多份巫术起诉躺在那里，情节几乎是同一个模板：穷邻居上门讨点吃的或借样东西，被拒；走时嘟囔了几句；不久拒绝的那家牛死了、孩子病了；那句嘟囔于是成了证据。第二年 Thomas 给出同一套读法：旧的邻里互助义务正在松动、拒绝又还让人心里过不去的时候，指控最密。被指控的人里，女性和年长者明显偏多。
批评也得如实说。后来的研究认为「拒绝施舍」这套模型盖不住所有案子：一波审判的起落，常常是谁在推动起诉、法条怎么改更有解释力；而法庭档案本身是被筛过的样本。稳的是那个形状，不是那套动机——把一段时间里所有指控的箭头画出来，它们会挤在关系密、互相欠着人情、又刚好有摩擦的那几对人之间。指控的分布是一张社会张力的地图，不是一张倒霉程度的地图。
这套东西也不是冻在「传统」里的。2024 年有研究者报告了 2014 到 2016 年在南苏丹 Zande 人地方法庭做的田野：mangu 这一类几乎从法庭里消失了，而 ngua 还在，法庭忙的是后者。同一群人，九十年，分类本身换了。还有一句得说明白：今天一些地方的巫术指控会导致真实的暴力和伤害；讲机制不是替它开脱，恰恰相反——正因为指控的落点有规律，才可能在事情发生之前看出谁会被推到那个位置上。
EN If witchcraft were merely a way of explaining misfortune, the choice of culprit should look random. It does not. The accusations Evans-Pritchard recorded fall between people who live close and rank about the same — neighbours, co-wives — and accusing upward, against nobles, tends to cost the accuser. A harder comparison comes from England: in 1970 Macfarlane worked through the Essex assize records for 1560–1680, where over five hundred witchcraft indictments survive, and found a recurring template — a poor neighbour asks for food or a loan, is refused, mutters on the way out, and when the refusing household's cow dies the muttering becomes evidence. Thomas published the same reading a year later: accusations cluster where old obligations of neighbourly help are loosening while refusal still stings. Women and the old were accused disproportionately. The criticism must be stated too: the charity-refused model does not cover every case, and the rise and fall of prosecutions often tracks legal and political factors more than neighbourly tension; court records are also a filtered sample. What is robust is the shape, not the motive — the distribution of accusations maps social tension rather than misfortune. Nor is any of this frozen: fieldwork in South Sudanese Zande courts in 2014–2016, published in 2024, reports that mangu has all but vanished from court business while ngua continues. Accusation-driven violence is a real harm today; explaining the mechanism is the opposite of excusing it, since a patterned target can be seen coming.