「要尊重不同的文化」今天温和得几乎没有内容，像贴在墙上的标语。但它一百年前是一句有攻击性的科学主张：当时的正统认为人类社会排成一条队，有的在前有的落后，而名次可以用身体量出来。推翻它的人不是靠喊平等，是靠测量。
麻烦紧接着就来。如果每套做法只能按它自己的标准理解，那「尊重」本身也只是某一套标准，从外面批评任何事的语言就全没了。这一页拆三件事：它当初怎么立起来，边界卡在哪一句话上，「站在他们的角度」又为什么比听起来难得多。
1883 年，一个二十五岁的德国人上船去巴芬岛（今加拿大北部）住了大约一年。他叫 Franz Boas，原本学物理，博士论文做的是海水的颜色——同一片水，为什么你看是这个色、我看是那个色。这个问题他后来整套搬到了人身上。
他要推翻的东西当时是正统：十九世纪的主流把人类社会排成一条线，从蒙昧到野蛮到文明；博物馆就按这套摆，全世界的鱼钩放一个柜子，从粗糙到精巧排成一列，参观者一眼看出「进化」。1887 年 Boas 在《科学》杂志上跟史密森学会的 Otis Mason 为此打了一场笔仗：形状相同的东西，在不同地方长出来的原因可以完全不同——不同的原因能产生相同的结果。所以一件器物首先是它那套生活的一部分，不是某个类型的标本，该按部落和地域来摆。这个立场叫历史特殊论（historical particularism）。
这就是整套文化相对主义的地基，而它是一条朴素的机制：一个做法的意义由它在整套安排里的位置决定。同一个动作在两套安排里不是同一件事，因为它各自连着的东西不一样。
把正统打出一个洞的是一次测量。1908 到 1910 年 Boas 受美国移民委员会委托，给上万名移民和他们在美国出生的子女做体质测量，结果是：在美国出生的那一代，头型指数（当时用来分「种族类型」的头骨长宽比）发生了可测的变化。一个一代之内就会动的指标，不配当固定的种族标志。证据强度如实说：这批数据后来被重算过两次，结论相反，至今没定论；但「头型不会变」这个更强的说法没再站起来。
还有一段得如实交代，它是这门学科的问题史。1897 年 Boas 想请一位格陵兰北部的因纽特人来他任职的美国自然史博物馆接受研究，探险家 Robert Peary 带回来的是六个人，四人死于结核等疾病；其中 Qisuk 七岁的儿子 Minik 被博物馆用一场假葬礼骗过去，父亲的骨骸进了馆藏，他索要未果，四人的遗骨 1993 年才送回格陵兰下葬。一个人可以在理论上反对把人排成等级，同时在操作上把人当标本。
EN In 1883 a twenty-five-year-old German spent about a year on Baffin Island. Franz Boas had trained as a physicist, with a doctorate on the colour of seawater and why two observers see it differently — a question he later moved wholesale onto people. The orthodoxy he attacked put societies on one line from savagery through barbarism to civilisation, and museums displayed the world's fishhooks in a single case ordered crude to refined. In an 1887 exchange in Science with Otis Mason of the Smithsonian, Boas argued that unlike causes can produce like effects, so an object is first of all part of a way of life rather than a specimen of a type, and belongs arranged by tribe and region. That position is historical particularism, and its mechanism is plain: what a practice means is set by its position in the whole arrangement. What punctured the orthodoxy was a measurement. Commissioned by the US Immigration Commission between 1908 and 1910, Boas measured over ten thousand immigrants and their American-born children and reported that the cephalic index — then used to sort "racial types" — shifted measurably in the American-born generation. An index that moves within one generation cannot be a fixed racial marker. Honest caveat: the data have been reanalysed twice with opposite conclusions, one group in 2002 finding a high hereditary component and judging Boas to have overstated environmental effects, another in 2003 largely vindicating him; the stronger claim that head form is immutable never recovered. One episode belongs to the discipline's record. In 1897 Boas asked the explorer Robert Peary to bring one Inuk from northern Greenland to the American Museum of Natural History, where he worked; Peary brought six, and four died of tuberculosis and other illnesses. Qisuk's seven-year-old son Minik was deceived with a staged burial while his father's skeleton entered the collection, and his later public demand for the remains failed; the four were returned to Greenland for burial only in 1993. A person can oppose ranking humanity in theory while treating humans as specimens in practice.
1934 年 Boas 的学生 Ruth Benedict 出《文化模式》，比喻很好记：人类可能的做法摆成一道巨大的弧，每个社会只从弧上选一小段，再把这一小段整合成一种风格。她拿美国西南部的 Zuni 和西北海岸的 Kwakiutl 对比——前者的场面是收敛、不出头，后者是竞争、夸示，在仪式上比谁散得掉更多财物。
这本书把「不一样」变成了可理解的东西，而不是缺陷；挨的批评是整合被夸大——一整个社会被写成一个人格。这类工作是解释性的模式归纳，不是可检验的测量；同门 Margaret Mead 那本《萨摩亚人的成年》(1928) 半世纪后被 Derek Freeman 指为受了受访者玩笑的误导，也是同一种脆——单点田野对单点田野，谁都没有第三方数据。
然后是这一页最要紧的一刀。有两个东西必须分开。方法上的相对主义：描述时先把自己的判准搁一边，否则你看不见人家的逻辑——这是工具。道德上的相对主义：任何做法都不能从外部评价——这是一个规范主张。前者推不出后者。「这件事在他们那套安排里自洽」是关于事实的话，里面不含任何「所以该怎样」。
这一步在历史上真的滑过去了。1947 年 Melville Herskovits 起草了一份《关于人权的声明》，发在《美国人类学家》上，回应联合国正在做的世界人权宣言：人的标准只能相对于他所属的文化，一套「普遍人权」装的其实是欧美关于个人与财产的观念。它的动机得看清——当时殖民体系还在运转，这是在替被殖民者的文化完整性挡一刀。反对一方最强的表述是：这么一挡也挡掉了从外部批评暴行的语言，而最需要它的往往是任何一套安排里权力最小的人。两边都不是稻草人；学科后来自己换了位置，1999 年美国人类学会的宣言转向主动主张人权。本页不替任何一边下判断，但那两句话得分开——Boas 反对纳粹的种族科学，这种反对也必须站在某个跨文化的立场上。
EN Ruth Benedict, a student of Boas, published Patterns of Culture in 1934 with a memorable image: the practices humanity could adopt form one great arc, and each society selects a short stretch of it and integrates that stretch into a style. She contrasted the restraint of the Zuni with the competitive display of the Kwakiutl, who vied at ceremonies over how much property they could give away. The book made difference intelligible rather than defective; the criticism is equally real, since integration was overstated and a whole society written as one personality, with internal conflict and change flattened. Such work is interpretive pattern-finding, not testable measurement. For how fragile the evidence is, take another quarrel in the same lineage. Margaret Mead spent nine months on a small Samoan island in 1925–26 and argued in Coming of Age in Samoa (1928) that adolescent turmoil is no biological necessity; Derek Freeman's 1983 book claimed she had been misled by two young women's joking. Paul Shankman, who has documented the controversy most closely, judges that Mead over-generalised from one island, that Freeman's counter-evidence is itself weak, and that her core claim still stands. Single-site fieldwork against single-site fieldwork, with no third-party data on either side. Now the cut that matters. Methodological relativism — setting your own standards aside while describing, or you will not see the other logic — is a tool. Moral relativism — that no practice may be judged from outside — is a normative claim, and the first does not imply the second: "this is coherent within their arrangement" is a factual sentence containing no "therefore one ought". The slide happened. In 1947 Melville Herskovits drafted a Statement on Human Rights in American Anthropologist, responding to the UN's universal declaration then in preparation: standards are relative to culture, and "universal human rights" encode Euro-American notions of the individual and property. Read its motive fairly — the colonial system was still running, and the statement shielded colonised peoples' cultural integrity. The strongest objection is that the same shield removes the language for criticising abuses from outside, and those who most need that language are usually the least powerful inside any arrangement. Neither side is a straw man. The discipline moved: the AAA's 1999 declaration actively asserts human rights. This page takes no side, but separating those two sentences is the precondition for reading any of it — and note that Boas publicly opposed Nazi race science, which requires standing somewhere across cultures.
1956 年 Horace Miner 在《美国人类学家》上发表了一篇格式正经的报告，讲一个叫 Nacirema 的族群：家家都有一个神龛，柜里放着巫医开的魔法药剂，没人敢扔；信徒每天早上在圣水前做一套嘴部仪式，还定期去见一种嘴巫医，任对方用尖锐工具扎牙。读着就发现不对——Nacirema 倒过来拼是 American，那套仪式就是刷牙。这是玩笑，也是个精密仪器：猎奇感不是从被描述的做法里冒出来的，是描述的角度带进去的。
民族中心主义（ethnocentrism）这个词 1906 年由 William Graham Sumner 在《民俗》里推广开，指把自己所属的群体当成一切的中心和尺度。机制不神秘：任何分类系统都得有个原点，而你手上那套的原点就是你自己那套生活。这跟「种族与族群」那期的边界机制是一件事的两面：边界总要有一侧当基准，默认是自己。
正面的田野例子很漂亮。1949 到 1953 年间 Laura Bohannan 在尼日利亚中部的 Tiv 人当中做田野，去时带着一个信念：莎士比亚是普遍的。她把《哈姆雷特》讲给几位长老听，长老们一边听一边纠正她：死人不会回来，那个「鬼」必然是巫术或凶兆；至于叔叔娶了寡嫂、当上首领，这有什么问题，本该如此。讲完长老们的评语是，你们那边的人该多听听我们讲，才不至于把事情理解错。要点不是「他们不懂莎士比亚」：长老们做的事和文学教授做的是同一件，拿一套自洽的理论解释情节——换掉的是理论，不是解释能力。
相对主义自己也会翻车，翻成另一种他者化：1983 年 Johannes Fabian 挑出民族志爱用现在时的习惯（「Tiv 人相信……」），说它把被研究的人放进一个没有历史、跟作者不同处一个时代的位置。所以比较的单位从哪来，本身要被查一遍——查今天最常用的那个工具就知道这句不空。公司里最普及的跨文化比较版本是 Geert Hofstede 的国别文化维度：1967 到 1973 年他在 IBM 内部做员工价值观调查，十万份以上问卷，最后用四十个国家的结果算出几个维度（权力距离、个体与集体的偏向等），几十年里被商学院和跨国公司拿去做培训。批评也硬，Brendan McSweeney 2002 年那篇最有名：样本全是一家公司的白领、多数是男性，四十国里只有六国的样本量过千；更根本的是它把「国家」当成文化的单位、把文化当成静态刻度。Schwartz 的价值观量表和 GLOBE 研究切出的维度也并不重合。横截面问卷、相关性，外加一个可疑的假定：当粗糙提示可以，别当尺子。
EN In 1956 Horace Miner published a straight-faced report in American Anthropologist on a people called the Nacirema: every household keeps a shrine whose cabinet holds magical potions prescribed by medicine men and which nobody dares empty; devotees perform a daily mouth-rite before holy water and periodically visit a mouth-specialist who probes their teeth with sharp instruments. Nacirema is American backwards and the rite is brushing your teeth. The joke is a precision instrument: exoticism is not emitted by the practice, it is carried in by the angle of description. Ethnocentrism, popularised by William Graham Sumner in Folkways (1906), means treating your own group as the centre and measure of everything, and the mechanism is unmysterious — any classification needs an origin point, and yours is your own way of life. It is the other face of boundary-drawing: a boundary needs one side as baseline, and by default that side is yours. The positive fieldwork case is lovely. Between 1949 and 1953 Laura Bohannan worked among the Tiv of central Nigeria, convinced that Shakespeare was universal. Telling Hamlet to a group of elders, she was corrected throughout: the dead do not return, so the ghost must be witchcraft or an omen; and an uncle marrying his brother's widow and becoming chief is exactly what ought to happen. Their closing verdict was that her people should listen to Tiv elders more, so as to stop misunderstanding things. The point is not that they missed Shakespeare — the elders did what a literature professor does, applying a coherent theory to a plot. The theory changed, not the capacity. Relativism can also fail into a subtler othering. Johannes Fabian's Time and the Other (1983) named the ethnographic present tense — "the Tiv believe…" — as a denial of coevalness, placing people outside history and outside the writer's own time; Edward Said's Orientalism (1978) attacked the same thing from another direction. Neither says stop comparing; both say audit where your units of comparison came from. That audit bites on today's standard tool. Geert Hofstede's national culture dimensions came from IBM employee value surveys between 1967 and 1973, over a hundred thousand questionnaires, with the published dimensions computed from forty countries and taught in business schools ever since. Brendan McSweeney's 2002 critique is the best known: respondents were white-collar staff at a single firm, mostly men, only six of the forty countries had more than a thousand respondents, and more fundamentally the nation is taken as the unit of culture and culture as a static scale. Schwartz's value surveys and the GLOBE project cut the dimensions differently and do not agree. Cross-sectional, correlational, resting on a questionable assumption: a rough prompt, not a ruler.