← Council on Local Relations · Primer 01 · v1.3 · last verified 2026-09-26Stamps: [C] certifiedCond. reconstruction or synthesis[P] projection
The evolution of the family · 10,000 BCE to 2025
The Long Flat Line
For almost all of human history, the family looked the same from one century to the next. Most children died young, most parents farmed, and the population barely grew. Then, in about two hundred years, nearly everything changed at once.
Growth rate, 10,000 BCE–17000.04%a year. One doubling every ~1,660 years.
Share of the 12,000 years1.9%is the stretch from 1800 to 2023.
Share of the growth88%of all people added since 10,000 BCE came in that stretch.
Children dying before age 543% → 3.7%world, 1800 to 2024.
I · Population
Twelve thousand years in one line
This is the world's population from the end of the last Ice Age to the UN's 2100 projection, drawn to scale. Across the full span, the breakout after 1800 is a near-vertical wall. Switch to the log scale to see the slow climb before it. Switch the range to 1700–2100 to see the wall up close.
World population, millions
Estimated (to 2023)UN medium projection (2024–2100)
Sources: before 1950, HYDE 3.3 (2023) and Gapminder v7 through Our World in Data, "Population (historical)" Cond. These are model estimates; there was no census. From 1950, UN World Population Prospects 2024, July 1 estimates [C]. Projection is the WPP 2024 medium variant, which peaks at 10.29 billion in 2084 and ends at 10.18 billion in 2100 [P]. Growth rates and shares were computed by the Analyst from these series.
United States resident population, decennial census, millions
Source: U.S. Census Bureau, 1990 CPH-2 Table 4, "Population: 1790 to 1990"; Historical Population Change Data 1910–2020; Vintage 2025 estimate, 341,784,857 on July 1, 2025 (NA-EST2025-POP) [C]. The U.S. grew about 64× from 1800 to 2025, against about 8× for the world.
II · The old baseline
The pre-modern family
Before about 1800, family life changed slowly because the conditions under it barely moved. Scholars reconstructing those centuries keep finding the same few facts.
Nearly half of children died before adulthood. A review of 43 historical populations and 20 hunter-gatherer groups found that about 27% of infants died in their first year and about 47.5% of children died before puberty (Volk & Atkinson, 2013) Cond.. The world's estimated under-five death rate was still about 43% in 1800 and about 41% in 1900 Cond.
Short lives were mostly short childhoods. World life expectancy at birth was about 28.5 years in 1800 (Riley, 2005) Cond.. Adults who survived childhood often lived into their fifties and sixties. The average was low because so many children died.
Households were the economy. In England and Wales, historian Peter Laslett found mean household size held at about 4.75 or a little below from the 1500s to 1911 Cond.. The first U.S. census, in 1790, recorded 5.79 people per household [C]. More than nine in ten Americans lived outside towns (U.S. urban share: 6.1% in 1800) [C]. The family farm or shop was the workplace, the school, the hospital, the retirement plan and the insurance policy.
Large families replaced people. They did not multiply them. American white women averaged about 7 births in 1800 (Haines, reconstructed) Cond.. With half of children lost, high fertility mostly held the population steady. That balance is why the line in Section I stays flat for so long.
III · The mechanism
Deaths fell first. Births followed.
The population explosion did not come from families having more children. It came from children surviving. Death rates fell first, and birth rates took generations to follow. In the gap between the two, population surged. Demographers call this the demographic transition. The three panels share a time axis.
Children dying before age 5, world, %
Gapminder (2015) to 1949 Cond.; UN IGME from 1950 via OWID; 2024 = 37.4 per 1,000, 4.9M deaths (UNICEF, Mar 2026) [C]
Life expectancy at birth, world, years
Riley (2005) to 1949 Cond.; UN WPP 2024 from 1950; 73.2 in 2023 [C]
Births per woman, world
UN WPP 2024 estimates 1950–2023 [C]; medium projection 2024–2100, dashed [P], falling to 1.84 by 2100. Peak 5.31 in 1963; 2.25 in 2023; replacement ≈ 2.1.
Deaths dropped first and births took generations to follow. That lag is where the extra seven billion people came from.
IV · The American household, 1790–2025
Smaller, later, less often married, more often alone
Every panel shares the same time axis, 1790 to 2025, so a change in one panel can be read against the others. Points are the years the source actually measured. Lines only connect them. Hover or tap a point for its value and source.
V · Structure
What the household used to do, and where it went
The shrinking household is one side of a larger shift. Jobs the family once did inside its own walls moved to firms, schools and government. This is a structural description, not a verdict. Each move bought something real, often survival itself, and each one changed what a family is for.
ProductionFarm and shop moved to wage work. The share of Americans in cities rose from 6.1% (1800) to 80% (2020).
EducationHome and church moved to the public school. Compulsory attendance began in Massachusetts in 1852 and reached every state by 1918.
Old-age supportSupport from grown children moved to public pension. The Social Security Act passed in 1935.
ChildcareMother and kin moved to school and paid care. 73.9% of U.S. mothers with children under 18 were in the labor force in 2025.
SurvivalFolk care and luck gave way to clinics, vaccines, clean water and sanitation. U.S. infant deaths fell from about 100 per 1,000 (1900) to 5.6 (2023).
VI · The ledger
Progress made, measured
Each row gives a starting value, the latest value, the source and a stamp. "Cond." marks historical reconstructions and scholarly syntheses, which are the best available estimates, not counts. "[P]" marks projections. Read the start and end years together, because they differ from row to row.
Measure
Then
Now
Source
Stamp
VII · The work ahead
What the numbers put in front of us
In two centuries, the share of children dying before age five fell from about 43% to under 4%. That work is not finished, and the change also raised new questions that no earlier generation had to answer. These are the open items, stated as the data states them. The reader concludes.
4.9Munder-5 deaths, world, 2024 · [C]
The survival work is unfinished
The world under-five death rate is down to 37 per 1,000. That still means about 4.9 million children died in 2024 (UN IGME). About 824 million people, 10.1%, lived in extreme poverty in 2024 (World Bank Poverty and Inequality Platform, release of Sept 22, 2026) [C].
1.60U.S. births per woman, 2024 · record low · [C]
Below replacement
A population holds steady at about 2.1 births per woman. In 2023, 130 of 237 countries and areas were below that level, and they held 67.3% of the world's people (Analyst computation from UN WPP 2024 estimates) [C]. The UN projects the world total peaks at 10.29 billion in 2084 and then declines [P].
late 2070s65+ outnumber under-18s, world · UN WPP 2024 medium · [P]
More grandparents than grandchildren
The share of people 65 and older doubled from 5.1% (1950) to 10.0% (2023) [C]. Pensions and elder care were built for a wide base of young workers. Who carries the old when the base narrows is a question every household and every budget will meet.
29.5%U.S. one-person households, 2025 · [C]
More people living alone
This share was 13.1% in 1960. Married-couple households fell from 78.2% of all households (1950) to 46.6% (2025). Median age at first marriage is now 30.8 for men and 28.4 for women.
70.4%U.S. children living with two parents, 2025 · [C]
Children's households have changed
This was 87.7% in 1960. Births to unmarried mothers rose from 5.3% (1960) to a 41.0% peak (2009) and stood at 39.5% in 2024. The CLR standard applies here: systems, not scapegoats.
Held pending re-verification from primary census microdata: the multigenerational household series. It will appear here once re-run and certified.
Everything before 1950 at world scale (population, life expectancy, child mortality, literacy, poverty, urban share) is a model reconstruction. Treat it as the best available estimate, not a count.
U.S. household size: before 1940 the census used an older "family" definition. 1940 and 1950 are decennial census counts. 1947 onward is the CPS survey. 2025 uses new Vintage 2025 population controls, so part of the 2024–25 change is methodological.
U.S. fertility: 1800 and 1900 are reconstructions for white women only. 1957 is adjusted for under-registered births. 1976 onward uses registered births.
Births to unmarried women before 1980 are estimated from states that reported marital status. NCHS changed method again in 2017.
Women's labor force: 1900 is from the June 1900 census, women 15 and older, on the older "gainful worker" concept (HSUS-1975 Series D 58). It was verified against the page image on Sept 26, 2026 [C]. 1948 onward is the CPS civilian labor force, ages 16+. The two are close but not identical measures.
One-person household shares for 1790–1950 are from HSUS-1975 Series A 343 and were verified against the page image and the underlying counts on Sept 26, 2026 [C]. The 1950 row is a March CPS figure that the table flags as not strictly comparable with the census years. 1960 onward is CPS table HH-4, which has no 1930 or 1940 row, so A 343 is the only national source for those years. The dip from 7.9% (1930) to 7.1% (1940) is shown as published. The HSUS headnote says the 1930 and 1940 household definitions were the same, so the dip is not a definition change.
Recent national divorce and marriage rates are provisional and exclude several states, including California.
The world extreme-poverty series changes method. 1820 uses Moatsos's cost-of-basic-needs measure. 2024 uses the World Bank $3.00/day line (2021 prices, release of Sept 22, 2026). The WDI database still shows 10.4% for 2024.
World urban share, 2025: 58% under national definitions. Under the UN's new harmonised method, 45% live in cities and 36% in towns. The two are not comparable.