The Family as a Historical Form

Almost every claim about the family — that it is in decline, that it has been transformed, that something has been lost — is a comparison. And the comparison is almost always with a particular arrangement that people believe was the norm for centuries.

It was the norm for about thirty years, in a small number of countries, under conditions that no longer exist.

And the account of what preceded it, which was standard in sociology for decades, turned out to be wrong — demonstrated wrong by counting, which is why this lesson is also a demonstration of what 7.6.1 said about description.

The parish listings, and the theory they demolished.

For most of the twentieth century, sociology told a story about the family that went like this.

Before industrialisation, people lived in large extended households: several generations under one roof, married sons and their wives, unmarried siblings, grandparents, working a farm or a trade together. Industrialisation broke this up. Factory work required mobility; the family lost its productive functions to the workplace; and what remained was the small, mobile, isolated nuclear family — parents and children, cut off from wider kin, specialising in emotional functions.

It was a good story. It fitted the theory of modernisation. It explained a great deal.

And in the 1960s a group of historians in Cambridge began counting.

They had a source nobody had systematically exploited: listings of inhabitants compiled by parishes, household by household, name by name, over more than two centuries. They coded them.

The mean household size in England between the 1570s and the 1820s was about 4.75 people.

It did not decline over the period. It was strikingly stable across two and a half centuries, across regions, and across the entire process that was supposed to have transformed it.

And the household composition was not what the theory required. The overwhelming majority of households were nuclear — a married couple and their children. Households containing kin beyond that — a grandparent, a married sibling — were a small minority, in the region of one in ten.

The extended family that industrialisation was supposed to have destroyed had not been there to destroy.

Now the more interesting half: what those households actually contained that ours do not.

Servants. Not a small elite phenomenon — a large share of young people in early modern England spent years living and working in another household, between leaving their parents and marrying. They appear in the listings as household members. Households were not small because families were small; they were the size they were partly because they included non-kin.

Very few grandparents. In a high-mortality regime, most people did not live to see their grandchildren reach adulthood. The multi-generational household was demographically difficult before it was culturally anything.

Late marriage and a great deal of never marrying. Across a broad region of north-western Europe, women married in their mid-twenties, men slightly later, and a substantial fraction — sometimes one in ten or more — never married at all. Couples set up their own household on marriage rather than joining a parental one, which is precisely why households were nuclear.

And step-families everywhere. Not because of divorce, which barely existed, but because of death. A large share of marriages were remarriages, and children commonly grew up with a step-parent and half-siblings.

The pre-industrial family was small, nuclear, formed late, frequently reconstituted after a death, and contained people who were not relatives.

None of that is what the theory said, and all of it was in the parish records the whole time.

What is being compared to what?

Three objects get called "the family" and they are different.

The household — people living under one roof, sharing some resources. This is what censuses count, and it is a residence unit.

The family — a set of people related by descent, marriage or adoption, whether or not they live together.

And kinship — the whole system of recognised relationships, obligations and descent reckoning, which extends far beyond both.

A great deal of confusion follows from measuring one and concluding about another. "Households are getting smaller" is a fact about residence. It is compatible with kin contact increasing, decreasing or staying the same — which is 7.2.2's operationalisation problem in the most emotionally loaded domain in the discipline.

And there is a second confusion, which is 7.6.4's.

A cross-sectional count of households understates how many people ever live in a given arrangement. Households have a developmental cycle: a couple forms, has children, the children leave, a parent is widowed and moves in, the household dissolves.

So a count showing that only a small percentage of households at any moment contain three generations is entirely compatible with a majority of people living in a three-generation household at some point in their lives. The snapshot and the life course give different answers, and only one of them is usually reported.

The variation

How much the form varies, and what is actually universal.

Something recognisable as a family exists everywhere. What varies is nearly everything about it.

Residence. Couples may set up a new household (neolocal), join the husband's family (patrilocal), join the wife's (matrilocal), or maintain separate residences. The north-western European pattern of neolocal residence at late marriage is one arrangement among many, and it is the one that produced the small nuclear households in the parish listings.

Descent. Membership and inheritance may be reckoned through the father's line, the mother's line, or both. Matrilineal systems are not matriarchies — authority typically rests with a woman's brothers rather than her husband — and they demonstrate something important: that the co-resident biological parents plus children is not a universal unit. In several documented systems, a man's primary obligations are to his sisters' children rather than to his own.

Marriage form. Polygyny is permitted in a majority of the societies in the standard ethnographic samples and practised by a minority of men within them, for the arithmetically obvious reason. Polyandry is rare and real.

And the joint family, which matters because it is the standard contrast case for South Asia. The classical joint household — married brothers and their families under one head, holding property in common — was always a minority arrangement at any given moment, and it was strongly associated with having property to hold jointly. Landless households were rarely joint.

Which is the developmental cycle point again: a household passes through a joint phase and then partitions, so a snapshot and a life history give very different pictures — and the "decline of the joint family" is a much more complicated claim than it appears, because the baseline was never what it is assumed to have been.

The thirty-year family

The arrangement people mean by "traditional", and the conditions it required.

Male breadwinner in stable full-time employment. Wife not in paid work, or working intermittently. Married before children. Two or three children. Own home. Lifelong.

This arrangement was widespread in several rich countries for roughly the three decades after the Second World War, and it was historically unusual in every respect.

It required a particular wage structure. A single wage sufficient to support a household — the family wage — was itself the product of a political settlement and of union bargaining, and it did not exist for most workers before it and has not since. In the pre-industrial households of the story, everyone worked, including children and including wives.

It required particular housing. Mass owner-occupation was constructed, in most countries, through specific mid-century policy — subsidised mortgage insurance, public housebuilding, tax treatment — much of it allocated unequally (see 8.4.3).

It required a particular age structure and mortality regime. Enough life expectancy for lifelong marriage to mean fifty years rather than fifteen.

And it required the exclusion of married women from many occupations, by explicit rule.

This last point is worth stating concretely, because it is frequently forgotten. Marriage bars — requiring women to resign on marriage — operated in teaching, the civil service, banking and many private firms across a range of countries into the middle of the twentieth century, and in some cases considerably later. The arrangement in which married women were not in the labour force was not simply a preference. In substantial sectors it was a condition of employment.

So the baseline against which family change is measured was a thirty-year arrangement resting on a particular wage settlement, a particular housing policy, a particular mortality regime and an explicit legal exclusion.

Comparing anything to it and calling the difference decline is comparing to an anomaly — which is 8.6.1's problem about baselines, in another domain.

What the theory said, and what survives

The functionalist account, and its two claims.

Parsons's argument was that the modern family had lost most of its historic functions — production, education, welfare, religious instruction — to specialised institutions, and retained two that could not be transferred.

The primary socialisation of children: the formation of a person capable of participating in society, which he located in the early years and in the family specifically.

And the stabilisation of adult personalities: the family as the place where adults are restored, sustained and held, against the impersonality of a differentiated society.

What survives of this.

The functional differentiation observation is sound and important. Households did lose productive, educational and welfare functions to firms, schools and states — that part of the story is historically correct, and it is what makes the family's remaining functions so concentrated.

The historical narrative is wrong, as the parish listings showed: there was no large extended household to be broken up in the region the theory was built from.

The "isolated" nuclear family is wrong empirically. Studies of kin contact from the 1950s onwards consistently found extensive, frequent and materially significant contact with extended kin — help with childcare, money, housing, illness — which the theory had written out.

And the account omits everything the family also does. It is a site of production (see 8.5.3), of unequal distribution (see 8.1.2 on pooling), of the transmission of advantage (see 8.3.3) — and of violence, which the functionalist literature is almost entirely silent about and which subsequent research established as pervasive rather than exceptional.

The feminist critique made the omission the point: an institution described as performing warmth and stabilisation was also the place where a great deal of unpaid labour was extracted and a great deal of harm occurred, and a theory that could describe the first without the second was describing an ideal rather than an institution.

Because almost every public argument in this area is conducted against a baseline that was neither traditional nor typical.

Four questions.

Compared to when? If the answer is the mid-twentieth century, the comparison is with a thirty-year arrangement that required a family wage, a housing settlement and a legal exclusion.

Household, family, or kinship? Shrinking households are not the same as weakening kinship, and the data on the two are different data.

Snapshot or life course? A small proportion of households in an arrangement at one moment is compatible with a large proportion of people passing through it.

And which society? The nuclear household of north-western Europe is one arrangement among many, and it is the one that most family theory was unknowingly built from (see 6.8.1).

One thing to carry from the parish listings, beyond the substance.

A theory held for decades by serious people was overturned by counting things that had been written down and filed for two hundred years. Nobody had looked, because the theory made looking seem unnecessary — the extended household was known to have existed, so what would be the point?

That is what 7.6.1 meant by description being undervalued, and it is the single best example of it in this course.

Sociology told a story: large extended households before industrialisation, broken up into isolated nuclear families by it. Parish listings covering England from the 1570s to the 1820s showed mean household size stable at about 4.75 across two and a half centuries, with the overwhelming majority of households nuclear and kin beyond the nuclear family present in something like one household in ten. The extended family had not been there to destroy.

What those households did contain: servants — a large share of young people lived and worked in another household before marrying; very few surviving grandparents, because of mortality; late marriage and substantial never-marrying, with couples forming their own households on marriage; and step-families formed by death rather than divorce, since a large share of marriages were remarriages.

Three objects are called the family: the household (a residence unit, which is what censuses count), the family (relationships regardless of residence), and kinship. And the developmental cycle means a cross-sectional count of households understates how many people ever live in a given arrangement — the snapshot and the life course give different answers.

The variation is enormous: neolocal, patrilocal and matrilocal residence; patrilineal, matrilineal and bilateral descent — matrilineal systems demonstrating that co-resident biological parents plus children is not a universal unit; polygyny permitted widely and practised by a minority. And the South Asian joint household was always a minority arrangement at any moment, strongly associated with having property to hold in common.

The "traditional" family — male breadwinner, non-employed wife, marriage before children, own home, lifelong — was widespread for roughly thirty post-war years and required a family wage, a constructed housing settlement, a particular mortality regime, and marriage bars that made married women's exclusion from many occupations a condition of employment rather than a preference.

Parsons's functional differentiation observation is sound; his historical narrative is wrong; the "isolated" nuclear family is contradicted by the kin contact evidence; and the account omits production, unequal distribution, transmission of advantage and violence.

And a theory held for decades was overturned by counting records that had been filed for two hundred years, because the theory made looking seem unnecessary.

Household / family / kinship — a residence unit; relationships regardless of residence; the whole system of recognised relations and obligations.

Nuclear / extended / joint / stem household — couple and children; plus other kin; married siblings under one head; one married child remaining with the parents.

Neolocal / patrilocal / matrilocal residence — new household on marriage; joining the husband's; joining the wife's.

Patrilineal / matrilineal / bilateral descent — reckoning membership and inheritance through one line or both.

Developmental cycle — the phases a household passes through, which make snapshots misleading about life courses.

Life-cycle service — young people living and working in another household before marriage.

European marriage pattern — late marriage, substantial permanent celibacy, and neolocal residence across north-western Europe.

Family wage — a single wage sufficient to support a household; a political and bargaining product, not a natural feature.

Marriage bar — a rule requiring women to resign on marriage.

Isolated nuclear family — Parsons's model, contradicted by evidence of extensive kin contact and support.

One — reconstruct three generations. For your own family, write down the household composition of each generation at the same age. Note what changed and what did not.

Two — check the baseline. Take any claim about family decline and identify the year it is comparing to. Then find out what was legally and economically true of that year.

Three — count servants. Look at a nineteenth-century census return for any street, if your country's archives are digitised. Note how many households contain people who are not relatives.

Four — separate the measures. Find one statistic about household size and one about contact with relatives in your country. Ask whether they support the same conclusion.

Five — apply the cycle. Take the proportion of households in your country currently containing three generations. Then estimate what proportion of people will live in one at some point. The two numbers are very different, and only one is ever quoted.