Counting People Who Were Never Counted
Before the nineteenth century almost nobody counted populations for the sake of knowing them. States counted for taxing and conscripting, which means every pre-modern number was produced by a process the counted had excellent reasons to defeat. Yet we now have reasonably confident English population figures back to 1541 — and the technique that produced them is one of the discipline's genuine triumphs.
Family Reconstitution
Parish registers recorded baptisms, marriages, and burials for religious reasons, with no demographic intent whatsoever. Family reconstitution (Louis Henry's method, industrialized by the Cambridge Group) links those three registers back together person by person: this baptism and that marriage and this burial are one woman, and these baptisms are her children.
Do it across a parish and you recover what no source ever recorded: age at first marriage, birth intervals, completed family size, infant mortality, remarriage speed after widowhood. The labour is brutal — thousands of hand-linked records per parish — and the payoff reversed a central assumption of European history.
The finding: north-west Europe had a distinctive marriage pattern — late marriage (women at 24–26) and a large proportion never marrying at all. Since fertility within marriage was largely uncontrolled, the age of marriage was the population's brake. Push marriage later or leave more women unmarried, and births fall — no contraception required. Population responded to the economy through the wedding, because couples waited for a viable household. That is a control loop nobody designed and nobody noticed, discovered entirely by re-linking records kept for a different purpose.
Every Pre-Modern Number Is a Negotiation
| Source | Made for | Systematic distortion |
|---|---|---|
| Tax rolls | Extraction | Under-count: people hide |
| Muster rolls | Conscription | Under-count: sons vanish |
| Grain requisition lists | Supply | Under-report the harvest |
| Parish registers | Religion | Miss the unbaptized, dissenters, the mobile |
| Bills of mortality | Plague policy | Cause of death is a searcher's guess |
The rule: ask what the count was for, and the direction of the bias follows. A tax roll undercounts because the purpose of a tax roll, from below, is to be defeated. Nobody was confused about this at the time; only later readers mistake the roll for a census.
The Multiplier Problem
You have 400 hearths in 1377 and want people. Multiply by a household size — but which? Estimates run from 4.0 to 5.5, and that spread alone swings your answer from 1,600 to 2,200: a 38% range, before any other error. Then compound it with unknown exemption rates and evasion, and confident-looking population figures for pre-modern Europe rest on a chain of assumptions any link of which can move the total by a third.
The honest practice: report the assumption with the number, and give the range rather than the point. "About 1,600–2,200, assuming 4.0–5.5 per hearth and no allowance for evasion" is a better sentence than "1,900" — because it tells the reader what would have to change for the figure to move, which is the only thing that makes a number arguable rather than decorative. A spurious decimal place is not precision; it is a claim about evidence you do not have.
Linking the Dead Back Together
The animation opens on three columns of raw entries — baptisms, marriages, burials — kept by a parish priest with no interest in demography whatsoever. Nothing about them looks like data.
Then the links begin drawing. A baptism entry reaches across to a marriage; the marriage reaches to a cluster of baptisms; a burial closes a line. One family assembles out of scattered records, then ten, then the parish. What was three lists becomes a population with structure.
Now the measurements appear, none of which any register ever recorded: age at first marriage, birth intervals, completed family size, infant mortality.
The final movement makes the discovery visible. A marriage-age dial appears beside a population curve. Drag marriage age from 20 to 26 and the births thin out and the curve bends downward — no contraception anywhere in the model. The animation holds on it: in north-west Europe the wedding was the brake, because couples waited for a viable household, so population answered the economy through the age of marriage.
Nobody designed that control loop and nobody at the time noticed it. It was recovered four centuries later by re-linking records that had been kept for the salvation of souls.