What the dates do

Read down the date columns instead of one life at a time and things appear that no single life shows: three wedding-day regimes, a July peak in the deaths, and a bride who gets seven years older, once.

Patterns, not people. Every figure here is counted from the dates gathered in this history and can be recounted; each says how many records it rests on, because several rest on few.

A family history is usually read one person at a time. Read down the columns instead, every marriage date, every death date, every birth, and things appear that no single life shows.

They married on Sundays, and then they stopped

Weekday of marriage, for every union in this model with a full date:

Three regimes, and each is nearly complete in its own era.

Before 1850, more than half of all marriages happen on a Sunday: the congregation was already assembled, the minister was already there, and a couple far out in the veld married on the day the district came to church.

Then Sunday collapses to a single instance in fifty years and Monday takes over. And from 1900 it is Saturday, which is where it has stayed.

Nobody in this family decided that. It is the church's own calendar, and later the working week's, printed through one family's dates. It is also usable: an undated marriage here can be placed in its era by the day of the week alone.

Half of them reached their seventies

For everyone born between 1850 and 1899 who lived past fifteen:

"Everybody died young back then" is wrong here, and it is wrong in a specific way. Childhood was the danger. Once past it, this was a long-lived population, and the men outlast the women by five years, the reverse of the modern pattern, and what you would expect where childbirth is the risk.

The infant deaths are not in that figure and cannot be. Visible under-five deaths here run at one to three per cent, which is a floor and not a rate: the children who died before anyone wrote them down are missing, and the reused names are where they show through.

The Karoo winter is in the death dates

Month of death, every dated death in the model (n = 551):

An even spread would put about 46 in each month. July carries 65, some forty per cent above the average month, and March, at 33, is the quietest.

July is midwinter here. The Koup gets hard frost, the wind comes off the Swartberg, and a cold house at 900 metres is a cold house. The peak is old people and small children, and it is the climate showing up in a column of dates.

The teenage bride disappears, once

Women's median age at first marriage, by the decade they were born:

One step, not a slope. Between the women born before 1850 and those born after it, the median rises nearly seven years, and then stops. The third cohort is flat against the second, and with only twelve women in it nothing should be read into that column at all.

So the change is real, it is large, and it happens once, in the generation that grew up as the district filled in and the church came closer. Frontier daughters married at eighteen; their daughters married at twenty-five.

Twins are rare, and they are how you catch a duplicate

Six sets of twins here, across five surnames, and every one of them carries unrelated names: Hendrik and Beatrix Annis; Bertram Henry and Sophia Johanna Mocke, born to the same couple twenty-six months later; Matthys Andries and Ockert Almero; Johanna Aletta Jacomina and Stephanus Johannes; Elizabeth Maria Susanna and Johanna Jacomina Petronella Catharina; Catharina Isabella and Hester Maria Josina.

Which is why six other same-day pairs are not twins at all. Andries Petrus beside Andrew Peter. Gert Adriaan beside Gerrit Adriaan. Aletta Francina beside Aletta Franzina. Cornelis Andries beside Cornelis Andreas. And twice the name is not even varied: a Pieter Hendrik Mocke and a Herbert Andrew van der Hoven each stand twice on their own birthday, the second time with the surname capitalised differently. Those are one child written twice, and the rule that catches them is that a family does not give two living children the same name.

A name is given again

The rule has a second edge, and it is the older one. When a child died, the name went to the next child born.

Jacobus Nicolaas, born 12 November 1755, died 1769. A Jacobus is baptised 8 October 1769, the same year, to the same parents.

Dina Johanna, born 1762, died 1771. Dina Johanna Magdalena is born in 1771, to the same parents.

Both are children of Jacobus Oosthuizen and Dina Carolina Britz, married at Tulbagh in 1749, twelve children, both dead in 1802 and their estate the one that survives here complete. The custom is already used as a dating tool, a name reappearing being evidence that the earlier child is dead, and these two show it working inside a single year.

It is also the source of most of the confusion in these registers. Two people of one name three years apart is not a mistake in the paper; it is a family remembering a child.

Counted from the marriages, births and deaths gathered here, and from the parentage that links them. Only an entry with a full date can be counted, so every n is smaller than the true number, 56 dated marriages before 1850 against far more marriages than that. These are one family's records across one district. The patterns are real in them; whether they held for the Karoo at large is beyond what has been gathered here.

Research notes

The research behind this page, kept separate from it: dated working notes, open questions, and what is not yet settled.