Estimating a Town's Population Across a Register Gap
A local historian is studying a market town, 1601–1700, from its parish registers. The registers are continuous except for a gap from June 1643 to February 1646. Baptisms average 38 per year in the 1620s and 24 per year in the 1660s. Burials average 22 per year, but 1644 records 97. Two later sources survive: a 1662 hearth tax listing 412 households and a 1676 ecclesiastical census recording 1,140 communicants. A nearby parish, never garrisoned, shows baptisms falling only from 30 to 27 across the same decades. The historian proposes to estimate the 1650 population by applying standard multipliers to the 1662 and 1676 figures, then back-projecting to 1650 using the baptism series, and to conclude that the town's population fell by about 30 percent because of the civil war. A referee must decide whether the paper's central claim is supported.
1. Assess the proposed method: multipliers on the 1662 and 1676 sources, then back-projection to 1650. What does it actually deliver?
Read the worked answer
Step 1 - the multipliers. A household multiplier of roughly 4.3–4.7 on 412 households gives c. 1,770–1,940 people in 1662, with the important qualification that hearth-tax returns typically list only chargeable households, so the poor and exempt are undercounted; true population is likely higher. A communicant multiplier of about 1.5–1.8 on 1,140 gives c. 1,700–2,050 for 1676. The two independent sources land in the same broad band, which is reassuring but not precise - the honest output is 'c. 1,700–2,000,' not a figure.
Step 2 - back-projection. This is where the method fails for the intended claim. Back-projection assumes a closed population changing only by births and deaths. A market town in a garrisoned war zone is the worst case for that assumption: troops, refugees, and displaced craftsmen move in and out, and baptisms measure baptisms, not births (omitted emergency baptisms, nonconformist families, stillbirths).
Verdict: the method supports a population bracket around 1662–1676 and a plausible range for 1650, but it cannot isolate a 30 percent war-caused decline. The 38→24 baptism fall is consistent with registration change, nonconformist separation, or migration, not only with deaths.
2. How should the historian handle the 1643–1646 gap and the 97 burials recorded in 1644?
Read the worked answer
Do not interpolate the gap as if it were normal. Filling three missing years with the 1620s average manufactures data and hides exactly the event of interest. Treat the gap as a documented fact about the record and say so in the text.
Test the 1644 spike before interpreting it:
- Seasonality and age structure. Plague and typhus produce summer–autumn burial clusters concentrated in adults and young adults; combat deaths skew heavily male and young adult. Marginal notes ('pestis,' 'soldier') in the register are the first thing to read.
- Burial-to-baptism ratio in the surrounding years. A mortality crisis followed by a baptism trough and then a rebound is typical of epidemic; a permanent fall with no rebound suggests flight or registration collapse.
- Comparators. The ungarrisoned neighbouring parish is a control: if its burials also spiked, the cause is regional, not the garrison.
Independent sources for the gap years: bishop's transcripts (often a separate copy of the same register, and sometimes surviving where the original is damaged), wills and probate inventories proved in the church court, churchwardens' accounts, muster or quartering records, and lay subsidy or hearth-tax material before and after. These distinguish epidemic mortality from troop movement, flight, or a register simply not being kept because the incumbent had been ejected.
3. The historian wants to keep the conclusion 'population fell by about 30 percent by 1650 because of the civil war.' What is the logical problem, and what would a defensible conclusion look like?
Read the worked answer
The logical problem is the baseline. There is no direct observation of the town in 1650 - the nearest anchors are 1662 and 1676, on the far side of the gap. The '30 percent' is derived from comparing 1620s baptisms with 1660s baptisms, which is a comparison of two decades' registration practice, not a measurement of population at two dates. A garrison town is precisely where registration practice is most likely to change: a separatist congregation gathers, an incumbent is ejected, baptisms move to another record. The comparison parish (30→27) shows what an unperturbed series looks like and makes the town's 38→24 look dramatic, but does not by itself attribute the difference to mortality rather than migration or recording.
A defensible conclusion would:
- State a population bracket for the 1660s–1670s (c. 1,700–2,000) and note that no comparable baseline exists for the 1620s, so the size of any fall is unknown.
- Assert what the record does show: a mortality crisis in 1644 of at least 97 recorded burials - a minimum, since the register is incomplete around it - and a sustained fall in baptisms relative to a control parish.
- Attribute causes with explicit uncertainty: 'consistent with epidemic disease and with war-related displacement; the two cannot be separated on these registers alone.'
- Name the source that would settle it - a return of households for the town in the 1640s or 1650s, or probate evidence for adult mortality.