Marriage and kinship, oriented

Start with what a definition is for. Edvard Westermarck, in 1891, defined marriage as a union lasting beyond mating until after the birth of the child, which included every form he knew and excluded passing liaisons. Later anthropologists found cases that strained any single definition: societies where a woman's brothers, not her partners, supported her children, and where a husband was a visitor. The field's answer was to stop looking for the one true form and to ask instead which rights a union transfers, to whom. Rights over children, over labor, over property, over sexual access, and the duty of support can be bundled together or split among relatives.

Two distinctions carry the whole subject. Consanguines are related by descent; affines are related through a marriage. And the social father, who gives a child a name and a lineage, need not be the biological one. Adoption, godparenthood and blood brotherhood make kin by rite, and the obligations are real.

The usual beginner's mistake is to treat a custom as a relic. Nineteenth-century writers read the levirate, in which a widow passes to her husband's brother, as a survival of ancient polyandry. Westermarck replied that it fits inheritance of a widow as property just as well. David Schneider's 1984 critique pushed further, arguing that the field had exported Western ideas of blood and marriage and found them everywhere by definition. Marshall Sahlins's answer, kinship as "mutuality of being", widens the definition to whatever makes people share existence.

Rule: when a custom looks strange, ask what it does now, which rights it moves, and what the society itself calls kin, before you ask what stage it came from.

Every language sorts relatives into categories, and the categories differ. English lumps every cousin under one word and treats both parents' sides alike; other languages split cousins by which parent's sibling they come from. Lewis Henry Morgan's 1871 survey made the first cut, between classificatory systems, which group relatives into broad categories, and descriptive ones, which specify each relative by chaining primary terms. Kinbank, a database of more than 210,000 kin terms from 1,229 languages, now lets the systems be compared at scale.

Anthropologists reduce the possibilities to six types named for the peoples among whom each was first described. Eskimo, the English pattern, singles out the nuclear family. Hawaiian uses the fewest terms, so cousins are siblings. Iroquois distinguishes cross cousins, the children of a parent's opposite-sex sibling, from parallel cousins. Sudanese gives nearly every relative a word of its own. Crow and Omaha skew generations, calling a cousin on one side by a parent-generation term. In the parental generation, a system that merges father with father's brother but splits off mother's brother is called bifurcate-merging.

Here is the trap. Textbooks pair each type with a social rule: Iroquois with cross-cousin marriage, Omaha with patrilineal descent. A 2020 study of three language families found terminologies descending along the language tree with no universal direction of change, and only patchy support for 18 such pairings across 176 societies. Kinbank's authors found no evidence that bifurcate-merging co-evolves with cross-cousin marriage in Bantu languages.

Rule: read a terminology first as inherited language, and treat any inference from it to marriage or descent as a hypothesis to test against the society's own practice.

Everyone has two parents, so a society that wants bounded groups must choose. Patrilineal descent puts children in the father's lineage or clan; matrilineal puts them in the mother's, so a man's heirs are his sister's children; bilateral descent counts both sides equally and produces no corporate groups at all, only each person's own overlapping kindred. Most present-day European societies are bilateral, and the household is their largest kin group. A lineage can trace its founder; a clan claims one it cannot trace. Clans group into phratries, and a society split in two halves is organized by moieties, which usually intermarry.

Residence is a separate rule. About 71 percent of societies in the Ethnographic Atlas are patrilocal, with the couple living among the husband's kin, and about 11 percent matrilocal. Neolocal residence, a new household apart from both families, is the familiar modern pattern and a rarity in the cross-cultural record. Avunculocal residence, with the husband's mother's brother, is matriliny's way of keeping an heir near his uncle. The Mosuo are duolocal: spouses never share a household.

The trap is reading one rule from the other. Ancient DNA from Neolithic Europe showed high male relatedness, and several studies concluded descent was patrilineal. A 2025 simulation found the data compatible with patrilocal residence, but the data alone cannot show patrilineal descent, since bilateral societies with the same residence rule leave the same signature. Residence rules also persist: Nama and Cederberg communities in South Africa stayed mostly matrilocal through 150 years of patrilocal colonial pressure.

Rule: name descent and residence separately for any society, and expect genetics to speak to residence before it speaks to descent.

A ban on unions between parents and children and between siblings is close to universal; beyond that, the line between too close and allowed is drawn by each society. Westermarck's explanation was that people raised closely together from early childhood develop an aversion to sex with each other, and that because such people are usually kin, the aversion shows up as a horror of incest. He supported it with cases like Greenland, where a boy and girl raised in one household were not expected to marry. Rival positions ground the taboo in the alliances that spouse exchange creates between groups, in the role confusion sex would cause inside a household, or in the lower survival of children born to close kin.

Above the taboo sit the positive rules. Exogamy requires marrying outside a group: Morgan counted a rule against marrying within the clan among the defining features of the Iroquois gens. Endogamy keeps marriage inside a caste, faith or class. Most societies have both at different levels. Cousin marriage is practiced by more than a tenth of the world's population, and most rules favor cross cousins, who belong to another lineage, over parallel cousins, who under unilineal descent belong to Ego's own. The levirate and the sororate keep an alliance alive after a death by marrying a widow to her husband's brother or a widower to his wife's sister.

Consider a learner who reads that cousin marriage is simply harmful. A study of the Dogon and the French nobility found reduced child survival in both, alongside higher fertility for one sex in each. The practice persists where its social gains offset its genetic costs.

Rule: for any marriage rule, identify the group it keeps people in or pushes them out of, and then ask what that group gains.

Keep three words apart. Monogamy allows one spouse at a time; polygyny, one husband with several wives; polyandry, one wife with several husbands. Polygamy covers the two plural forms. In a standard cross-cultural sample, 153 of 186 societies permit polygyny, 31 are monogamous and 2 polyandrous. Westermarck made the complementary point: polygyny occurs among most peoples, yet monogamy is by far the most common form of human marriage, because plural marriage is almost everywhere confined to a minority of men. Polyandry is rare and is nowhere the exclusive form; its best-known case, Himalayan brothers sharing a wife, keeps a family's land undivided.

The popular claim that polygyny locks many men out of marriage was tested against 84 million census records from 30 countries. The association between polygyny and the share of unmarried men was negative or null almost everywhere, and the counties of the 1880 American West with Mormon polygyny had fewer unmarried men than other Western counties. A 20-year study in rural Tanzania found monogamy not consistently associated with better outcomes for children. Both findings are specific to their data, and both unsettle an assumption.

Law is a separate question. In Reynolds v. United States (1879) the Supreme Court upheld a bigamy conviction against a religious-duty defense, holding that laws may not reach belief but may reach practice; its description of polygamy as "odious" reads today as the prejudice of its era rather than as law. In 2020 Utah made bigamy between consenting adults an infraction while keeping coerced bigamy a felony. Decriminalizing is not recognizing: no United States jurisdiction licenses a plural marriage. Elsewhere, religious or customary family law can make polygyny lawful for some citizens and not others in one country.

Rule: when you read a claim about polygamy, ask whether it counts societies or marriages, and whether it is about a practice, a crime or a legal status.

Four words, four things. A matriarchy would be a society governed by women. A matrilineal society traces descent through mothers. A matrilocal society settles couples with the wife's kin. A matrifocal household is organized around a mother, often with no resident husband. Any of the last three can exist while men hold councils, chiefships and warfare, and in the ethnographic record they usually do. No society ruled by women as a class has been documented by anthropologists; what they have documented are societies where women own houses, control lineage property and choose heirs.

The confusion has a history. Bachofen's 1861 Mutterrecht proposed an early stage of mother right, with descent through mothers and a rule of women, later overthrown by father right. Morgan took female-line descent to be the archaic rule, and Engels built on both the claim that the downfall of maternal law was the historic defeat of the female sex. Marija Gimbutas read Neolithic Europe as a goddess-centered, matrilineal civilization overrun by patriarchal steppe pastoralists. Ancient DNA now supports the steppe migration; nothing supports the matriarchy.

The genetics are the interesting part. Fifty-seven genomes from Iron Age Dorset revealed a community built around one maternal line, with incoming spouses mostly male, a matrilocal pattern unexpected in European prehistory. A Neolithic cemetery in eastern China holds two maternal clans that buried separately and intermarried for ten generations. The living comparisons are the duolocal Mosuo, the matrilocal Lahu and the Tlingit. The matrilineal puzzle, a man's split between his sister's household where his heirs live and his wife's where his children live, is one reason the system is less common than patriliny.

Rule: before calling any society a matriarchy, state which of descent, residence, household focus and political rule you have evidence for.

Before judging a claim that a society is patriarchal, find the definition it uses. Morgan used "patriarchal family" narrowly, for the Hebrew pastoral household under paternal power. Later sociology used the word for any system in which men hold a disproportionate share of authority and resources, defined sometimes by law, sometimes by norms, sometimes by outcomes such as who owns land. The open literature on gendered conflict defines gender norms as rules that apportion resources, roles, power and entitlements by perceived sex, which gives a way to measure rather than to label.

Three origin theories are worth knowing in their own terms. Engels argued that as herds and other wealth built up under men's control while inheritance still passed through the mother's lineage, men overthrew maternal law to leave property to their own children, producing a monogamous family founded on male supremacy to secure undisputed paternity. Barbara Smuts applied sexual-conflict theory, arguing that women's ability to resist male coercion is weakest where men control more resources, as after the shift to agriculture, and where patrilocal residence separates women from their kin. The critique of both is essentialism: the editors of a recent collection changed their own term from "sexual" to "gendered" conflict to keep culture, sanctions and learning in view.

Notice where rivals agree. Both the materialist and the evolutionary accounts point to the same two levers: who controls subsistence resources, and whether women live among their own relatives. Sex ratios shift bargaining power, but norms decide who counts in the marriage market. Bride price is the worked example: one line of research finds it constrains women's reproductive autonomy; ethnographers of Ghana answer that within a patriarchal context it legitimizes the marriage and secures the wife's rights.

Rule: state each position fairly, name the evidence that would decide between them, and keep the comparison of societies separate from any political conclusion.

Bride price, or bridewealth, moves from the groom's family to the bride's and is the commonest marriage transaction in the record. It is usually not a sale: in many Melanesian societies the bride's family may waive it, and the marriage stays valid. What it does is legitimize the union, place the children in the husband's lineage with his rights over them, and tie two families together. Dowry moves the other way and is concentrated in the stratified societies of Europe and Asia; Westermarck listed its purposes as a return gift, the wife's contribution to the household, and her settlement if the marriage ends. Indirect dowry, such as the Germanic morning gift or the Islamic mahr, is paid by the husband but belongs to the wife. Where payments are heavy, families route a child into religious celibacy, more often the sex facing the higher payment, which a study of 352 societies reads as a way to stop siblings fragmenting an estate.

Whether bride price harms women is contested, and the contest is instructive. It can trap a woman in an unhappy marriage, because her kin may be asked to repay it on divorce. It can also protect her, because her relatives, having been paid, can bring claims against a husband who mistreats her. International bodies have called for bans, while ethnographers answer that the effects depend on age, kinship networks and residence.

Who chooses follows the same logic. Families weigh rank, wealth, labor and fertility, the things a family can evaluate; couples weigh attraction, the thing only they can judge. Betrothal, elopement and courtship all appear alongside arranged systems, and the Supreme Court has described the decline of arranged marriage as one of marriage's deep transformations. Romantic love is not new; what changed is whether it must come before the wedding.

Rule: a marriage arranged with both parties' free and full consent is an arranged marriage, and one without it is a forced marriage, whatever wealth changed hands.

The Census Bureau defines a household as all the people who occupy one housing unit, related or not, and a family as two or more people related by birth, marriage or adoption, one of them the householder, living together. A family household is maintained by a householder living with relatives; a nonfamily household is a person alone or a householder with roommates or an unmarried partner only. A subfamily is a married couple or single parent living in someone else's household, which is how a three-generation stem household appears in the data. Since 1980 the Bureau has used "householder" instead of "head" and no longer assumes the husband is the reference person.

Definitions make numbers. A Census step family counts a stepchild only relative to the householder, so the Bureau says plainly that it undercounts step families. A cohabiting couple without children is a nonfamily household, so a rise in cohabitation appears as a rise in nonfamily households. In December 2025 the Bureau reported that fewer than half of United States households were married-couple households, a statement about households, not about how many people ever marry.

Rates have the same problem. The National Center for Health Statistics (NCHS) reports, for provisional 2023, 2,041,926 marriages, a crude rate of 6.1 per 1,000 population, and 672,502 divorces from 45 reporting states, a rate of 2.4. Divide one by the other and you get the "half of marriages fail" myth, which compares this year's divorces to this year's weddings. The honest measure follows a cohort: the National Survey of Family Growth found that by age 30 about 56 percent of women and 50 percent of men had married for the first time, and by 40 about three quarters of each, with over half of adults under 50 having ever cohabited.

Rule: before quoting a family statistic, find the definition of the unit it counts and the population it divides by.

The Court wrote in Windsor that the regulation of domestic relations is a virtually exclusive province of the states, so who may marry, at what age, and how a marriage ends are state questions. The constitutional floor is the right to marry. Loving v. Virginia (1967) struck down bans on interracial marriage under both the Equal Protection and Due Process Clauses, calling marriage one of the basic civil rights of man. Zablocki v. Redhail (1978) held that a state may not condition marriage on proof that child support is paid up, and Turner v. Safley (1987) held that prisons may not forbid inmate marriages without a legitimate penological reason, listing what survives incarceration: emotional support, public commitment, religious meaning and legal benefits.

The history the Court tells is one of continuity and change. Under coverture a married man and woman were one legal person, and that person was the husband; the doctrine was abandoned as women gained property and political rights. Windsor (2013) struck down the Defense of Marriage Act's federal definition of marriage, Obergefell (2015) required every state to license and recognize same-sex marriages on four principles of why marriage is fundamental, and the Respect for Marriage Act of 2022 wrote federal recognition and interstate full faith and credit into statute. The dissents read the same history as a reason for caution; be able to state both readings.

Divorce ends a valid marriage, where annulment declares that none existed. Fault regimes required proof of a wrong such as adultery or cruelty; no-fault regimes let either spouse end a marriage on a finding of breakdown, which changes bargaining because neither can withhold consent. Several faiths keep their own procedures, so a person can be divorced in one system and married in another. Since January 1996 NCHS has not collected detailed divorce data, so national rates rest on provisional counts from reporting states.

Rule: for any marriage-law question, ask first whether it is a state rule, a constitutional limit, or a federal consequence of a state marriage.

The United Nations definition used in current research is marriage before age 18, counting formal marriages and informal unions alike, and that research cites an estimate of 640 million women and girls alive today who were married as children. Among 14 countries in one study the share of girls married before 18 ran from about 21 percent in the Philippines to about 77 percent in Bangladesh. The determinants are structural and slow-moving: a systematic test of tropical cyclones as a trigger found little measurable change in national child-marriage rates in the three years after a storm.

The legal text most often cited is Article 16 of the Convention on the Elimination of All Forms of Discrimination against Women. It gives women the same right as men to enter marriage and to choose a spouse with free and full consent, the same rights during marriage and at its dissolution, and it declares the betrothal and marriage of a child to have no legal effect, requiring states to set a minimum age and make registration compulsory. Registration is the enforcement point: an unregistered customary or religious union cannot be checked for age or consent.

Two distinctions keep a learner out of trouble. A forced marriage lacks the free and full consent of a party; an arranged marriage has it, however the match was made, and human-rights law condemns only the first. And a minimum age of 18 in a statute often comes with exceptions for parental consent, court approval or pregnancy, which is where most legal child marriages occur. In the United States the age and the exceptions are set by each state and keep changing.

Consider a reader who sees bride price and child marriage listed together as harmful practices. The harm finding for child marriage rests on a child's inability to consent; the finding for bride price concerns adults and is contested in the ethnography.

Rule: judge a marriage practice first by whether each party could and did consent, and date any legal threshold you quote.