Max Weber, Cavalli-Sforza, Ethnicity, and Population Genetics

Ok, below is a complicate and attenuated definition of ethnicity by the classical sociologist Max Weber.  Variations of this definition are found in many anthropology and sociological textbooks, though he is by far not the only source of wisdom.  But be aware that as with most classical literature, it is often difficult to read.  But for the purposes of this discussion with Population geneticists, I want to highlight Weber’s emphasis in beliefs about heredity and blood relationships in determining endogamy and exogamy.  All ethnic groups encourage the youth to have babies with people who are “like us,” however defined.  The result indeed is that in a rough way, genes are inherited within “ethnic groups,” or at least there are belief systems indicating that this happens.  I wrote about this a bit earlier at here.  Note, this version is suitable for use with undergrads—it is easier to read than what follows.

Anyway, I like the broad brush analysis of blood alleles, and glotto-chronology that people like Cavalli-Sforza use to map deep history and very general relationships (OK I know that glotto-chronology is also known for its limitiations).  This is the only effective way of studying such migrations, given the paucity of archaeological and historical data.  Ok, so fine.

But we know a lot more from the studies of people like Weber (and his successors) about the overwhelming role that ideology, inequality, racism, etc. play in structuring mating habits.  A sampling of Max Weber’s thoughts appears below in all its complexity.  My question for the people following in Cavalli-Sforza’s tradition like Razib Khan is, how would you go about including such “variables” as Weber describes in mathematical models?  My feeling is that given the inherently fluid nature of such definitions, and the compromises necessary to simplify research questions so that they fit into something that is “countable,” are a step too far.  And as a result, you get the reactions of myself, and most social scientists that we should not depend too much on such quantitative data which inherently simplifies social complexity—ethnographic data is at least as important.

Anyway: Here is Weber’s description/definition of ethnicity.  Links to the original articles are below.  There version here is a translation I participated in, and appeared in the peer-reviewed Journal of Classical Sociology in 2010.

“When the most extreme consequences of stratification are pressed, the Stand evolves into a closed ‘caste’. That means, apart from the conventions and legal guarantees, rituals develop guaranteeing Stände-related distinctions. This is achieved by restricting any physical contact of members of higher castes with members of castes regarded as “lower,” and protects the higher caste …Therefore, the individual castes partly develop distinctive cults and gods.

As a result of these consequences, the Stände-related stratification only then lead to the development of castes where underlying differences can be found which are held to be “ethnic”. Particularly the “caste” is the normal form of Gemeinschaft communities which are the precursors of the Gesellschaft type-societies created who live along the lines of “ethnicity,” and therefore believe in blood relationship, and restrict both exogamous marriage and social intercourse. These aspects can be found among pariah peoples around the world….

Ethnic and caste segregation also differ regarding their effects. Ethnic coexistence, which implies mutual rejection and disdain, also permits any ethnic community to value its personal honor as the highest. However, caste stratification is accompanied by a ‘vertical social gradation’, and acknowledges a socially accepted higher “honor” to the benefiting privileged castes and Stände. This is typically explained by arguing that ethnic differences were transformed into differences of “function” within a politicized Gesellschaft-like social order (warrior, priest, and craftsmen who are politically important for war, and building trades, and so on). Even the most despised pariah people somehow cultivate what is peculiar to them, in the same manner that ethnic and ‘Stände’-related communities do. They especially continue to cultivate the belief in their own unique “honor” (as do the Jews).

However, Stände which are both despised and negatively privileged show a specific deviation regarding the “sense of dignity” …But to understand this, it is necessary to focus on the position of the privileged. Their “sense of dignity” is the subjective precipitation in social honor and of conventional demands which a positively privileged “Stand” requires for the deportment of its members. As a result, it can be said that the positively privileged ‘Stände’ sense of dignity, naturally relies on its “who they are”, they do not rely on transcending values, but they refer to their own “beauty and excellence”. Their kingdom is “of this world”, and they live for the present and justify their privilege by referring to a glorious past.

Naturally the negatively privileged status group can only draw its sense of dignity by referring to a future which lies beyond the present, and is temporal or transcendent. In other words, this sense of dignity is nourished from the belief in a providential “mission”, or a specific honor before God as the “chosen people”. Therefore, the idea arises that “the last will be the first” beyond this life, or that in the present life a messiah will arrive who will shine a light upon the honor of the pariah people (Jews) or ‘Stand’, which has before been concealed from the world. These simple facts are the source of a pariah ‘Stände’s’ character of religiosity. …

This is to say that the ethnic origin of Stände formation is by no means a normal phenomenon. On the contrary, since objective “racial differences” are not based on every subjective “ethnic” mutual feeling, a racialized justification for ‘Stände’-related stratification is ultimately tested with concrete individual cases. Quite frequently, the ‘Stand’ itself creates ‘pure-breds’ [or stereotypes] which are an anthropological type. The Stand functions on a highly exclusive manner and is based on a selection of individuals who are personally qualified for membership (e.g. the Knighthood), based on their martial, physical, and psychological eligibility.

So, from a practical point of view, the stratification by Stände goes hand in hand with a monopolization of ideal and material goods or opportunities, in a manner which we have come to know as typical. Besides the specific honor of Stand, which always bases itself upon distance and exclusiveness, there are all sorts of material monopolies”


Thank you for reading this far!  It is work to read this far.  (Now those of you who are mathematically inclined know how we feel when we deal with your elegant mathematical formulations!)  Anyway, if you want to read more, please look at the entire translation of Weber’s work at the Journal of Classical Sociology (2010), as well as our commentary, which is also there.  In my view, a meeting of minds between the population geneticists, and the sociologists/anthropologists would be useful for understanding such matters.  I’m just not sure how it is going to happen.


An afterthought and a comment for Razib Khan:

Razib Khan over on one of Michael Scroggins posts linked two blogs of his from Discover Magazine.  I read them, and appreciated that he was careful in his discussion of race, even though he did not cite the relevant anthropologists or sociologists (Note to Razib: Need Weber in there, or perhaps Cornell and Hartmann’s textbook Ethnicity and Race).  I believe he even used the term “social construction” at one point, which hearkens back to the work of Weber and others.

Razib continues

“the biology is more interesting than the sociology, which can be decomposed pretty easily.”


Ok I will let him have his own opinion on what is “more interesting,” but I look forward to his deconstruction of a classical text like Weber, or even a more contemporary approach like Cornell and Hartmann.  Weber of course is difficult to read, but generations of sociology undergrads have somehow gotten through.  Cornell and Harmann though is well-written and hardl

84 thoughts on “Max Weber, Cavalli-Sforza, Ethnicity, and Population Genetics

  1. Erm, could you please explain how that post argues that racial categories are real? Because I see lines like:

    ” The main point which I think we can all agree upon is that colloquial understanding of race has only a partial correlation with any genetic understanding of race. I myself have ranted against the confusions which have ensued because of the conflation of the two classes, and it is certainly a legitimate area of study, but it is not my primary concern.”


    “And that’s the key: racial typologies are coarse reflections of genuine history. In other words, race is a reflection and reification of genuine lower level dynamics, it is not the prior phenomenon.”

    I don’t see how it does. Can you explain what your argument is without just posting to articles which either do not address, or reinforce the point I was trying to make?

  2. @dad
    My sense is that you don’t understand the point I’m making, because nothing you’ve posted so far actually addresses it. But you do illustrate my point that when scientists use the word race as an analytic category, people will read popular understandings of race into it – even if they explicitly warn that popular understandings of racial categories are wrong.

  3. As for the comments on intelligence, again, you’re mixing topics. We are talking about the scientific construct as discussed in the psychometric literature. This construct is not “culturally specific”. That is, intelligence is defined in terms of the covariance of test scores. The factorial structure is common across cultures. You said: “It would be a silly idea though to administer the SAT in a farming community where cognitive skills are defined differently.” IQ and other cognitive tests directly measure acquired knowledge. In circumstances where the opportunity to acquire knowledge is relatively egalitarian, then IQ and ability tests end up indirectly measuring ability. So, to address what you said specifically, using the SAT to measure intelligence would be perfectly fine in the case of a “farming community where cognitive skills are defined differently” so long as the SAT questions indexed information commonly obtainable in the community. If not, you would have to change the question. Ask farm related questions: “If a hen lays 1 egg per 2 days how many eggs would the hen lay in 23 days?” This all is obvious and everyone recognizes it.

  4. @ KbH

    I almost lost this in the line of comments.

    Have you ever read Ingold on attention or cognition? He has been writing on both topics over the last few years and has been very critical of Sperber, in particular. (

    His argument about following rules and directions (be they genetic or textual) is virtually identical to that put forward by Garfinkel in latter years. My feeling here is that, even if you wish to assign a range to genetic action you still have to account for the incompleteness of directions at all levels. It isn’t quite the same as limits imposed by inert materials. Living things are different than inert things.

    Though, I suspect you are right and the difference between us is one of abstract level.

  5. Justaguy said:

    “What I’m disputing is not the uneven distribution, but the idea that categories exist on the genetic level, or that those categories are best described by the word “race”.”

    Firstly, “race” is a term that refers to an umbrella of concepts. Secondly, there are a number of scientific race concepts, some biological and some not. Thirdly, there are a number of biological race concepts e.g., race qua breeding population, race qua ecotype, and race qua geographic subspecies. So stop reifying the term and name the specific race concept that you feel poorly fits humans. Here was one of the better defenses of human biological races (HBR) that I have seen in a number of years:

    “How to Be a Biological Racial Realist
    Quayshawn Spencer

    In this talk, I will show that the case for biological racial realism is more formidable than philosophers have thought, provided that one adopts the right semantic, metaphysical, and biological assumptions. Specifically, I will argue that given a referentialist account of the meaning of ‘race’, a genuine kind account of biologically real kindhood, a fuzzy graph- theoretic account of populations, and the landmark results from Noah Rosenberg et al. (2002; 2005) on human population substructure, one can fashion a respectable account of race
    as the “B-partition”of metapopulations in a species(or “BPM race theory”).”

    Of course, this is basically the “race theory” that intellectual giants such as Mayr employed — so nothing new in that respect.

  6. @ Chuck

    In “fixing” the IQ test by asking farming related questions you have just moved the basis of the measurement from a “population” with shared heredity, to a community of people who share in the activity of farming. Additionally you have just moved the basis of the test from measuring abstract concepts to measuring concrete activities.

    This demonstrates the point we have all been making that context is prior to any discussion of “higher cognition” or “intelligence.”

  7. unless that context is a western one. in that case, you reject any definition of intelligence at all.

  8. Michael,

    “In “fixing” the IQ test by asking farming related questions you have just moved the basis of the measurement from a “population” with shared heredity, to a community of people who share in the activity of farming.”

    I can’t make heads or tails out of this. For one, I don’t see why being defined by “a shared activity of farming” would preclude being defined by “shared heredity.” Amish come to mind — they represent a breeding population. The first leads to the second. For another, as noted before, social construction does not preclude genetic differences — and you don’t need relative genetic relatedness for mean genetic difference. Finally, none of this is relevant to what I said, which was: SAT/IQ measures information obtained. In situations where the flow of information is not restricted, because more mentally able people are more capable of processing, integrating, storing, and retrieving information, more mentally able people will do better on SAT/IQ tests. SAT/IQ then measures mental ability.

  9. It seems dad’s knowledge of human genetics consists entirely of factoids and snippets half-baked wisdom he has gleaned from reading blogs such as Steve Sailer, vdare, etc. As he has nothing useful to contribute to this conversation, he should be best ignored.

    Speaking as a long time reader of gnxp, there is absolutely nothing about justaguy’s recent posts which I or Razib would find objectionable. Had I not bothered to peruse the rest of the comments below mine, I fear I would have composed enough text to complete a doctoral thesis worth of soporific prose. :-)

  10. ok, i read through your 3 posts. all you did was retype stuff razib already said and that all gnxp (like me) readers already know. that’s hwy i had skipped all of your posts until now. i, however, actually have my “name” in one of the titles of Tony’s posts. so i’m pretty sure i’ve been worth more so far. anyway, what have i said that you disagree with? it seems like you’re a really pretentious, pedantic nerd who just doesn’t like me yet never actually says why. get back to me about this, i worried we won’t be friends

  11. ^^ you see? Not much meat in the coconut, is there? I rest my case.

    On another thread here I posted a compete essay denouncing Stephen Jay Gould, the arch-nemesis of hereditarians everywhere, in no uncertain terms – had you even bothered to read it, you would not have made a fool of yourself by imputing false motives where none exist whatsoever.

  12. oh, wait, that was a good post. seriously, tho, why are you so angry with me? it’s hilarious how you are mad but STILL WON’T TELL ME WHY

  13. wait, i did read that and even complimented you on it. what is wrong with you? you are weird, sam

  14. wow, i now realize that your post on Gould is in the thread with my handle IN THE TITLE. wow, you are dumb. also, you said “by the way,” to start almost every paragraph in that editorial.

  15. @Michael,

    “Though, I suspect you are right and the difference between us is one of abstract level.”

    Absolutely, we seem to agree on interactionism. I was citing Garfinkel and Sperber to get away from Durkheim and Bourdieu specifically because they reify categories as objective, but since we’re here, we can go further. Here goes.

    Ingold’s criticism of Sperber makes some good points re: that there can can exist complete truth-conditional information in symbolic form outside of context (as in the recipe example). Ingold is right, here and this has been demonstrated by Searle with regard to literal language. I also think Ingold is correct in general in eschewing the symbol processing model of cognition. In fact, I’m quite sympathetic to the embodiment approach he draws upon from Gibson, Heidegger, and Merleau-Ponty. (But we’ll see that Ingold cannot abide embodiment because he denies some features of the body.) Finally, my endorsement of Sperber relates specifically to his interactionist framework in general opposition to the Durkheimian reification of categories. Having said that, I do indeed prefer Ingold’s approach to cognition over Sperber’s, but we’re not done here. This is all true, but doesn’t really address innatism (for lack of a better word).

    What I find problematic about Ingold’s developmental approach is that it hits a wall as soon as we examine non-‘normal’ cognition. More to the point, some conditions are spectral- like autism. We simply can’t say, in that case, that those instances don’t count and that we’re only talking about normal processing because we don’t have a hard line to distinguish. And we simply cannot just excuse ourselves by saying that that is an anomaly because it’s spectral and heterogeneous (also genetic). Which is all to demonstrate the radical development approach assumes a universality that empirically doesn’t exist. (Consider also the cases of sexual orientation.)

    Or he is not assuming universality, but then his argument must be that these conditions require interaction to manifest, the observation is trivial and is just restating that genes are process. Of course this is true, as gravity is process, but only skirts the issue that processes aren’t arbitrary. Also, if his argument is that since all matter is permeable and procedural there is no valid concept of “innate”, that is another distraction. Even if we shift and talk about attractors of processes, we still have to talk about the particular nature of any given attractor – which incidentally maps to a human with a particular nature. At best he can say that “innate” is only a heuristic- but it’s an empirically robust heuristic and he doesn’t account for that.

    Finally, Ingold seems to be more than willing to admit that environmental features have particular qualities (he must if he follows Gibson). The qualities affect practice – and are not all subject to human interaction, like magnetism or melting point. Insofar as humans are material, they too have particular properties. In the end, he is merely asserting uniformity though he doesn’t need to. His interactionist approach doesn’t depend on it and ironically he is asserting the very Cartesian dichotomy he seeks to negate. For Ingold, the mind is developed interaction. Interaction of what? All materials. What is the nature of the materials? It depends on the material… except for if the material is inside the human. It’s as if we were dealing with a geologist that was content to stop at “all rocks are rocks”. Either he’s treating the human as some special case of material, distinct from any other material or he is treating all materials as processes and only admitting some arbitrarily chosen characteristics. See what I mean?

    Incidentally, you mention “It isn’t quite the same as limits imposed by inert materials.” I don’t think Ingold believes in inert materials at all. All materials are process (or Lines) for him. I don’t think he makes the ontological distinction you do between living things and non-living things, but I could be misreading him.

    BTW, dad is really distracting.

  16. @ KbH

    I don’t think he would use the term inert. But, he does make a distinction between the living and the non-living. I read him as pretty far from a perspective like ANT on this account.

    I think the large difference between Ingold and Sperber, and the one that your objections speak to, is Ingold’s firm insistence on the organism (and its development in time) as the unit of analysis.

    My other thought here is that, again as you point to, Ingold hasn’t yet worked it all out. He still has some loose threads out there which he is trying to synthesize. Ultimately, I think quite a bit of it will run through evo-devo and/or development systems biology.

    The varnish is starting to wear off the modern synthesis and fields like population genetics are just the first to feel the effects.

  17. Justaguy said:
    “Modern conceptions of race are derived from pseudoscientific ideas about there being different categories of humans.”

    This is more silliness. The modern taxonomic concept of race as applied to humans and to non-humans has the same origin. To the extent you claim that the former is “pseudoscientific”, you must claim the latter is. Reductio ad absurdum. And doing so is horribly anachronistic. Pseudoscience means something like “scientific fraud.” Yet that doesn’t characterize the situation at all. Linnaeus or Kant might have been working from a different model e.g., categorical versus fuzzy set — though I think the difference is more nuanced — but that not eq

    That is different from a modern genetics understanding, which looks at the degrees of relatedness of different populations. Those are two fundamentally different ways of looking at genetic variation. One is a measure of relatedness between two socially and/or geographically separate groups of people amidst a continuum of variation, and the other poses socially and/or geographically separate groups of people as being manifestations of distinct racial essences.

    Of course, the people who would be categorized as belonging to the same race are genetically more related to each other by some ways of measurement than they are to other groups. So, what’s the problem of a geneticist using racial categories to discuss groups which are more internally related to each other?

    The big problem is that when scientists use the word race it reinforces popular understandings of race as being scientifically valid. But popular understandings of racial difference are never reducible to simple genetic similarity – they always involve a moral and mental dimension that are not present in a spreadsheet which analyzes admixture. So, why retrofit a pseudoscientific concept for use as an analytic category within science? Especially when the pseudoscientific meaning is so pervasive that it can easily lead to misunderstanding, such as when Khan states that there are “biologically separate races”.

    When culture is used effectively within anthropology it doesn’t describe cultures as static, monolithic and uniform. That is, you can only talk about American culture in a very broad and general sense, individuals within America do not share the same beliefs, practices and social institutions as each other. Good anthropology tend not to use culture as a distinct thing, but to describe practices, beliefs and institutions as cultural to discuss the extent to which what people do is colored by shared sentiments, practices, symbols, meanings, etc..

    Just because they are both categories used to explain human difference doesn’t mean they are equivalent to each other, or that they are used in the same way.

  18. After 29 comments to this thread, I am still waiting on Razib’s “easy” deconstruction of Weber’s definition of ethnicity and race.

    i told tony this via email when he contacted me, but for the record, blogging is much a secondary responsibility this term for me between research, motivating myself to write a paper, teaching, etc. so sorry i haven’t been able to participate, though i don’t know if i have much to add. i’ve talked about this stuff for 10 years.

    the main problem that the non-pop gen ppl on this thread have is that i suspect they lack an intuition for how crisp clustering is, or isn’t, in real populations. whether a particular *genetic* concept of a population is useful in an instrumental sense, and it’s always instrumental, depends on how useful it’s been in the past. it’s iterative.

    so, for example, i know that if you gave me any 100,000 SNPs ascertained in europeans (100,000 base pair markers) and a population of 100 germans from frankfurt (defined as people whose ancestors were on church rolls in 1850, and NOT of hugenot extraction) and 100 lithuanians from villages outside vilnius, the two would separate out into clear and distinct clusters using PCA. PCA just pulls out the variation in the data set, and generates dimension out of it which can be visualized. if you repeated this assertion with nigerians and lithuanians, i could be willing to go down to 1,000 markers. if we repeated this assertion with latvians and lithuanians, perhaps 1,000,000 markers would do the trick, but i would be less sure going down toward 100,000 markers. this is all informed by my *experience* running these data sets hundreds of times (at this point thousands depending on how you count a replicate).

  19. That is different from a modern genetics understanding, which looks at the degrees of relatedness of different populations.

    a minor point which i think is important today. for reasons of data limitation humanists used to have a priori populations, and then compare differences between those populations. today, it is more common to actually looking at all *individuals* within both populations, and construct the populations *after* making sure that all the individuals within separate populations are actually genetically differentiable. IOW, “outliers” are removed. both methods have drawbacks, but i think this must be kept in mind.

  20. Justaguy said:
    “Modern conceptions of race are derived from pseudoscientific ideas about there being different categories of humans.”

    Oops–accidentally quick-key posted when writing a reply. To continue:— but that’s not equivalent to engaging in artifice. That is, given the models they were working from and the data on hand, the conclusions were, for the most part, scientifically respectable. You continue:

    “One is a measure of relatedness between two socially and/or geographically separate groups of people … the other poses socially and/or geographically separate groups of people as being manifestations of distinct racial essences.”

    Classic taxonomy a la Aristotle was, indeed, essentialist. The metaphysics of modern taxonomy is more ambiguous. We would have to discuss specific systems and authors. Whatever the case, there is no necessary contraction between a “modern genetic understanding” and a racial essential understanding, since the former deals with accidents (the physical) and the latter deals with substance (the meta-physical). That is, one could just as well argue that populations are “distinct racial essences” in process of manifesting. No one does, because everyone works from an evolutionary frame.

    You say: “The big problem is that when scientists use the word race it reinforces popular understandings of race as being scientifically valid.”

    One could make a similar argument against speaking of “species” or “kingdoms” since both were originally conceptualized in a pre- evolutionary framework. I imagine you would grant this but note that speciesism isn’t a pressing problem. And you continue:

    “So, why retrofit a pseudoscientific concept for use as an analytic category within science?”

    The scientific categories e.g., breeding populations or ecotypes or subspecies are not, in any meaningful sense, pseudoscientific. And they are independent of a metaphysical frame so there is no “retrofitting.” But that’s not the reason one would acknowledge human races. One would do so for the sake of internal consistency (assuming it would be consistent to do so)– since the race concept, now evolutionarily framed, is widely employed in biology.

  21. since the race concept, now evolutionarily framed, is widely employed in biology.

    animal biologists are being pressured now to stop using the term actually. in that they actually get feedback to use the word “population” or “breed” as a substitute. i know this from personal experience.

  22. @Razib
    I see how having more markers can refined the capacities for identifying differences between populations, in the long run. But how could this be used to add understanding to past relationships rooted in domination and subordination, such as Weber writes about? My sense is that the technique could possibly be used, at least in the bigger picture. Indeed, indirectly, the mitochondrial and Y chromosome comparisons on Iceland have already informed understandings of past relationships between Norse men and Celtic women in an interesting fashion which could be understood better in the context of archaeological and historical sagas.

  23. @Chuck
    I wasn’t talking about Linnaeus, I was talking about late 19th and early 20th century scientific racism. My understanding is that contemporary popular understandings of race came to prominence in the context of social darwinism and colonialism – placing the political hierarchy established under colonialism into scientific terms. While I tend to think of race science of that era as being pseudoscience, it is true that a lot of it was fairly mainstream.

    I understand that there is a debate as to the extent to which race works the same way as a category within humans as it does in other species, and I’m not really qualified to weigh in on it. I’m also fairly agnostic as to the analytic frameworks people in fields I’m unfamiliar should use. And I don’t think it really matters for the point that I’d make, because whatever a good definition of a biological subspecies of humans would look like, it would not resemble popular understandings of race.

    I’m starting with the assumption that science should be used to inform society. So, why then use vocabulary which will reinforce popular misconceptions? Especially when those popular misconceptions have negative impacts on society.

    Just to be clear, I’m not suggesting that scientists should base their research programs on political priorities. I’m suggesting that they should be mindful of the influence the work they do has on the broader public, and should communicate their work in a way that promotes more accurate understandings of what they study.

    And I’m not sure what’s at stake with wanting to retain the word race as an analytic category.

Comments are closed.