The Cambridge Group for the History of Population and Social StructureAlice Reid & Jim Oeppen
Looking backwards in time gives a mistaken impression that family sizes in the past were larger than they actually were. This blog explains why this happens, and explores the differences between the picture of the past painted by genealogies and the past as it actually was.
Looking backwards at our families
Alice’s grandmother, Margaret, had six children, of whom five survived to adulthood. She had 14 grandchildren and (so far) 25 great-grandchildren. She also had two sisters, Kathleen and Moira. Moira had two children, four grandchildren and six great grandchildren. Kathleen remained single and childless throughout her life. On average, the three sisters (Margaret, Kathleen and Moira) had 2.7 children apiece.
Kathleen, Moira and Margaret with their mother Agnes (also known as Nan) in 1929. Family photograph, courtesy of Colin Reid.
Of the seven offspring in the next generation who survived to adulthood, five of them came from a family of six, and two from a family of two. If you were to gather them all in a room and ask how many children their mothers had (imagine they were not related and therefore did not worry about whether or not siblings should all answer the question), the answer would be 4.9 children. The view from the children’s point of view is very different, because there are more of Margaret’s children to remember their big family. The fact that Kathleen had no children means that her family size (of zero) cannot be represented in a calculation of mothers’ family size as reported by children.
In the next generation the difference is larger still, with the grandchildren’s point of view suggesting that their grandmothers’ generation had 5.2 children on average, nearly double the real number of 2.7.
Looking back at previous generations of our own families can therefore give an inflated view of how large family sizes were in the past, and can produce distorted impressions of families and family formation.
Alice’s grandmother Margaret (centre), with her surviving children and her husband. Family photograph, courtesy of Colin Reid.
Family history and genealogy
Demography takes a “descendant” viewpoint. The average family size is calculated from the mother’s viewpoint – the 2.7 children in the example above, not the ascendant 5.2. By contrast almost all genealogies are ascendant: i.e. a survivor works backwards, recording the generations in their main line of ascent. (Descendant genealogies select a person in the past and follow their kin forward in time – a future blog will discuss Chinese genealogies, which are usually descendant.) The extent to which a genealogist follows collateral kin in each generation, such as aunts and uncles etc., is variable – depending on the available records and enthusiasm.
King Henry IV of England, in armour and wearing crown, wielding his sword so as to cut Richard II from the Plantagenet family tree (England, 1460 – c. 1470). British Library Harley 7353.
Campop’s work on reconstructing the demography of English families allows us to calculate the ascendant bias in family size from 1550 to 1850 (i.e. the extent to which ascendant genealogies overstate family sizes). The simple formula that links the averages for the ascendant and descendant views has been known for over a century.
To simplify the picture, we start by removing the effect of celibacy (women remaining unmarried) and mortality. Assume that every woman married, and both she and her husband survived to at least her 50th birthday. The descendant average number of children over the period varied between about four and six children, but the ascendant view adds 1.5 to two extra children. This is like comparing the average number of children from Margaret and Moira (four) with the average from their children’s point of view (5.2).
Including women such as Kathleen in the example above, who did not marry or have children increases this bias still further. Celibacy in the past among females surviving to age 50 is thought to have been about 10-15 percent. Adding these women with no descendants to the calculation raises the ascendant bias to about 2.5 children. Similar biases have been found for Basque villages 1800-1969, Brazil 1960-2000, France 1830-1896, the USA 1867-1955, and a variety of late 20th century, high fertility populations.
Genealogy showing the descendants of Adam and Eve (London, 1611). British Library C.35.l.13.(2).
So, women with descendants, who are more likely to appear in genealogies, are not typical of women in general. Their experience should not be used to characterise the experience of the overall population.
Nevertheless, these women with descendants did exist, and it is also worth considering how they managed to fit larger than average numbers of children into their child-bearing histories.
The maximum reproductive span for a woman is 35 years (between the ages of 15 and 50). But women in the British past were aged about 25 when they married for the first time (see blog on marriage), and the typical age at last birth in a non-contracepting population of women surviving to age 50 is 41 years, reducing the average fertile period to 26 years.
Tony Wrigley and colleagues at Campop calculated that average inter-birth intervals were 2.5 years: typical of a population with long breast-feeding. Thus, women in an ascendant genealogy would need an extra 6.25 years of reproduction. They must have married young, lived to 50, or had short birth-intervals (or multiple births), or all three.
Children born to Andrew and Janet Gray, great great grandparents of Agnes (Nan) in the photograph above. Janet’s young age at marriage, survival beyond age 50, and very short birth-intervals enabled her to have 16 singleton births. Image courtesy of Colin Reid.
How do we know?
This knowledge uses ‘family reconstitution’: the reconstruction of families by linking the baptisms, marriages, and burials recorded in parish registers. This process starts with a marriage and locates the baptisms of bride and groom to establish their birth dates and age at marriage. The births of their children are identified, enabling the age of the mother at birth to be calculated. Finally, the deaths of husband and wife are located in the records, yielding age at death.
The same process is undertaken for the marriages of each of the children of the original couple, making inter-generational comparisons possible. Campop created a number of family reconstitutions for a variety of communities across England. These have to be treated very carefully to yield accurate demographic measures, but they are our best source of information about the population of England between the mid-16th and mid-19th centuries.
The bias in ascendant genealogies can be calculated by comparing the average number of children per woman using all women in the population (the descending point of view), with the sibship sizes of those women who had children. In other words, by performing a similar comparison to the example in the first section of this blog.
Further reading
E. A. Wrigley, R.S. Davies, J.E. Oeppen, and R.S. Schofield, English Population History from Family Reconstitution, 1580–1837 (Cambridge University Press, 1997).
